Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Comparison of Multiple Surface Ocean Wind Products with Buoy Data over Blue Amazon (Brazilian Continental Margin)

Comparison of Multiple Surface Ocean Wind Products with Buoy Data over Blue Amazon (Brazilian... Hindawi Advances in Meteorology Volume 2021, Article ID 6680626, 19 pages https://doi.org/10.1155/2021/6680626 Research Article Comparison of Multiple Surface Ocean Wind Products with Buoy Data over Blue Amazon (Brazilian Continental Margin) 1 1 2 Vitor Paiva , Milton Kampel , and Rosio Camayo Earth Science Coordination, Earth Observation and Geoinformatics, National Institute for Space Research, Sao Jose dos Campos 12227-010, Brazil Earth Science Coordination, Numerical Modelling of the Earth System, National Institute for Space Research, Cachoeira Paulista 12630-000, Brazil Correspondence should be addressed to Milton Kampel; milton.kampel@inpe.br Received 10 December 2020; Revised 24 May 2021; Accepted 31 August 2021; Published 14 September 2021 Academic Editor: Maria Angeles Garc´ıa Copyright © 2021 Vitor Paiva et al. +is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Remote sensing data for space-time characterization of wind fields in extensive oceanic areas have been shown to be increasingly useful. Orbital sensors, such as radar scatterometers, provide data on ocean surface wind speed and direction with spatial and temporal resolutions suitable for multiple applications and air-sea studies. Even considering the relevant role of orbital scat- terometers to estimate ocean surface wind vectors on a regional and global scale, the products must be validated regionally. Six different ocean surface wind datasets, including advanced scatterometer (ASCAT-A and ASCAT-B products) estimates, nu- merical modelling simulations (BRAMS), reanalysis (ERA5), and a blended product (CCMP), were compared statistically with in situ measurements obtained by anemometers installed in fifteen moored buoys in the Brazilian margin (8 buoys in oceanic and 7 in shelf waters) to analyze which dataset best represents the wind field in this region. +e operational ASCAT wind products −1 −1 presented the lowest differences in wind speed and direction from the in situ data (0.77ms <RMSE <1.59ms , spd −1 −1 ° ° 0.75 < R <0.96, −0.68ms <bias <0.38ms , and 12.7 <RMSE <46.8 ). CCMP and ERA5 products also performed well spd spd dir −1 −1 in the statistical comparison with the in situ data (0.81ms <RMSE <1.87ms , 0.76 < R <0.91, spd spd −1 −1 ° ° −1.21ms <bias <0.19ms , and 13.7 <RMSE <46.3 ). +e BRAMS model was the one with the worst performance spd dir −1 (RMSE >1.04m·s , R <0.87). For regions with a higher wind variability, as in the southern Brazilian continental margin, spd spd wind direction estimation by the wind products is more susceptible to errors (RMSE >42.4 ). +e results here presented can be dir used for climatological studies and for the estimation of the potential wind power generation in the Brazilian margin, especially considering the lack of availability or representativeness of regional data for this type of application. forecasting, navigation, marine engineering, and offshore 1. Introduction wind energy [1, 12, 13]. +e vast extension of the global ocean imposes technical Ocean surface wind is one of the main drivers of several oceanic, atmospheric, and climate processes, thus being an and financial limitations on in situ ocean surface winds sampling, which is necessary and valuable in remote sensing important indicator of climate change [1]. Ocean surface wind vector variability is intrinsically linked to that of calibration and validation process. Direct measurements are oceanic processes, such as coastal upwelling [2, 3], primary acquired by anemometers on moored buoys, research cruises, productivity [4, 5], deep water formation [6, 7], advective and vessels of opportunity, or by light detection and ranging volume transport, air-sea momentum, and heat fluxes (Lidar) sensors, providing precise and accurate data. How- [8–11]. Having the ability to measure wind speed and di- ever, these techniques are limited to point measurements that rection with high accuracy becomes essential to these are not able to provide satisfactory spatial and temporal processes’ studies, also considering the impacts on weather coverage to resolve variability at different scales [14]. 2 Advances in Meteorology and correlations) between wind products and in situ mea- Today, more than 10 global and regional atmospheric reanalysis and wind products are available for the scientific surements. However, significant systematic differences do exist in the wind products derived from different scatter- community [15]. Numerical prediction models are capable of simulating ocean surface winds on a global scale but ometers [20, 30, 31] and from different producers [32], depend on the existing knowledge about the physical pro- which may result in detrimental impacts on different ap- cesses and on the availability of data used to parametrize the plications, if unknown. +erefore, characterizing errors and initial conditions of the simulations [16, 17]. In weather inconsistencies among different wind products is important, prediction, the ability to make a skilful forecast requires both not only for guiding wind product users, but also to motivate realistic representation of the atmosphere by numerical research by the producers to solve inconsistencies problems. +e Brazilian oceanic area represented by the “Territorial models and accurate initial conditions [18]. Reanalysis products integrate simulated data from nu- Sea” and the “Exclusive Economic Zone” corresponds to approximately 3.6 million km merical prediction models with atmospheric, oceanic, and . If the “Extended, External, other measurements, but with relatively low spatial reso- Jurisdictional” or “Legal Continental Shelf” claimed to the lutions (250 to ∼31km). +is can be insufficient to ade- Commission on the Limits of the Continental Shelf of the quately represent small and mesoscale meteorological United Nations in three “Revised Partial Proposals” for the features and accurately characterize local wind regimes [19]. outer limit of the Brazilian Continental Shelf (2.1 million Offshore wind blended products usually consider a first km ) is added to this area, the total oceanic area called the guess of the wind vector as the background and assimilate “Blue Amazon” may be approximately 5.7 million km . +is multiple sources of measured wind data from buoys, ships, immense oceanic area contains an enormous amount of and satellites [20]. economically, socially, and strategically relevant resources Orbital remote sensors provide wind data on a global and is extremely important for climate stability and envi- scale, covering remote regions normally characterized by the ronmental quality of the adjacent coast [47]. availability of sparse in situ data, with relatively good ac- +e South Atlantic Subtropical Anticyclone (SASA) is curacy and high spatial and temporal resolution [21]. Ocean the main feature of the atmospheric circulation over the surface wind vectors with high spatial and temporal reso- South Atlantic Ocean, with a seasonal variability extending lution obtained by spaceborne scatterometers are utilized in over south-eastern Brazil during the austral winter and various fields of meteorology, oceanography, and climate retracting eastward during the summer [48]. SASA con- studies [22, 23]. Surface wind and stress fields derived from tributes with northeast winds to the Brazilian continental scatterometer observations can be applied to drive ocean margin that are more constant during austral spring and circulation models on various scales and can also be as- summer [49]. similated into regional and global numerical weather pre- Over the Tropical Atlantic, the Intertropical Conver- diction models [21, 24]. +ese data have also been used to gence Zone (ITCZ) associated with the confluence zone of infer about offshore wind energy potential [20, 25]. the northeast and southeast trade winds migrates seasonally Scatterometers measure the radar cross section of the from its most northerly position, around ∼10 N in August- ocean surface, and numerical inversion of the geophysical September, to its southerly position, around ∼2 S in March- model function (GMF) yields the scatterometer wind April in the Atlantic Ocean [50, 51]. Associated with the measurement. +e present constellation of scatterometers mesoscale oscillations, the frontal systems move with cy- maps the surface wind field globally, with a typical spatial clones and post-frontal anticyclones, changing the fields of resolution of 25–12.5km [26], and has been successfully atmospheric pressure, wind, and other atmospheric vari- ° ° used in weather forecasting applications [27], long-term ables along their trajectory [52]. Between 30 S and 50 S, the climate studies [28], and air-sea interactions [29]. +e main zones with the highest occurrence of cold fronts are con- limitations of scatterometers are contamination by rain centrated on the South-Southwestern Atlantic, mainly in the (depending on the frequency of the transmitted signal), lack coastal areas of Argentina and Uruguay [53]. +e south and of data near the coast (typically within 15km), and poor south-eastern coasts of Brazil and the region between the temporal sampling [21]. extreme south of Brazil and the coast of Argentina are Although the number of space-borne scatterometers is favourable regions for the formation of cyclones [54]. +ese increasing in recent years, regional validation of wind systems cause strong winds, precipitation and decrease in products from different scatterometers [30, 31] and pro- temperature [55]. In this region, the atmospheric circulation ducers [32] remains a challenge [33]. Moored ocean buoys is characterized, on a synoptic scale, mainly by the move- provide the absolute calibration reference for satellite wind ment of transient systems such as cyclonic vortices, frontal retrievals [22]. While the development of GMF and wind systems, and the South Atlantic Convergence Zone (SACZ) retrieval algorithms rely on many inputs (numerical models, [56]. wind retrievals for other satellites, statistical constraints, For better understanding the wind regimes on the etc.), finalized satellite wind retrievals always need to be Brazilian continental margin, and to aid in decision-making verified by comparisons with buoys. for the installation of offshore wind turbines, there is a need Several studies evaluated the accuracy of wind products to conduct a study that contemplates this vast oceanic area. over the ocean comparing them with in situ measurements +e dataset used must be adequate to characterize the spatial globally and regionally [19, 24, 34–46]. In general, these and temporal variability of the local wind fields, and also to authors observed good agreement (mean square error, bias, estimate the potential offshore wind energy production. Advances in Meteorology 3 10°N +is study aims to evaluate and compare different al- ternative sources of offshore wind data, including the Ad- vanced SCATterometer (ASCAT) on board MetOP-A (ASCAT-A) and MetOP-B (ASCAT-B) European satellites, the Cross-Calibrated Multi-Platform wind product, the Eu- 0° ropean Centre for Medium-Range Weather Forecasts ERA5 Reanalysis, and numerical simulations from the Brazilian Regional Atmospheric Modelling System (BRAMS/CPTEC) with in situ ocean surface wind measurements, aiming to 10°S analyze the capacity of these wind data sources in describing the wind field in the Blue Amazon region. 2. Materials and Methods 20°S 2.1. Buoy Wind Data. Time series of in situ wind data measured by anemometers installed in 15 oceanographic buoys moored on the Brazilian continental shelf and in 30°S deeper oceanic waters were obtained from the Brazilian National Buoy Program (PNBOIA) and the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA) project, respectively. +e geographical locations and sum- marized information regarding these buoys are presented in 40°S 60°W 50°W 40°W 30°W 20°W Figure 1 and Table 1, respectively. Longitude Eight PIRATA buoys used in this study are moored at ° ° depths between 1,500 and 6,000m and are equipped with a Figure 1: Map of the study area between 10 N and 40 S and be- ° ° tween 20 W and 60 W, showing the geographical locations of the 15 R. M. Young mechanical anemometer, model 05103 (http:// buoys used for in situ measurements of wind speed and direction. www.youngusa.com/products/11/5.html), or with a Gill Symbols denote different organizational affiliations and domains: WindSonic ultrasonic anemometer (http://gillinstruments. circle, Brazilian National Buoy Program (PNBOIA), shelf waters; com/data/datasheets/windsonic-1405-027-iss7.pdf), installed triangle, Prediction and Research Moored Array in the Tropical 4m abovesea level.Wind direction is measuredwith a Sparton Atlantic (PIRATA), oceanic waters. SP3004D compass (https://sparton.com/) or an EG&G flux- gate compass model 63764 or KVH LP101-5 (https://www. pmel.noaa.gov/gtmba/sensor-specifications). Data with a 10- which in turn is calculated from an iterative process using minute temporal resolution, containing measurements since the wind speed measured by the buoys. For more details, 1997, are available from NOAA’s Pacific Marine Environ- please refer to References [57, 58]. mental Laboratory (https://www.pmel.noaa.gov/tao/drupal/ disdel/). Seven PNBOIA buoys used in this study are moored 2.2. Scatterometer Wind Data. Ocean surface wind data close to the continental shelf break (∼200m isobath). +e estimated by orbital scatterometers are derived from the PNBOIA data, maintained by the Brazilian Navy Hydrog- ASCAT sensor, installed on board MetOp-A and MetOp-B raphy Centre, are available on the Internet (https://www. satellites. Both satellites have been in operation since 10/2006 marinha.mil.br/chm/dados-do-goos-brasil/pnboia-mapa). and 09/2012, respectively, until the present. MetOp-C data Time series of wind data are provided with hourly time were not used due to the lack of concomitant buoy data resolution. +e temporal coverage of the data ranges from during the development of this study. Wind data were ob- 04/2009 to 11/2018. +e buoys contain 2 anemometers, one tained from the EUMETSAT Ocean and Sea Ice Satellite mechanical installed at 4.7m above sea level (used as the Application Facility (OSI SAF) through the Royal Nether- main reference) and another ultrasonic sensor installed at lands Meteorological Institute (KNMI). +ree ASCAT 3.7m. More information about the sensors of each buoy can products were acquired for this study: (i) ASCAT-A at be found at the same Internet address for data access. processing level 2 (L2), with 12.5km of spatial resolution, To adjust the wind measured by the buoys to 10m, a from the period 08/2010 to 10/2018 (https://projects.knmi.nl/ steady-state neutrally stable logarithmic vertical wind profile scatterometer/ascat_osi_co_prod/ascat_app.cgi); (ii) ASCAT- relation was considered. In this case, the wind speed (U) at B, at L2 processing level, spatial resolution of 12.5km, from height z is given by the period 10/2012 to 10/2018 (https://projects.knmi.nl/ scatterometer/ascat_b_osi_co_prod/ascat_app.cgi); and (iii) u z U(z) � ln􏼠 􏼡, (1) ASCAT Coastal Wind Data Record (ASCAT-CDR), L2 k z processing level, 12.5km spatial resolution, from the period where k is the von Karm ´ an ´ constant (k �0.41) and z is the 01/2007 to 03/2014 (https://projects.knmi.nl/scatterometer/ surface roughness that depends on the friction velocity (u ), ascat_cdr/). Latitude 4 Advances in Meteorology Table 1: Main characteristics of in situ measurement buoys. Dataset Buoy Location Distance to the coast (km) Time coverage ° ° ° 8N38 W 8 N; 38 W 1219 1998-01 2014-11 ° ° ° 4N23 W 4 N; 23 W 1082 2006-06 2017-03 ° ° ° 4N38 W 4 N; 38 W 788 2000-03 2013-10 ° ° ° 0N23 W 0 N; 23 W 1491 1999-03 2016-03 PIRATA (OW) ° ° ° 0N35 W 0 N; 35 W 563 1998-01 2015-10 ° ° ° 8S30 W 8 S; 30 W 531 2005-08 2015-12 ° ° ° 14S32 W 14 S; 32 W 614 2005-08 2016-01 ° ° ° 19S34 W 19 S; 34 W 562 2005-09 2014-08 ° ° Fortaleza 3.21 S; 38.40 W 54 2016-11 2017-12 ° ° Recife 8.15 S; 34.56 W 35 2012-09 2016-04 ° ° Porto Seguro 15.99 S; 37.94 W 99 2012-07 2016-12 ° ° PNBOIA (SW) Cabo Frio 2 23.63 S; 42.20 W 72 2016-07 2017-07 ° ° Santos 25.44 S; 45.04 W 165 2011-04 2017-08 ° ° Itaja´ı 28.50 S; 47.39 W 125 2009-04 2018-11 ° ° Rio Grande 31.57 S; 49.86 W 101 2009-04 2017-01 OW: oceanic waters; SW: shelf waters. Wentz et al. [64]. +is dataset uses the variational analysis +e ASCAT-A and ASCAT-B products were processed using the CMOD-5n geophysical model function [59] until method (VAM) to fill gaps in the gridded product [65–67]. September 2018. From then on, the GMF CMOD-7 function VAM generates a surface wind grid, minimizing an objective [60] started to be used (Verhoef, personal communication). function, which measures the mismatch of the in situ and +e ASCAT-CDR product is the result of reprocessing satellite wind data overlaid on reanalysis bottom base, ASCAT-A data using the latest GMF CMOD-7. +e three considered as an initial estimate [63]. Wind data at 10m products, regardless of the GMF used, are optimized for from the CCMP product are available in regular grids for coastal regions [23, 61, 62], containing wind data at 10m dates after 1987 with a spatial resolution of 25km and a above sea level in along-track grids (along the satellite path) temporal resolution of six hours. Data were obtained for the with a spatial resolution of 12.5km, coming from 14 orbital period from 01/1998 to 12/2018 (http://www.remss.com/ cycles per day. measurements/ccmp/). Table 2 summarizes the main characteristics of all al- ternative offshore wind products used in this study. 2.5. ERA5 Reanalysis. +e ERA5 Reanalysis product is de- veloped by the Copernicus Climate Change Service, 2.3. BRAMS Numerical Forecasting Model. +e Brazilian implemented by the European Centre for Medium-Range Regional Atmospheric Modelling System (BRAMS) is a Weather Forecasts (ECMWF) (https://www.ecmwf.int/en/ numerical simulation system designed for atmospheric forecasts/datasets/archive-datasets/reanalysis-datasets/ forecasting and research from regional to hemispheric era5). Reanalysis data are produced by combining short- scales. Based on the Regional Atmospheric Modelling Sys- term numerical predictions with observational data. ERA5 is tem (RAMS), originally developed at Colorado State Uni- produced using the assimilation of atmosphere and surface versity, BRAMS was until recently developed and data in four dimensions through the ECMWF Integrated maintained by the Centre for Weather Forecasting and Forecast System with 137 vertical levels in sigma coordinates Climate Studies (CPTEC) of the Brazilian National Institute (which use atmospheric pressure on the surface as a ref- for Space Research (INPE), the University of São Paulo, and erence) in grids with 31km of resolution for atmospheric other institutions in Brazil and abroad. +e computational levels. Hourly ERA5 data are available for several atmo- code of BRAMS is free software with General Public License spheric variables, including wind zonal and meridional (CC-GPL). More information about BRAMS can be found at components, and air temperature at 10m above sea level. http://brams.cptec.inpe.br/. BRAMS simulations for Monthly updates of ERA5 hourly data on single levels from 2013–2015 were provided by CPTEC/INPE, containing wind 1979 to present are published within 3 months of real time. data at 10m in grids with 5km of spatial resolution and Data were obtained for the period from 01/2000 to 12/2017. hourly temporal resolution. +e coverage area is between ° ° ° ° 7.6 N–37.3 S and 62.5 W–29.6 W. 2.6. Statistical Analysis. To compare the different sources of 2.4. Blended CCMP Data. +e Cross-Calibrated Multi- offshore wind speed data with the in situ buoy measure- Platform (CCMP) is a level 3 (L3) wind product which uses ments, the root mean square error (RMSE ), bias , spd spd satellite and in situ data, and numerical modelling simu- Pierson correlation coefficient (R ), and standard deviation spd lations. +e first CCMP version, described in the study of (σ ) were calculated considering the simultaneous and spd Atlas et al. [63], was discontinued in 2012. +e second valid wind speed data records among all datasets, according CCMP version used in this work is described in the study of to the following equations: Advances in Meteorology 5 Table 2: Main characteristics of the considered wind products. Wind Type of Processing Spatial Temporal Time Included data sources product dataset level resolution resolution coverage ASCAT-A 3–5 swaths/ Satellite ASCAT/MetOP-A L2B 12.5km 2007–present Coastal day ASCAT-B 3–5 swaths/ Satellite ASCAT/MetOP-B L2B 12.5km 2012–present Coastal day ASCAT- 3–5 swaths/ Satellite ASCAT/MetOP-A L2B 12.5km 2007–2014 CDR day BRAMS NWP Forcing data from reanalysis — 5km Hourly 2013–2015 ECMWF analyses, QuikSCAT, ASCAT/ CCMP V2.0 Blended MetOP-A, WindSat, SSM/I, SSMIS, AMSR-E, L3 0.25lat/lon 6 hours 1987–present AMSR2, GMI ERA5 Reanalysis NWP and observations — 31km Hourly 1979–present 􏽶�������������� � 􏽐 sin􏼐θ − θ 􏼑sin􏼐θ − θ 􏼑 r,i r e,i e N i�1 1 2 􏽱�������������������������� R � . dir (8) (2) 2 2 RMSE � 􏽘 􏼐x − x 􏼑 , N spd e,i r,i 􏽐 sin 􏼐θ − θ 􏼑sin 􏼐θ − θ 􏼑 r,i r e,i e N i�1 i�1 Please, refer to the study of Jammalamadaka and Sen- gupta [68] for more information about circular statistics. +e bias � 􏽘􏼐x − x 􏼑, (3) spd e,i r,i bias and RMSE formulas for wind direction data are the i�1 same as for wind speed data. However, the minor angle difference (θ ) was derived as follows: x − x x − x 􏽐 􏼐 􏼑􏼐 􏼑 i�1 e,i e r,i r 􏽱������������� �􏽱������������� � ° ° R � , (4) spd ⎧ ⎪ ⎫ ⎪ θ − θ , if − 180 <􏼐θ − θ 􏼑 <180 2 2 ⎪ e,i r,i e,i r,i ⎪ N N ⎪ ⎪ ⎪ ⎪ 􏽐 􏼐x − x 􏼑 􏽐 􏼐x − x 􏼑 i�1 e,i e i�1 r,i r ⎨ ⎬ θ � θ − θ − 360, if 􏼐θ − θ 􏼑 >180 . d,i e,i r,i e,i r,i ⎪ ⎪ 􏽶������������ � ⎪ ⎪ ⎪ ⎪ 􏽴 ⎪ ⎪ ⎩ ° ⎭ θ − θ + 360, if 􏼐θ − θ 􏼑 < − 180 e,i r,i e,i r,i (5) σ � 􏽘 x − x􏼁 , spd i (9) i�1 +erefore, RMSE and bias for direction can be read as where x refers to the wind speed values at instant i, r refers to follows: 􏽶���������� the in situ buoy data used as a reference, and e refers to the data estimated by the alternative sources of wind data. 1 (10) RMSE � 􏽘 􏼐θ 􏼑 , dir d,i Regarding wind direction data, the circular statistics was i�1 applied. +e mean direction (θ) is estimated as follows: − 1 ⎧ ⎪ ⎫ ⎪ N ⎪ tan , if C >0, S >0 ⎪ ⎪ ⎪ ⎪ ⎪ bias � 􏽘 􏼐θ 􏼑. (11) ⎪ ⎪ dir d,i ⎪ ⎪ ⎪ N ⎪ ⎪ i�1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎬ −1 +e comparisons considered the buoys geographic co- θ � . (6) tan + 180 , if C <0 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ordinates and the respective grid point or pixel of the other ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ datasets. Each buoy is identified by its name (PNBOIA, shelf ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ waters) or geographic coordinates (PIRATA, oceanic wa- −1 ⎩ ⎭ tan + 360 , if C >0, S <0 ters), as adopted by the respective monitoring programs. An interval of 30 minutes was considered around the mea- Considering each angle (θ) in an instant i as a vector, surement time of the in situ PNBOIA data, and of 5 minutes N N C � 􏽐 cos(θ ) and S � 􏽐 sin(θ ) are the respective x- i�1 i i�1 i in the case of PIRATA buoys, according to the respective and y-axis representations of the resultant vector length (r). temporal resolution of the in situ data recording (hourly or +e circular standard deviation (σ ) is estimated by the dir 10 minutes, respectively). Two approaches were taken for following equation: data comparison, prioritizing different criteria: (i) the first 􏽰������ approach compared data that were obtained simultaneously (7) σ � 2 ln(r), dir by all the five datasets—ASCAT, CCMP, ERA5, BRAMS, √������ 2 2 where r � (r/N) � ( C + S /N) would be the mean re- buoys—(ii) a second approach compared all available paired data separately for each buoy. When comparing sultant length. +e circular correlation (R ) is defined as dir 6 Advances in Meteorology −1 simultaneous data (i), the Taylor diagrams [69] were pro- −0.48m·s for BRAMS and ERA5. +e highest correlations duced, and ASCAT-A, ASCAT-B, and ASCAT-CDR data (R >0.87) were obtained for ASCATproducts considering spd were considered separately, due to the difficulty of matching all and SW buoys, or CCMP considering OW buoys. them over time. Wind roses diagrams were also generated. In general, relatively higher differences were observed in relation to wind direction, especially for shelf water buoys. Best results were obtained for OW with RMSE values dir 3. Results ° ° ° <23.2 , bias between −0.09 and 0.19 , and correlations dir >0.83, except for the BRAMS product (RMSE �31.2 , 3.1. In Situ Buoy Data. Wind roses of oceanic water buoys dir bias � −1.40 , R �0.72, in the best cases). Considering all (PIRATA) show a low variability in the wind direction, dir dir buoys, the values of RMSE and bias varied between 46.8 dir dir except for the buoys moored in the north-south oscillation ° ° ° and 53.4 , and between −1.06 and 4.88 , respectively. Also, region of the Intertropical Convergence Zone [44, 70] the best correlations (R >0.87) were obtained for the (Figure 2). +e northernmost buoy (8N38W) has dominant dir CCMP product. On the contrary, high RMSE values NE winds and the buoys 4N23W and 4N38W have two dir ° ° ° >73.5 , bias between −14.34 and −0.09 and lower cor- predominant wind directions: NNE–SSE and NE–SE, re- dir relations (<0.35) were obtained for SW. spectively. +ese buoys also have relatively higher standard ° ° +e comparisons of all available paired data for each deviations in wind direction (52.6 and 82.1 , Table 3). +e oceanic water buoy did not differ much with respect to wind 0N35W and 08S30W buoys have dominant winds from ESE speed (Figure 6 and Table S2 in Supplemental File). +e and the 0N23W buoy from SE. Buoys 14S32W and 19S34W −1 lower RMSE values varied between 0.77m·s (ASCAT-A, have dominant E-SE and E-NE winds, respectively. +e spd −1 0N23W) and 1.13m·s (ASCAT-B, 19S34W), except for higher mean wind speeds were observed for buoys 8S30W −1 −1 −1 −1 BRAMS (1.04m·s to 1.90m·s ), with bias between spd (7.36m·s ) and 14S32W (6.92m·s ). −1 −1 −0.73m·s (ERA5, 4N23W) and 0.43m·s (ASCAT-B, For shelf water buoys PNBOIA, the higher mean wind −1 8N38W), and correlation coefficients between 0.61 (BRAMS, speeds were observed in Fortaleza (8.91m·s ), Rio Grande −1 −1 −1 4N38W) and 0.92 (ASCAT-A, 8N38W). Regarding wind (8.34m·s ), Itaja´ı (7.68m·s ), and Cabo Frio 2 (7.52m·s ) direction, the lower RMSE values varied between 13.7 (Figure 2, Table 3). Fortaleza, Recife, and Cabo Frio 2 buoys dir (CCMP, 8S30W) and 32.5 (ASCAT-B and ASCAT-CDR, showed a lower variability in wind direction, with domi- 4N23W), with bias between −9.18 (ASCAT-A, 4N38W) nance from the SE-E and NE quadrants, respectively. dir and 19.5 (BRAMS, 19S34W), and correlation coefficients However, in Cabo Frio 2, the standard deviation of the mean ° between 0.57 (BRAMS, 0N35W) and 0.92 (CCMP, 4N38W). wind direction was 75.0 (Table 3), while for Fortaleza and ° ° In general, the lower RMSE and bias values and the higher R Recife, the standard deviations were lower (23.1 and 33.3 , values, both for wind speed and direction, were obtained for respectively). the ASCAT and CCMP products. +e comparisons of all available paired data for each 3.2. Ocean Surface Wind Statistical Comparisons. shelf water buoy did not differ much with respect to wind +roughout the statistical comparisons presented below, a speed (Figure 7 and Table S3 in Supplemental File). +e −1 negative systematic bias was observed for the BRAMS wind lower RMSE values varied between 0.78m·s (ASCAT-A, spd −1 speed data, which was corrected by applying the scaled Cabo Frio 2) and 1.09m·s (ASCAT-B, Itaja´ı), with bias spd −1 −1 distribution mapping method [71]. For oceanic waters (OW) between −1.21m·s (ERA5, Fortaleza) and −0.004m·s buoys comparisons, the correction of the systematic bias (ASCAT-B, Rio Grande), and R between 0.51 (BRAMS, spd −1 resulted in lower RMSE Recife) and 0.96 (ASCAT-A, Fortaleza, and ASCAT-B, Cabo values (a decrease of 1.8m·s in spd the best case), while for shelf waters (SW) buoys, there was Frio 2). Regarding wind direction, the lower RMSE values dir ° ° no relevant change with slightly higher RMSE values for varied between 12.7 (ASCAT-B, Fortaleza) and 79.6 spd two buoys (Figure S1 in Supplemental File). +erefore, (CCMP, Santos), with bias between −28.5 (BRAMS, dir BRAMS wind speed data with systematic bias correction Santos) and 40.5 (BRAMS, Rio Grande), and circular were used only for comparisons with OW buoys. correlation coefficients between 0.58 (CCMP and ERA5, +e simultaneous comparisons of the ocean surface wind Itaja´ı) and 0.94 (ASCAT-B, Fortaleza), except for compar- products with in situ measured buoys data did not differ isons with the Santos buoy (≤0.34). much with respect to wind speed (Figures 3–5 and Table S1 in Supplemental File). Considering the comparison of all buoys together, the RMSE values ranged between 3.3. Error Dependence on Measured Wind Speed. spd −1 −1 Scatterometers limitations in representing low and high 0.98m·s and 1.82m·s . Considering the OW and SW wind speeds are well documented in the literature [72, 73]. buoys separately, the RMSE values were relatively lower spd −1 −1 (0.91m·s and 1.02m·s in the best cases, respectively). +e To analyze if wind speeds from the alternative wind data sources show any variation or dependence with the in situ highest values of RMSE were obtained for the product spd BRAMS, even after removing the systematic bias. For all buoy data, four different bins were considered for statistical −1 −1 −1 −1 −1 comparison: 3–6m·s ; 6–9m·s ; 9–12m·s and >12m·s buoys, the bias varied between ±0.30m·s . For OW spd −1 buoys, the bias varied between ±0.19m·s . For SW, the (Figures 8 and 9 and Table S4 in Supplemental File). In spd −1 −1 −1 general, the lowest RMSE (0.76m·s ) and bias bias varied between −0.18m·s and −0.25m·s for spd spd spd −1 −1 ASCAT and CCMP products and between −0.70m·s and (−0.05m·s ), and highest R values (0.70) were observed spd Advances in Meteorology 7 N N N 38 % 25 % 20 % NW NE NW NE NW NE 30 20 22 15 15 10 8 5 W E W EE W SW SE SW SE SW SE S S S N N N 38 % 40 % 50 % NW NE NW NE NW NE 30 40 22 30 15 20 8 10 W E W EE W SW SE SW SE SW SE S S S N N N 38 % 20 % 40 % NW NE NW NE NW NE 15 30 10 20 5 10 W E W EE W SW SE SW SE SW SE S S S N N N 30 % 9 % 30 % NW NE NW NE NW NE 6 20 3 10 W E W EE W SW SE SW SE SW SE S S S N N N 12 % 9 % 9 % NW NE NW NE NW NE 6 6 3 3 W EE W W E SW SE SW SE SW SE S S S –1 Wind Speed (m.s ) 0 - 3 6 - 9 12 - 15 3 - 6 9 - 12 > 15 Figure 2: Wind roses of oceanic (PIRATA) and shelf (PNBOIA) water buoys. Santos Recife 14S32W 0N35W 8N38W 25.44°S; 45.04°W 8.15°S; 34.56°W 14°S; 32°W 0°N; 23°W 8°N; 38°W Itajaí Porto Seguro 19S34W 0N23W 4N38W 28.50°S; 47.39°W 15.99°S; 37.94°W 19°S; 34°W 0°N; 35°W 4°N; 23°W Rio Grande Cabo Frio 2 Fortaleza 8S30W 4N23W 31.57° S; 49.86°W 2 23.63° S; 42.20°W 3.21°S; 38.40°W 8°S; 30°W 4°N; 38°W 1.0 1.0 1.0 0.99 0.99 0.99 0.95 0.95 0.95 0.9 0.9 0.9 0.8 0.8 0.8 Correlation Correlation Correlation 0.7 0.7 0.7 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 8 Advances in Meteorology Table 3: Mean (±standard deviation) and maximum and minimum wind speed and circular mean (±circular standard deviation) wind direction from the in situ measurement buoys. Mean Max Min Dataset Buoy −1 −1 −1 Speed (ms ) Dir. ( ) Speed (ms ) Speed (ms ) 8N38W 6.78 ±2.60 244 ±60.0 20.6 0.00 4N23W 5.38 ±2.17 307 ±82.1 32.3 0.00 4N38W 6.31 ±2.21 274 ±52.6 17.9 0.00 0N23W 5.67 ±1.89 306 ±35.5 19.0 0.00 PIRATA (OW) 0N35W 6.27 ±2.07 288 ±38.2 17.1 0.00 8S30W 7.36 ±1.70 290 ±20.5 16.3 0.00 14S32W 6.92 ±1.90 276 ±28.1 17.1 0.00 19S34W 6.15 ±2.17 254 ±52.3 16.5 0.00 Fortaleza 8.91 ±2.19 287 ±23.1 16.0 0.43 Recife 6.53 ±1.76 298 ±33.3 47.4 0.11 Porto Seguro 5.53 ±2.42 280 ±111 16.1 0.11 PNBOIA (SW) Cabo Frio 2 7.52 ±3.19 264 ±75.0 16.3 0.11 Santos 6.88 ±3.07 290 ±102 20.1 0.11 Itaja´ı 7.68 ±3.41 242 ±146 36.3 0.11 Rio Grande 8.34 ±3.58 218 ±113 48.9 0.11 OW: oceanic waters; SW: shelf waters. 0.0 0.0 0.0 0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0 6 12 18 24 30 36 42 48 54 Standard deviation Standard deviation Standard deviation PIRATA Bias PIRATA Bias PIRATA Bias CCMP 0.08 CCMP 0.10 CCMP 1.12 ERA5 -0.08 ERA5 0.54 ERA5 -0.09 BRAMS 0.02 BRAMS 0.04 BRAMS 7.92 ASCAT 0.17 ASCAT 0.29 ASCAT 0.19 RMSE RMSE RMSE (a) (b) (c) Figure 3: Continued. 0.6 0.6 1.2 1.2 8 1.0 1.0 1.0 1.0 1.0 1.0 0.99 0.99 0.99 0.99 0.99 0.99 0.95 0.95 0.95 0.95 0.95 0.95 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 Correlation Correlation Correlation Correlation Correlation Correlation 0.7 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.2 Advances in Meteorology 9 0.0 0.0 0.0 0 6 12 182430364248 54 0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0 8 16 24 32 40 48 56 Standard deviation Standard deviation Standard deviation PIRATA Bias PIRATA Bias PIRATA Bias CCMP 0.19 CCMP 0.10 CCMP 1.78 ERA5 -0.30 ERA5 -0.07 ERA5 1.27 BRAMS 6.34 BRAMS 0.02 BRAMS 7.96 ASCAT 0.40 ASCAT 0.18 ASCAT 0.83 RMSE RMSE RMSE (d) (e) (f) Figure 3: Taylor diagrams for statistical comparisons between CCMP, BRAMS, ERA5, and ASCAT-A (a, b), or ASCAT-B (c, d), or ASCAT- CDR (e, f) that have coincident records with oceanic water buoy data. (a) Speed-ASCAT-A. (b) Direction-ASCAT-A. (c) Speed-ASCAT-B. (d) Direction-ASCAT-B. (e) Speed-ASCAT-CDR. (f) Direction-ASCAT-CDR. 0.0 0.0 0.0 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 0 15 30 45 60 75 90 105 120 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 Standard deviation Standard deviation Standard deviation PNBOIA Bias PNBOIA Bias PNBOIA Bias CCMP -0.24 CCMP -6.15 CCMP -0.25 ERA5 -0.69 ERA5 -6.87 ERA5 -0.70 -0.49 -1.37 -0.60 BRAMS BRAMS BRAMS ASCAT -0.23 ASCAT -8.42 ASCAT -0.18 RMSE RMSE RMSE (a) (b) (c) Figure 4: Continued. 0.6 3.0 3.0 1.0 1.0 1.2 2.0 2.0 6 1.0 1.0 1.0 1.0 1.0 1.0 0.99 0.99 0.99 0.99 0.99 0.99 0.95 0.95 0.95 0.95 0.95 0.95 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 Correlation Correlation Correlation Correlation Correlation Correlation 0.7 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.2 10 Advances in Meteorology 0.0 0.0 0.0 0 15 30 45 60 75 90 105 120 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 0 15 30 45 60 75 90 105 120 Standard deviation Standard deviation Standard deviation PNBOIA Bias PNBOIA Bias PNBOIA Bias CCMP -6.11 CCMP -0.18 CCMP -13.52 ERA5 -4.16 ERA5 -0.65 ERA5 -13.67 BRAMS -0.09 BRAMS -0.48 BRAMS -5.51 ASCAT -3.84 ASCAT -0.20 ASCAT -14.34 RMSE RMSE RMSE (d) (e) (f) Figure 4: Taylor diagrams for statistical comparisons between CCMP, BRAMS, ERA5, and ASCAT-A (a, b), or ASCAT-B (c, d), or ASCAT- CDR (e, f) that have coincident records with shelf water buoy data. (a) Speed-ASCAT-A. (b) Direction-ASCAT-A. (c) Speed-ASCAT-B. (d) Direction-ASCAT-B. (e) Speed-ASCAT-CDR. (f) Direction-ASCAT-CDR. 0.0 0.0 0.0 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 0 1020304050607080 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 Standard deviation Standard deviation Standard deviation PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias CCMP -0.01 CCMP -1.26 CCMP -0.04 ERA5 -0.28 ERA5 -1.89 ERA5 -0.30 BRAMS -0.13 BRAMS 4.88 BRAMS -0.19 ASCAT 0.04 ASCAT -2.56 ASCAT 0.06 RMSE RMSE RMSE (a) (b) (c) Figure 5: Continued. 3.0 2.4 2.4 0.8 0.8 1.0 1.6 1.6 2.0 42 1.0 1.0 1.0 0.99 0.99 0.99 0.95 0.95 0.95 0.9 0.9 0.9 0.8 0.8 0.8 Correlation Correlation Correlation 0.7 0.7 0.7 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 Advances in Meteorology 11 0.0 0.0 0.0 10 20 30 40 50 60 70 80 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 0 8 16 24 32 40 48 56 64 72 Standard deviation Standard deviation Standard deviation PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias CCMP -1.99 CCMP 0.03 CCMP -2.39 ERA5 -1.64 ERA5 -0.23 ERA5 -2.80 BRAMS 4.12 BRAMS -0.11 BRAMS 4.29 ASCAT -1.06 ASCAT 0.07 ASCAT -3.30 RMSE RMSE RMSE (d) (e) (f) Figure 5: Taylor diagrams for statistical comparisons between CCMP, BRAMS, ERA5, and ASCAT-A (a, b), or ASCAT-B (c, d), or ASCAT- CDR (e, f) that have coincident records with all buoys data. (a) Speed-ASCAT-A. (b) Direction-ASCAT-A. (c) Speed-ASCAT-B. (d) Direction-ASCAT-B. (e) Speed-ASCAT-CDR. (f) Direction-ASCAT-CDR. −1 in the 6–9m·s interval, while higher differences were in BRAMS wind speed data, the highest RMSE con- spd −1 −1 −1 observed for winds >12m·s (1.11m·s <RMSE tinued to be obtained for this dataset (RMSE >1.56 m·s spd spd −1 −1 −1 >4.59m·s , −3.55m·s <bias > −0.59m·s ). For all in simultaneous comparisons). spd speed intervals and all datasets considered in the analysis, the Indeed, if analysis is restricted to OW buoys, best per- lower RMSE values and higher R values were obtained formance is observed for CCMP product (Figure 6 and spd spd for ASCAT-A and ASCAT-B products. Regarding wind Table S2), showing the lowest errors in terms of RMSE and dir −1 direction, higher differences were observed in the 3–6m·s the best values of R and R for 5 or more buoys among spd dir interval for OW buoys, considering all wind products. For the 8 OW buoys compared. Although the data from the SW buoys, higher differences were also observed for the PIRATA buoys are assimilated in the generation of the lower wind speed interval, considering ASCAT-A, ASCAT- CCMP dataset, these results agree with previous studies B, and ERA5 products. Best comparisons were obtained for that compared CCMP and PIRATA wind data [41–43]. −1 CCMP product with oceanic water buoys (RMSE <23.3 , Suzuki et al. [42] obtained an RMSE value of 1.04m·s , dir spd ° ° −0.01 <bias <0.61 , R >0.91), and for ASCAT-B with and Yan et al. [43] obtained RMSE , bias , and R dir dir spd spd spd −1 −1 ° ° ° shelf water buoys (RMSE <56.6 , −0.47 <bias <7.66 , values of 1.29m·s , 0.24 m·s , and 0.90, respectively. For dir dir R >0.59), in all speed intervals. wind direction, these authors [43] obtained dir ° ° RMSE �22.4 , bias �0.89 , and R �0.97. In the dir dir dir present study, ASCAT products showed the best results 4. Discussion after CCMP, with a performance close to other recent In the present study, data from six different ocean surface comparisons [44]. +e lowest errors in terms of RMSE spd wind products were compared with in situ measurements for 5 of the 8 OW buoys were obtained for ASCAT-A. from 15 moored buoys in oceanic and shelf waters of the Verhoef et al. [61] validated the ASCAT-A product opti- Brazilian margin. In general, there is a good correspon- mized for coastal regions using in situ buoy data from the dence between the wind speeds from the simultaneous tropical oceans and along North American and European records of wind products and buoy observations coasts. For buoys located more than 50km from the coast, −1 (Figures 3–5 and Table S1). Without considering the these authors obtained RMSE of 2.2 m·s and bias of spd spd −1 −1 BRAMS model, comparisons of simultaneous records are −0.29m·s . In the present study, RMSE <1.25m·s and spd −1 −1 even better. It can be noted that ASCAT and CCMP bias between −0.33m·s and 0.25 m·s were computed spd products had slightly better results than ERA5. +e spatial for ASCAT-A in oceanic waters. In the ASCAT-B vali- −1 resolution of the databases seems to influence their per- dation report [62], bias values of 0.02 m s (CMOD5.n) −1 formance, with the finest resolution products (satellite- and 0.04 m s (CMOD7) were reported, whereas in the −1 derived databases) showing better results than the coarser present study, bias values varied between −0.24m·s and −1 ERA5 database. Despite the removal of the systematic bias 0.26 m·s . In the ASCAT-CDR validation report [23], bias 2.4 0.8 1.6 4 12 Advances in Meteorology 2.0 1.5 1.0 0.5 0.0 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 0.4 0.0 –0.4 –0.8 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 1.00 0.75 0.50 0.25 0.00 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W –10 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 1.00 0.75 0.50 0.25 0.00 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W OW buoys surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 Figure 6: Barplots representing the statistical metrics computed for each oceanic water (OW) buoy. RMSE and RMSE : mean square spd dir error for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s correlation coefficient for spd dir spd dir wind speed and direction. –1 –1 R bias (°) RMSE (°) R bias (m s ) RMSE (m s ) dir dir dir spd spd spd Advances in Meteorology 13 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Rio Grande Itajaí 0.0 –0.4 –0.8 –1.2 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande 1.00 0.75 0.50 0.25 0.00 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Rio Grande Itajaí Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande –20 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande 1.00 0.75 0.50 0.25 0.00 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande SW buoys surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 Figure 7: Barplots representing the statistical metrics computed for each shelf waters (SW) buoy. RMSE and RMSE : mean square error spd dir for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s correlation coefficient for wind spd dir spd dir speed and direction. –1 –1 R bias (°) RMSE (°) R bias (m s ) RMSE (m s ) dir dir dir spd spd spd 14 Advances in Meteorology 0 0 0.5 7.5 0.0 5.0 –0.5 2.5 –1.0 –1.5 0.0 –2.0 –2.5 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 3-6 6-9 9-12 >12 3-6 6-9 9-12 >12 wind speed intervals surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 −1 Figure 8: Barplots representing the statistical metrics computed for all buoys in four wind speed bins (m·s ). RMSE and RMSE : mean spd dir square error for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s correlation spd dir spd dir coefficient for wind speed and direction. −1 −1 values ranged between −0.3 m·s and 0.2m·s , whereas in shows the differences between the wind roses of Santos buoy −1 the present study, bias values varied between −0.37m·s and ASCAT-A paired series, which are also observed for the −1 other datasets considered in this study (Figures S2–S16 in and 0.15m·s . In terms of wind direction, relatively higher differences Supplemental File). It is worth remembering that PNBOIA data (SW) are hourly averages, while PIRATA data (OW) are were observed in the present study, even without consid- ering BRAMS comparisons. Most oceanic water buoys are 10 minutes averages, which can also contribute to the rel- located in the Tropical Atlantic region where winds are more atively greater differences observed for southern shelf water stable [50, 51]. However, highest statistical differences were buoy comparisons [75]. For satellite-derived wind products, observed for shelf water buoys (Figure 2 and Table S1). Wind land masking limitations can be an issue [45]. In addition, direction comparisons were considered more satisfactory more coastal winds are very dependent on the surrounding only for SW buoys located in the northern part of the topography, land-ocean gradients, and discontinuities. Brazilian continental margin (Figure 7 and Table S3). For Previous studies reported overestimations of wind speed Santos, Itaja´ı, and Rio Grande buoys located in the southern by coastal ASCAT-A and ASCAT-B products in the Iberian, Irish, and Japanese continental shelf regions [20, 45, 46]. part of the study region, wind direction comparisons resulted in relatively higher differences. According to Lin Takeyama et al. [46] proposed that this overestimation is et al. [74], when comparing the wind vectors classified as inversely proportional to the distance from the coast due to variable or stable, the estimated error variances of the increased backscatter by land contamination. However, variable winds are much higher than those of the stable Verhoef et al. [61] observed that the ASCAT-A product winds. +e southern Brazilian continental margin is well optimized for the coastal region underestimated buoy wind known in the literature for frequent incursions of frontal speed measurements in the northern hemisphere. In the systems and relevant cyclogenetic activity, which increases present study, not only ASCAT products but all wind the variability of the wind field in this region [52–56]. +is products underestimated buoy wind speed for shelf waters. variability pattern for these 3 buoys can be depicted from the Payan [76] observed that the ASCAT-A product over- wind roses shown in Figure 2. As an example, Figure 10 estimated buoy wind speeds in the northern hemisphere and –1 –1 R bias (m s ) RMSE (m s ) spd spd spd R bias (°) RMSE (°) dir dir dir Advances in Meteorology 15 oceanic waters shelf waters –1 –2 –3 –4 0.75 0.50 0.25 0.00 –0.25 –5 –10 –15 1.00 0.75 0.50 0.25 0.00 3-6 6-9 9-12 >12 3-6 6-9 9-12 >12 wind speed intervals surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 −1 Figure 9: Barplots representing the statistical metrics computed for oceanic and shelf water buoys in four wind speed bins (m·s ). RMSE spd and RMSE : mean square error for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s dir spd dir spd dir correlation coefficient for wind speed and direction. –1 –1 R bias (°) RMSE (°) R bias (m s ) RMSE (m s ) dir dir dir spd spd spd 16 Advances in Meteorology N N NW NE NW NE 15 15 10 10 5 5 W E W E SW SE SW SE S S –1 –1 Wind Speed (m.s ) Wind Speed (m.s ) 0 - 3 6 - 9 12 - 15 0 - 3 6 - 9 12 - 15 3 - 6 9 - 12 > 15 3 - 6 9 - 12 > 15 (a) (b) Figure 10: Wind roses of the Santos buoy and ASCAT-A paired series (Please see Figures S2–S16 for all buoys and wind datasets). underestimated speeds measured in the southern hemi- 5. Conclusions sphere. In general, the comparisons of wind speed obtained +is study evaluates and compares satellite-derived ocean in the present study for shelf waters agree with previous surface wind data (ASCAT-A and ASCAT-B scatter- studies. +e best performances are distributed between ASCATproducts (Figure 7, Table S3). CCMP and ERA5 also ometers), reanalysis (ERA5), BRAMS-modelled surface winds, and a satellite-reanalysis-buoy-blended product show good performance for wind speed, with exception for Recife buoy, which is the closest to the coast (35km). (Cross-Calibrated Multi-Platform ocean wind vectors), to analyze the performance of these datasets in representing the Regarding the comparisons by wind speed intervals (Figures 8 and 9 and Table S4), ASCAT-A and ASCAT-B in situ measurements obtained by moored buoys along the Brazilian margin. +e results presented here can be of great showed a better statistical performance in terms of RMSE value for climate and oceanographic studies, among others, and R, considering all datasets. In general, the lowest especially where in situ measured wind data are not available RMSE and higher R values were observed in the spd spd −1 or are insufficient, therefore requiring alternative sources of 6–9m·s interval. +e poor wind speed accuracy for interval −1 wind data to be considered. >12m·s for OW buoys can be related to the relatively low Ocean surface wind estimations obtained with ASCAT amount of data recorded for this wind speed interval, as seen scatterometer optimized for coastal regions and with a in the wind roses shown in Figure 2. Almost all wind −1 −1 spatial resolution of 12.5km are the ones that best represent products overestimated speeds between 3m·s and 6m·s −1 and underestimated speeds >6m·s , so that as wind speed the in situ wind speed, with the lower RMSE spd −1 −1 (<1.59m·s ), lower bias (±0.68m·s ), and higher cor- increases, the negative bias increases proportionally. spd According to Carvalho et al. [19], weaker winds produce low relation (>0.75) in relation to the other products considered in the study. However, CCMP product also showed a good amounts of backscatter, which scatterometers have diffi- culties in measuring. Stronger winds can only produce statistical comparison with in situ buoy data −1 −1 (RMSE <1.87m·s , bias ±0.96m·s , R >0.82), as well as proportional backscatter in the ocean surface until a certain −1 −1 ERA5 (RMSE <1.84m·s , bias ±1.21m·s , threshold, after which no more backscatter is produced even spd spd R >0.76). Among all the compared ocean surface wind if the wind speed keeps increasing. Regarding wind direc- spd products, BRAMS is the least able to represent the wind field tion, best comparisons with OW buoys were obtained for −1 (RMSE >1.04m·s , bias ±2.89, R <0.87). In regions CCMP product. High values of RMSE and low correla- dir spd spd spd with more variable winds, as in the southern Brazilian tions for SW buoys can be related to subcell wind variability in southern Brazilian shelf, as mentioned above [74]. In continental margin, the estimation of wind direction is more susceptible to errors (RMSE >42.4 ). general, the statistical performance for wind direction in- dir creases towards the higher speed intervals. According to Considering the relatively higher temporal resolution of ERA5 and CCMP products (1 and 6 hours, respectively), and Chakraborty et al. [37], the scatterometer errors associated with wind direction are highly random and are dependent the extension of the available time series (>30 years), it is suggested to use these datasets to build a regional clima- on wind speed. In general, the lower the wind speeds, the tology and to analyze the space-time variability of wind fields higher the wind direction errors. Santos (%) ASCAT-A (%) Advances in Meteorology 17 [3] C. Aguirre, S. Garc´ıa-Loyola, G. Testa, D. Silva, L. Farias, and in the study region. Nevertheless, there are seven orbital J. E. Tremblay, “Insight into anthropogenic forcing on coastal scatterometers operating currently [77–81]. +is number of upwelling off south-central Chile,” Elementa Science of the operational scatterometers has been increasing in recent Anthropocene, vol. 6, no. 59, 2018. years and in near future. Once the scatterometers can give [4] H. Demarcq, “Trends in primary production, sea surface winds within 6 hours over the regional ocean, it may be good temperature and wind in upwelling systems (1998–2007),” to use combined winds from multiple scatterometers. +e Progress in Oceanography, vol. 83, no. 1–4, pp. 376–385, 2009. derived information for such satellite constellation and al- [5] M. G. Jacox, E. L. Hazen, and S. J. Bograd, “Optimal envi- ternative sources could be useful for several applications, as ronmental conditions and anomalous ecosystem responses: on the estimation of the potential wind power generation in constraining bottom-up controls of phytoplankton biomass in the Brazilian margin region. the California current system,” Scientific Reports, vol. 6, Ar- ticle ID 27612, 2016. [6] Y. Amitai, Y. Ashkenazy, and H. Gildor, “+e effect of wind- Data Availability stress over the eastern Mediterranean on deep-water for- mation in the Adriatic sea,” Deep Sea Research Part II: Topical In situ buoy wind data can be accessed at https:// Studies in Oceanography, vol. 164, pp. 5–13, 2019. www.marinha.mil.br/chm/dados-do-goos-brasil/pnboia- [7] J. L. Bullister, M. Rhein, and C. Mauritzen, “Deepwater mapa and https://www.pmel.noaa.gov/tao/drupal/disdel/; formation,” in Ocean Circulation and Climate—A 21st Cen- scatterometer wind data at projects.knmi.nl/scatterometer/; tury Perspective, vol. 103, pp. 227–253, Academic Press, blended CCMP data at http://www.remss.com/measure- Cambridge, MA, USA, 2013. ments/ccmp/; and ERA5 reanalysis data at www.ecmwf.int/ [8] W. G. Large and S. Pond, “Open ocean momentum flux en/forecasts/datasets/archive-datasets/reanalysis-datasets/ measurements in moderate to strong winds,” Journal of era5. +e authors agree to make data and materials sup- Physical Oceanography, vol. 11, no. 3, pp. 324–336, 1981. porting the results presented in the manuscript available [9] D. B. Chelton, M. G. Schlax, M. H. Freilich, and R. F. Milliff, upon reasonable request. “Satellite measurements reveal persistent small-scale features in ocean winds,” Science, vol. 303, no. 5660, pp. 978–983, Conflicts of Interest [10] C. Fairall, B. Barnier, and D. I. Berry, “Observations to quantify air-sea fluxes and their role in climate variability and +e authors declare that there are no conflicts of interest predictability,” in Proceedings of 2009 OceanObs: Sustained regarding the publication of this paper. Ocean Observations and Information for Society, vol. 2, Venice, Italy, September 2009. Acknowledgments [11] J. P. Machado, F. Justino, and L. P. Pezzi, “Impacts of wind stress changes on the global heat transport, baroclinic in- +e authors would like to thank the Brazilian Navy Hy- stability, and the thermohaline circulation,” Advances in drography Centre, INPE’s Weather and Climate Forecast Meteorology, vol. 2016, Article ID 2089418, 15 pages, 2016. Centre, the Pacific Marine Environmental Laboratory, the [12] W. T. Liu, W. Tang, and X. Xie, “Wind power distribution European Organization for the Exploitation of Meteoro- over the ocean,” Geophysical Research Letters, vol. 35, no. 13, logical Satellites, the Royal Netherlands Meteorological Institute, and the European Centre for Medium Range [13] J. Vogelzang and A. Stoffelen, “Scatterometer wind vector Weather Forecasts for the data used in this study. +is work products for application in meteorology and oceanography,” Journal of Sea Research, vol. 74, pp. 16–25, 2012. was supported by the Coordenação de Aperfeiçoamento de [14] C. M. Risien and D. B. Chelton, “A satellite-derived clima- Pessoal de N´ıvel Superior (CAPES), Brazil (Grant Finance tology of global ocean winds,” Remote Sensing of Environ- Code 001). VGP was supported by CAPES (PROEX 0487/ ment, vol. 105, no. 3, pp. 221–236, 2006. 1681024). [15] D. Dee, J. Fasullo, D. Shea, J. Walsh, and NCAR, >e Climate Data Guide: Atmospheric Reanalysis: Overview & Comparison Supplementary Materials Tables, National Center for Atmospheric Research, Boulder, CO, USA, 2016, https://climatedataguide.ucar.edu/climate- +e supporting information consists of detailed tables and data/atmospheric-reanalysis-overview-comparison-tables. figures of our results that are mentioned and discussed in the [16] A. J. Wallcraft, A. B. Kara, C. N. Barron, E. J. Metzger, text of the Results and Discussion sections. (Supplementary R. L. Pauley, and M. A. Bourassa, “Comparisons of monthly Materials) mean 10m wind speeds from satellites and NWP products over the global ocean,” Journal of Geophysical Research: At- mospheres, vol. 114, no. 16, 2009. References [17] S. R. Smith, P. J. Hughes, and M. A. Bourassa, “A comparison of nine monthly air-sea flux products,” International Journal [1] M. Bourassa, A. Stoffelen, H. Bonekamp et al., “Remotely sensed winds and wind stresses for marine forecasting and of Climatology, vol. 31, no. 7, pp. 1002–1027, 2011. [18] E. Kalnay, Atmospheric Modeling, Data Assimilation and ocean modeling,” in Proceedings of 2009 OceanObs: Sustained Ocean Observations and Information for Society, vol. 2, Predictability, Cambridge University Press, New York, NY, USA, 2003. Venice, Italy, September 2009. [2] J. Kampf ¨ and P. Chapman, “Seasonal wind-driven coastal [19] D. Carvalho, A. Rocha, M. Go´mez-Gesteira, and C. Silva Santos, “Comparison of reanalyzed, analyzed, satellite-retrieved upwelling systems,” in Upwelling Systems of the World, Chapter 8, pp. 315–361, Springer, Berlin, Germany, 2016. and NWP modelled winds with buoy data along the Iberian 18 Advances in Meteorology Peninsula coast,” Remote Sensing of Environment, vol. 152, [36] F. Pimenta, W. Kempton, and R. Garvine, “Combining me- pp. 480–492, 2014. teorological stations and satellite data to evaluate the offshore [20] D. Carvalho, A. Rocha, M. Gomez-Gesteira, ´ and C. Silva wind power resource of southeastern Brazil,” Renewable Santos, “Offshore winds and wind energy production esti- Energy, vol. 33, no. 11, pp. 2375–2387, 2008. mates derived from ASCAT, OSCAT, numerical weather [37] A. Chakraborty, R. Kumar, and A. Stoffelen, “Validation of prediction models and buoys—a comparative study for the ocean surface winds from the OCEANSAT-2 scatterometer using triple collocation,” Remote Sensing Letters, vol. 4, no. 1, Iberian Peninsula Atlantic coast,” Renewable Energy, vol. 102, pp. 433–444, 2017. pp. 84–93, 2013. [38] D. S. Dukhovskoy, M. A. Bourassa, G. N. Petersen, and [21] M. A. Bourassa, A. Stoffelen, H. Bonekamp et al., “Remotely sensed winds and wind stresses for marine forecasting and J. Steffen, “Comparison of the ocean surface vector winds ocean modeling,” Frontiers in Marine Science, vol. 6443 pages, from atmospheric reanalysis and scatterometer-based wind 2019. products over the Nordic seas and the northern north Atlantic [22] F. J. Wentz, L. Ricciardulli, E. Rodriguez et al., “Evaluating and their application for ocean modeling,” Journal of Geo- and extending the ocean wind climate data record,” IEEE physical Research: Oceans, vol.122, no. 3, pp.1943–1973, 2017. Journal of Selected Topics in Applied Earth Observations and [39] J. Yang and J. Zhang, “Comparison of Oceansat-2 scatter- ometer wind data with global moored buoys and ASCAT Remote Sensing, vol. 10, no. 5, pp. 2165–2185, 2017. [23] A. Verhoef, J. Vogelzang, and A. Stoffelen, ASCAT L2 Winds observation,” Advances in Meteorology, vol. 2019, Article ID Data Record Validation Report, Royal Netherlands Meteo- 1651267, 9 pages, 2019. rological Institute, De Bilt, Netherlands, 2016. [40] Z. Wang, A. Stoffelen, J. Zou et al., “Validation of new sea [24] N. Ebuchi, H. C. Graber, and M. J. Caruso, “Evaluation of surface wind products from scatterometers onboard the HY- wind vectors observed by QuikSCAT/SeaWinds using ocean 2B and MetOp-C satellites,” IEEE Transactions on Geoscience buoy data,” Journal of Atmospheric and Oceanic Technology, and Remote Sensing, vol. 58, no. 6, pp. 4387–4394, 2020. vol. 19, no. 12, pp. 2049–2062, 2002. [41] A. Okuro and M. Kubota, “Inter-comparison of daily mean [25] F. M. Pimenta, A. R. Silva, A. T. Assireu, V. d. S. e. Almeida, sea surface wind speeds in various global high-resolution and O. R. Saavedra, “Brazil offshore wind resources and at- products,” Journal of >e Remote Sensing Society of Japan, mospheric surface layer stability,” Energies, vol. 12, no. 21, vol. 34, no. 3, pp. 177–196, 2014. [42] N. Suzuki, M. Konda, K. Kutsuwada, and K. Utsunomiya, p. 4195, 2019. [26] A. B. Villas Boas, ˆ F. Ardhuin, A. Ayet et al., “Integrated “Comparison of the accuracy of various global wind speed observations of global surface winds, currents, and waves: datasets obtained from satellites and reanalyses,” Journal of requirements and challenges for the next decade,” Frontiers in Advanced Marine Science and Technology Society, vol. 24, Marine Science, vol. 6, no. 425, 2019. no. 2, pp. 31–37, 2018. [27] D. B. Chelton, M. H. Freilich, J. M. Sienkiewicz, and J. M. Von [43] Q. Yan, J. Zhang, J. Meng, J. Wang, and C. Fan, “Use of an Ahn, “On the use of QuikSCAT scatterometer measurements optimum interpolation method to construct a high-resolution of surface winds for marine weather prediction,” Monthly global ocean surface vector wind dataset from active scat- Weather Review, vol. 134, no. 8, pp. 2055–2071, 2006. terometers and passive radiometers,” International Journal of [28] D. Halpern, “Offshore Ekman transport and Ekman pumping Remote Sensing, vol. 38, no. 20, pp. 5569–5591, 2017. off Peru during the 1997–1998 El Nino,” Geophysical Research [44] A. Modi, S. K. Munaka, R. Harikumar, T. B. Nair, and Letters, vol. 29, no. 5, pp. 1–19, 2002. K. Srinivas, “Evaluation of winds from SCATSAT-1 and [29] D. Chelton and S.-P. Xie, “Coupled ocean-atmosphere in- ASCAT using buoys in the Indian ocean,” Journal of the teraction at oceanic mesoscales,” Oceanography, vol. 23, no. 4, Indian Society of Remote Sensing, vol. 49, pp.1915–1925, 2021. pp. 52–69, 2010. [45] T. Remmers, F. Cawkwell, C. Desmond, J. Murphy, and [30] A. Bentamy and D. C. Fillon, “Gridded surface wind fields E. Politi, “+e potential of advanced scatterometer (ASCAT) from Metop/ASCATmeasurements,” International Journal of 12.5km coastal observations for offshore wind farm site se- Remote Sensing, vol. 33, no. 6, pp. 1729–1754, 2012. lection in Irish waters,” Energies, vol. 12, no. 2, p. 206, 2019. [31] A. Bentamy, S. A. Grodsky, A. Elyouncha, B. Chapron, and [46] Y. Takeyama, T. Ohsawa, S. Shimada, K. Kozai, K. Kawaguchi, F. Desbiolles, “Homogenization of scatterometer wind re- and T. Kogaki, “Assessment of the offshore wind resource in trievals,” International Journal of Climatology, vol. 37, no. 2, Japan with the ASCAT microwave scatterometer,” Interna- pp. 870–889, 2017. tional Journal of Remote Sensing, vol. 40, no. 3, pp.1200–1216, [32] N. Ebuchi, “Intercomparison of four ocean vector wind 2019. products from OceanSat-2 scatterometer,” in Proceedings of [47] B. M. Castro, F. P. Brandini, M. Dottori, and J. F. Fortes, “A the 2013 IEEE International Geoscience and Remote Sensing Amazonia ˆ Azul: recursos e preservação,” Revista USP, Symposium—IGARSS 2013, Melbourne, Australia, 2013. vol. 113, pp. 7–26, 2017. [33] Z. Wang, A. Stoffelen, B. Zhang, Y. He, W. Lin, and X. Li, [48] M. S. Reboita, T. Ambrizzi, B. A. Silva, R. F. Pinheiro, and “Inconsistencies in scatterometer wind products based on R. P. Da Rocha, “+e south Atlantic subtropical anticyclone: ASCAT and OSCAT-2 collocations,” Remote Sensing of En- present and future climate,” Frontiers in Earth Science, vol. 7, vironment, vol. 225, pp. 207–216, 2019. p. 8, 2019. [34] M. C. Baptista, “An analysis of the surface wind field over the [49] X. Sun, K. H. Cook, and E. K. Vizy, “+e south Atlantic tropical and southern Atlantic ocean using data from the ERS subtropical high: climatology and interannual variability,” scatterometer,” Master’s thesis, INPE, San Jose dos Campos, Journal of Climate, vol. 30, no. 9, pp. 3279–3296, 2017. Brazil, 2000. [50] S. Hastenrath, Climate Dynamics of the Tropics, Vol. 8, [35] A. Bentamy, D. Croize-Fillon, and C. Perigaud, “Character- Springer Science & Business Media, Berlin, Germany, 1991. ization of ASCAT measurements based on buoy and [51] A. Melo, I. A. Cavalcanti, and P. Souza, “Zona de con- QuikSCAT wind vector observations,” Ocean Science, vol. 4, vergˆencia intertropical do Atlantico,” ˆ in Tempo e Clima no no. 4, pp. 265–274, 2008. BrasilOficina de Textos, Sao Paulo, Brazil, 2009. Advances in Meteorology 19 [52] J. M. Wallace and P. V. Hobbs, Atmospheric Science: An [68] S. R. Jammalamadaka and A. Sengupta, Topics in Circular Statistics, Vol. 5, World Scientific, Singapore, 2001. Introductory Survey, Vol. 92, Elsevier, Amsterdam, Nether- lands, 2006. [69] K. E. Taylor, “Summarizing multiple aspects of model per- formance in a single diagram,” Journal of Geophysical Re- [53] I. A. Cavalcanti and V. E. Kousky, “Frentes frias sobre o Brasil,” in Tempo e Clima no Brasil, pp. 135–147, Oficina de search: Atmospheres, vol. 106, no. 7, pp. 7183–7192, 2001. [70] M. P. Byrne, A. G. Pendergrass, A. D. Rapp, and Textos, Sao Paulo, Brazil, 2009. [54] B. J. Hoskins and K. I. Hodges, “A new perspective on K. R. Wodzicki, “Response of the intertropical convergence zone to climate change: location, width, and strength,” southern hemisphere storm tracks,” Journal of Climate, Current Climate Change Reports, vol. 4, no. 4, pp. 355–370, vol. 18, no. 20, pp. 4108–4129, 2005. [55] M. S. Reboita, M. A. Gan, R. P. d. Rocha, and T. Ambrizzi, [71] M. B. Switanek, P. A. Troch, C. L. Castro et al., “Scaled “Regimes of precipitation in south America: a bibliographical distribution mapping: a bias correction method that preserves review,” Brazilian Journal of Meteorology, vol. 25, no. 2, raw climate model projected changes,” Hydrology and Earth pp. 185–204, 2010. System Sciences, vol. 21, no. 6, pp. 2649–2666, 2017. [56] L. M. V. Carvalho, C. Jones, and B. Liebmann, “+e south [72] C. Anderson, H. Bonekamp, C. Duff, J. Figa-Saldana, and Atlantic convergence zone: intensity, form, persistence, and J. J. W. Wilson, “Analysis of ASCAT ocean backscatter relationships with intraseasonal to interannual activity and measurement noise,” IEEE Transactions on Geoscience and extreme rainfall,” Journal of Climate, vol.17, no.1, pp. 88–108, Remote Sensing, vol. 50, no. 7, pp. 2449–2457, 2012. [73] L. Zeng and R. A. Brown, “Scatterometer observations at high [57] W. T. Liu, K. B. Katsaros, and J. A. Businger, “Bulk pa- wind speeds,” Journal of Applied Meteorology, vol. 37, no. 11, rameterization of air-sea exchanges of heat and water vapor pp. 1412–1420, 1998. including the molecular constraints at the interface,” Journal [74] W. Lin, M. Portabella, A. Stoffelen, J. Vogelzang, and of the Atmospheric Sciences, vol. 36, no. 9, pp. 1722–1735, A. Verhoef, “ASCAT wind quality under high subcell wind variability conditions,” Journal of Geophysical Research: [58] S. D. Smith, “Coefficients for sea surface wind stress, heat flux, Oceans, vol. 120, no. 8, pp. 5804–5819, 2015. and wind profiles as a function of wind speed and temper- [75] J. C. May and M. A. Bourassa, “Quantifying variance due to ature,” Journal of Geophysical Research, vol. 93, no. C12, temporal and spatial difference between ship and satellite pp. 15467–15472, 1988. winds,” Journal of Geophysical Research: Oceans, vol. 116, [59] H. Hersbach, A. Stoffelen, and S. De Haan, “+e improved C-band geophysical model function CMOD5,” in Proceedings [76] C. Payan, “Improvements in the use of scatterometer winds in of the 2005 Envisat & ERS Symposium, Salzburg, Austria, the operational NWP system at Met´ ´ eo-France,” in Proceedings of the 2010 International Winds Workshop, vol. 10, pp. 22–26, [60] A. Stoffelen, J. A. Verspeek, J. Vogelzang, and A. Verhoef, Tokyo, Japan, 2010. “+e CMOD7 geophysical model function for ASCAT and [77] A. Verhoef and A. Stoffelen, ASCAT Wind Validation Report, ERS wind retrievals,” IEEE Journal of Selected Topics in Ap- Royal Netherlands Meteorological Institute, De Bilt, Neth- plied Earth Observations and Remote Sensing, vol. 10, no. 5, erlands, 2018. pp. 2123–2134, 2017. [78] J. Verspeek, A. Verhoef, and A. Stoffelen, ASCAT-C Wind [61] A. Verhoef, M. Portabella, and A. Stoffelen, “High-resolution Product Calibration and Validation, Royal Netherlands Me- ASCAT scatterometer winds near the coast,” IEEE Transac- teorological Institute, De Bilt, Netherlands, 2019. tions on Geoscience and Remote Sensing, vol. 50, no. 7, [79] S. Singh, R. K. Tiwari, H. S. Gusain, and V. Sood, “Potential pp. 2481–2487, 2012. applications of SCATSAT-1 satellite sensor: a systematic [62] A. Verhoef and A. Stoffelen, ASCAT Coastal Winds Valida- review,” IEEE Sensors Journal, vol. 20, no. 21, tion Report, Royal Netherlands Meteorological Institute, De pp. 12459–12471, 2020. Bilt, Netherlands, 2013. [80] H. Wang, J. Zhu, M. Lin, Y. Zhang, and Y. Chang, “Evaluating [63] R. Atlas, R. N. Hoffman, J. Ardizzone et al., “A cross-calibrated, Chinese HY-2B HSCATocean wind products using buoys and multiplatform ocean surface wind velocity product for mete- other scatterometers,” IEEE Geoscience and Remote Sensing orological and oceanographic applications,” Bulletin of the Letters, vol. 17, no. 6, pp. 923–927, 2020. American Meteorological Society, vol. 92, no. 2, pp. 157–174, [81] R. E. N. Yueming, LM-4B Successfully Launched HY-2C Satellite, China Aerospace Science and Technology Corpo- [64] F. J. Wentz, J. Scott, R. Hoffman, M. Leidner, R. Atlas, and ration, Beijing, China, 2020. J. Ardizzone, “Remote sensing systems cross-calibrated multi- platform (CCMP) 6-hourly ocean vector wind analysis product on 0.25deg grid,” 2015, https://www.remss.com/ measurements/ccmp/. [65] R. Atlas, R. N. Hoffman, S. C. Bloom, J. C. Jusem, and J. Ardizzone, “A multiyear global surface wind velocity dataset using SSM/I wind observations,” Bulletin of the American Meteorological Society, vol. 77, no. 5, pp. 869–882, 1996. [66] R. N. Hoffman, “SASS wind ambiguity removal by direct minimization,” Monthly Weather Review, vol. 110, no. 5, pp. 434–445, 1982. [67] R. N. Hoffman, “SASS wind ambiguity removal by direct minimization. Part II: use of smoothness and dynamical constraints,” Monthly Weather Review, vol. 112, no. 9, pp. 1829–1852, 1984. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Advances in Meteorology Hindawi Publishing Corporation

Comparison of Multiple Surface Ocean Wind Products with Buoy Data over Blue Amazon (Brazilian Continental Margin)

Loading next page...
 
/lp/hindawi-publishing-corporation/comparison-of-multiple-surface-ocean-wind-products-with-buoy-data-over-SWDM938iMS
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2021 Vitor Paiva et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1687-9309
eISSN
1687-9317
DOI
10.1155/2021/6680626
Publisher site
See Article on Publisher Site

Abstract

Hindawi Advances in Meteorology Volume 2021, Article ID 6680626, 19 pages https://doi.org/10.1155/2021/6680626 Research Article Comparison of Multiple Surface Ocean Wind Products with Buoy Data over Blue Amazon (Brazilian Continental Margin) 1 1 2 Vitor Paiva , Milton Kampel , and Rosio Camayo Earth Science Coordination, Earth Observation and Geoinformatics, National Institute for Space Research, Sao Jose dos Campos 12227-010, Brazil Earth Science Coordination, Numerical Modelling of the Earth System, National Institute for Space Research, Cachoeira Paulista 12630-000, Brazil Correspondence should be addressed to Milton Kampel; milton.kampel@inpe.br Received 10 December 2020; Revised 24 May 2021; Accepted 31 August 2021; Published 14 September 2021 Academic Editor: Maria Angeles Garc´ıa Copyright © 2021 Vitor Paiva et al. +is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Remote sensing data for space-time characterization of wind fields in extensive oceanic areas have been shown to be increasingly useful. Orbital sensors, such as radar scatterometers, provide data on ocean surface wind speed and direction with spatial and temporal resolutions suitable for multiple applications and air-sea studies. Even considering the relevant role of orbital scat- terometers to estimate ocean surface wind vectors on a regional and global scale, the products must be validated regionally. Six different ocean surface wind datasets, including advanced scatterometer (ASCAT-A and ASCAT-B products) estimates, nu- merical modelling simulations (BRAMS), reanalysis (ERA5), and a blended product (CCMP), were compared statistically with in situ measurements obtained by anemometers installed in fifteen moored buoys in the Brazilian margin (8 buoys in oceanic and 7 in shelf waters) to analyze which dataset best represents the wind field in this region. +e operational ASCAT wind products −1 −1 presented the lowest differences in wind speed and direction from the in situ data (0.77ms <RMSE <1.59ms , spd −1 −1 ° ° 0.75 < R <0.96, −0.68ms <bias <0.38ms , and 12.7 <RMSE <46.8 ). CCMP and ERA5 products also performed well spd spd dir −1 −1 in the statistical comparison with the in situ data (0.81ms <RMSE <1.87ms , 0.76 < R <0.91, spd spd −1 −1 ° ° −1.21ms <bias <0.19ms , and 13.7 <RMSE <46.3 ). +e BRAMS model was the one with the worst performance spd dir −1 (RMSE >1.04m·s , R <0.87). For regions with a higher wind variability, as in the southern Brazilian continental margin, spd spd wind direction estimation by the wind products is more susceptible to errors (RMSE >42.4 ). +e results here presented can be dir used for climatological studies and for the estimation of the potential wind power generation in the Brazilian margin, especially considering the lack of availability or representativeness of regional data for this type of application. forecasting, navigation, marine engineering, and offshore 1. Introduction wind energy [1, 12, 13]. +e vast extension of the global ocean imposes technical Ocean surface wind is one of the main drivers of several oceanic, atmospheric, and climate processes, thus being an and financial limitations on in situ ocean surface winds sampling, which is necessary and valuable in remote sensing important indicator of climate change [1]. Ocean surface wind vector variability is intrinsically linked to that of calibration and validation process. Direct measurements are oceanic processes, such as coastal upwelling [2, 3], primary acquired by anemometers on moored buoys, research cruises, productivity [4, 5], deep water formation [6, 7], advective and vessels of opportunity, or by light detection and ranging volume transport, air-sea momentum, and heat fluxes (Lidar) sensors, providing precise and accurate data. How- [8–11]. Having the ability to measure wind speed and di- ever, these techniques are limited to point measurements that rection with high accuracy becomes essential to these are not able to provide satisfactory spatial and temporal processes’ studies, also considering the impacts on weather coverage to resolve variability at different scales [14]. 2 Advances in Meteorology and correlations) between wind products and in situ mea- Today, more than 10 global and regional atmospheric reanalysis and wind products are available for the scientific surements. However, significant systematic differences do exist in the wind products derived from different scatter- community [15]. Numerical prediction models are capable of simulating ocean surface winds on a global scale but ometers [20, 30, 31] and from different producers [32], depend on the existing knowledge about the physical pro- which may result in detrimental impacts on different ap- cesses and on the availability of data used to parametrize the plications, if unknown. +erefore, characterizing errors and initial conditions of the simulations [16, 17]. In weather inconsistencies among different wind products is important, prediction, the ability to make a skilful forecast requires both not only for guiding wind product users, but also to motivate realistic representation of the atmosphere by numerical research by the producers to solve inconsistencies problems. +e Brazilian oceanic area represented by the “Territorial models and accurate initial conditions [18]. Reanalysis products integrate simulated data from nu- Sea” and the “Exclusive Economic Zone” corresponds to approximately 3.6 million km merical prediction models with atmospheric, oceanic, and . If the “Extended, External, other measurements, but with relatively low spatial reso- Jurisdictional” or “Legal Continental Shelf” claimed to the lutions (250 to ∼31km). +is can be insufficient to ade- Commission on the Limits of the Continental Shelf of the quately represent small and mesoscale meteorological United Nations in three “Revised Partial Proposals” for the features and accurately characterize local wind regimes [19]. outer limit of the Brazilian Continental Shelf (2.1 million Offshore wind blended products usually consider a first km ) is added to this area, the total oceanic area called the guess of the wind vector as the background and assimilate “Blue Amazon” may be approximately 5.7 million km . +is multiple sources of measured wind data from buoys, ships, immense oceanic area contains an enormous amount of and satellites [20]. economically, socially, and strategically relevant resources Orbital remote sensors provide wind data on a global and is extremely important for climate stability and envi- scale, covering remote regions normally characterized by the ronmental quality of the adjacent coast [47]. availability of sparse in situ data, with relatively good ac- +e South Atlantic Subtropical Anticyclone (SASA) is curacy and high spatial and temporal resolution [21]. Ocean the main feature of the atmospheric circulation over the surface wind vectors with high spatial and temporal reso- South Atlantic Ocean, with a seasonal variability extending lution obtained by spaceborne scatterometers are utilized in over south-eastern Brazil during the austral winter and various fields of meteorology, oceanography, and climate retracting eastward during the summer [48]. SASA con- studies [22, 23]. Surface wind and stress fields derived from tributes with northeast winds to the Brazilian continental scatterometer observations can be applied to drive ocean margin that are more constant during austral spring and circulation models on various scales and can also be as- summer [49]. similated into regional and global numerical weather pre- Over the Tropical Atlantic, the Intertropical Conver- diction models [21, 24]. +ese data have also been used to gence Zone (ITCZ) associated with the confluence zone of infer about offshore wind energy potential [20, 25]. the northeast and southeast trade winds migrates seasonally Scatterometers measure the radar cross section of the from its most northerly position, around ∼10 N in August- ocean surface, and numerical inversion of the geophysical September, to its southerly position, around ∼2 S in March- model function (GMF) yields the scatterometer wind April in the Atlantic Ocean [50, 51]. Associated with the measurement. +e present constellation of scatterometers mesoscale oscillations, the frontal systems move with cy- maps the surface wind field globally, with a typical spatial clones and post-frontal anticyclones, changing the fields of resolution of 25–12.5km [26], and has been successfully atmospheric pressure, wind, and other atmospheric vari- ° ° used in weather forecasting applications [27], long-term ables along their trajectory [52]. Between 30 S and 50 S, the climate studies [28], and air-sea interactions [29]. +e main zones with the highest occurrence of cold fronts are con- limitations of scatterometers are contamination by rain centrated on the South-Southwestern Atlantic, mainly in the (depending on the frequency of the transmitted signal), lack coastal areas of Argentina and Uruguay [53]. +e south and of data near the coast (typically within 15km), and poor south-eastern coasts of Brazil and the region between the temporal sampling [21]. extreme south of Brazil and the coast of Argentina are Although the number of space-borne scatterometers is favourable regions for the formation of cyclones [54]. +ese increasing in recent years, regional validation of wind systems cause strong winds, precipitation and decrease in products from different scatterometers [30, 31] and pro- temperature [55]. In this region, the atmospheric circulation ducers [32] remains a challenge [33]. Moored ocean buoys is characterized, on a synoptic scale, mainly by the move- provide the absolute calibration reference for satellite wind ment of transient systems such as cyclonic vortices, frontal retrievals [22]. While the development of GMF and wind systems, and the South Atlantic Convergence Zone (SACZ) retrieval algorithms rely on many inputs (numerical models, [56]. wind retrievals for other satellites, statistical constraints, For better understanding the wind regimes on the etc.), finalized satellite wind retrievals always need to be Brazilian continental margin, and to aid in decision-making verified by comparisons with buoys. for the installation of offshore wind turbines, there is a need Several studies evaluated the accuracy of wind products to conduct a study that contemplates this vast oceanic area. over the ocean comparing them with in situ measurements +e dataset used must be adequate to characterize the spatial globally and regionally [19, 24, 34–46]. In general, these and temporal variability of the local wind fields, and also to authors observed good agreement (mean square error, bias, estimate the potential offshore wind energy production. Advances in Meteorology 3 10°N +is study aims to evaluate and compare different al- ternative sources of offshore wind data, including the Ad- vanced SCATterometer (ASCAT) on board MetOP-A (ASCAT-A) and MetOP-B (ASCAT-B) European satellites, the Cross-Calibrated Multi-Platform wind product, the Eu- 0° ropean Centre for Medium-Range Weather Forecasts ERA5 Reanalysis, and numerical simulations from the Brazilian Regional Atmospheric Modelling System (BRAMS/CPTEC) with in situ ocean surface wind measurements, aiming to 10°S analyze the capacity of these wind data sources in describing the wind field in the Blue Amazon region. 2. Materials and Methods 20°S 2.1. Buoy Wind Data. Time series of in situ wind data measured by anemometers installed in 15 oceanographic buoys moored on the Brazilian continental shelf and in 30°S deeper oceanic waters were obtained from the Brazilian National Buoy Program (PNBOIA) and the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA) project, respectively. +e geographical locations and sum- marized information regarding these buoys are presented in 40°S 60°W 50°W 40°W 30°W 20°W Figure 1 and Table 1, respectively. Longitude Eight PIRATA buoys used in this study are moored at ° ° depths between 1,500 and 6,000m and are equipped with a Figure 1: Map of the study area between 10 N and 40 S and be- ° ° tween 20 W and 60 W, showing the geographical locations of the 15 R. M. Young mechanical anemometer, model 05103 (http:// buoys used for in situ measurements of wind speed and direction. www.youngusa.com/products/11/5.html), or with a Gill Symbols denote different organizational affiliations and domains: WindSonic ultrasonic anemometer (http://gillinstruments. circle, Brazilian National Buoy Program (PNBOIA), shelf waters; com/data/datasheets/windsonic-1405-027-iss7.pdf), installed triangle, Prediction and Research Moored Array in the Tropical 4m abovesea level.Wind direction is measuredwith a Sparton Atlantic (PIRATA), oceanic waters. SP3004D compass (https://sparton.com/) or an EG&G flux- gate compass model 63764 or KVH LP101-5 (https://www. pmel.noaa.gov/gtmba/sensor-specifications). Data with a 10- which in turn is calculated from an iterative process using minute temporal resolution, containing measurements since the wind speed measured by the buoys. For more details, 1997, are available from NOAA’s Pacific Marine Environ- please refer to References [57, 58]. mental Laboratory (https://www.pmel.noaa.gov/tao/drupal/ disdel/). Seven PNBOIA buoys used in this study are moored 2.2. Scatterometer Wind Data. Ocean surface wind data close to the continental shelf break (∼200m isobath). +e estimated by orbital scatterometers are derived from the PNBOIA data, maintained by the Brazilian Navy Hydrog- ASCAT sensor, installed on board MetOp-A and MetOp-B raphy Centre, are available on the Internet (https://www. satellites. Both satellites have been in operation since 10/2006 marinha.mil.br/chm/dados-do-goos-brasil/pnboia-mapa). and 09/2012, respectively, until the present. MetOp-C data Time series of wind data are provided with hourly time were not used due to the lack of concomitant buoy data resolution. +e temporal coverage of the data ranges from during the development of this study. Wind data were ob- 04/2009 to 11/2018. +e buoys contain 2 anemometers, one tained from the EUMETSAT Ocean and Sea Ice Satellite mechanical installed at 4.7m above sea level (used as the Application Facility (OSI SAF) through the Royal Nether- main reference) and another ultrasonic sensor installed at lands Meteorological Institute (KNMI). +ree ASCAT 3.7m. More information about the sensors of each buoy can products were acquired for this study: (i) ASCAT-A at be found at the same Internet address for data access. processing level 2 (L2), with 12.5km of spatial resolution, To adjust the wind measured by the buoys to 10m, a from the period 08/2010 to 10/2018 (https://projects.knmi.nl/ steady-state neutrally stable logarithmic vertical wind profile scatterometer/ascat_osi_co_prod/ascat_app.cgi); (ii) ASCAT- relation was considered. In this case, the wind speed (U) at B, at L2 processing level, spatial resolution of 12.5km, from height z is given by the period 10/2012 to 10/2018 (https://projects.knmi.nl/ scatterometer/ascat_b_osi_co_prod/ascat_app.cgi); and (iii) u z U(z) � ln􏼠 􏼡, (1) ASCAT Coastal Wind Data Record (ASCAT-CDR), L2 k z processing level, 12.5km spatial resolution, from the period where k is the von Karm ´ an ´ constant (k �0.41) and z is the 01/2007 to 03/2014 (https://projects.knmi.nl/scatterometer/ surface roughness that depends on the friction velocity (u ), ascat_cdr/). Latitude 4 Advances in Meteorology Table 1: Main characteristics of in situ measurement buoys. Dataset Buoy Location Distance to the coast (km) Time coverage ° ° ° 8N38 W 8 N; 38 W 1219 1998-01 2014-11 ° ° ° 4N23 W 4 N; 23 W 1082 2006-06 2017-03 ° ° ° 4N38 W 4 N; 38 W 788 2000-03 2013-10 ° ° ° 0N23 W 0 N; 23 W 1491 1999-03 2016-03 PIRATA (OW) ° ° ° 0N35 W 0 N; 35 W 563 1998-01 2015-10 ° ° ° 8S30 W 8 S; 30 W 531 2005-08 2015-12 ° ° ° 14S32 W 14 S; 32 W 614 2005-08 2016-01 ° ° ° 19S34 W 19 S; 34 W 562 2005-09 2014-08 ° ° Fortaleza 3.21 S; 38.40 W 54 2016-11 2017-12 ° ° Recife 8.15 S; 34.56 W 35 2012-09 2016-04 ° ° Porto Seguro 15.99 S; 37.94 W 99 2012-07 2016-12 ° ° PNBOIA (SW) Cabo Frio 2 23.63 S; 42.20 W 72 2016-07 2017-07 ° ° Santos 25.44 S; 45.04 W 165 2011-04 2017-08 ° ° Itaja´ı 28.50 S; 47.39 W 125 2009-04 2018-11 ° ° Rio Grande 31.57 S; 49.86 W 101 2009-04 2017-01 OW: oceanic waters; SW: shelf waters. Wentz et al. [64]. +is dataset uses the variational analysis +e ASCAT-A and ASCAT-B products were processed using the CMOD-5n geophysical model function [59] until method (VAM) to fill gaps in the gridded product [65–67]. September 2018. From then on, the GMF CMOD-7 function VAM generates a surface wind grid, minimizing an objective [60] started to be used (Verhoef, personal communication). function, which measures the mismatch of the in situ and +e ASCAT-CDR product is the result of reprocessing satellite wind data overlaid on reanalysis bottom base, ASCAT-A data using the latest GMF CMOD-7. +e three considered as an initial estimate [63]. Wind data at 10m products, regardless of the GMF used, are optimized for from the CCMP product are available in regular grids for coastal regions [23, 61, 62], containing wind data at 10m dates after 1987 with a spatial resolution of 25km and a above sea level in along-track grids (along the satellite path) temporal resolution of six hours. Data were obtained for the with a spatial resolution of 12.5km, coming from 14 orbital period from 01/1998 to 12/2018 (http://www.remss.com/ cycles per day. measurements/ccmp/). Table 2 summarizes the main characteristics of all al- ternative offshore wind products used in this study. 2.5. ERA5 Reanalysis. +e ERA5 Reanalysis product is de- veloped by the Copernicus Climate Change Service, 2.3. BRAMS Numerical Forecasting Model. +e Brazilian implemented by the European Centre for Medium-Range Regional Atmospheric Modelling System (BRAMS) is a Weather Forecasts (ECMWF) (https://www.ecmwf.int/en/ numerical simulation system designed for atmospheric forecasts/datasets/archive-datasets/reanalysis-datasets/ forecasting and research from regional to hemispheric era5). Reanalysis data are produced by combining short- scales. Based on the Regional Atmospheric Modelling Sys- term numerical predictions with observational data. ERA5 is tem (RAMS), originally developed at Colorado State Uni- produced using the assimilation of atmosphere and surface versity, BRAMS was until recently developed and data in four dimensions through the ECMWF Integrated maintained by the Centre for Weather Forecasting and Forecast System with 137 vertical levels in sigma coordinates Climate Studies (CPTEC) of the Brazilian National Institute (which use atmospheric pressure on the surface as a ref- for Space Research (INPE), the University of São Paulo, and erence) in grids with 31km of resolution for atmospheric other institutions in Brazil and abroad. +e computational levels. Hourly ERA5 data are available for several atmo- code of BRAMS is free software with General Public License spheric variables, including wind zonal and meridional (CC-GPL). More information about BRAMS can be found at components, and air temperature at 10m above sea level. http://brams.cptec.inpe.br/. BRAMS simulations for Monthly updates of ERA5 hourly data on single levels from 2013–2015 were provided by CPTEC/INPE, containing wind 1979 to present are published within 3 months of real time. data at 10m in grids with 5km of spatial resolution and Data were obtained for the period from 01/2000 to 12/2017. hourly temporal resolution. +e coverage area is between ° ° ° ° 7.6 N–37.3 S and 62.5 W–29.6 W. 2.6. Statistical Analysis. To compare the different sources of 2.4. Blended CCMP Data. +e Cross-Calibrated Multi- offshore wind speed data with the in situ buoy measure- Platform (CCMP) is a level 3 (L3) wind product which uses ments, the root mean square error (RMSE ), bias , spd spd satellite and in situ data, and numerical modelling simu- Pierson correlation coefficient (R ), and standard deviation spd lations. +e first CCMP version, described in the study of (σ ) were calculated considering the simultaneous and spd Atlas et al. [63], was discontinued in 2012. +e second valid wind speed data records among all datasets, according CCMP version used in this work is described in the study of to the following equations: Advances in Meteorology 5 Table 2: Main characteristics of the considered wind products. Wind Type of Processing Spatial Temporal Time Included data sources product dataset level resolution resolution coverage ASCAT-A 3–5 swaths/ Satellite ASCAT/MetOP-A L2B 12.5km 2007–present Coastal day ASCAT-B 3–5 swaths/ Satellite ASCAT/MetOP-B L2B 12.5km 2012–present Coastal day ASCAT- 3–5 swaths/ Satellite ASCAT/MetOP-A L2B 12.5km 2007–2014 CDR day BRAMS NWP Forcing data from reanalysis — 5km Hourly 2013–2015 ECMWF analyses, QuikSCAT, ASCAT/ CCMP V2.0 Blended MetOP-A, WindSat, SSM/I, SSMIS, AMSR-E, L3 0.25lat/lon 6 hours 1987–present AMSR2, GMI ERA5 Reanalysis NWP and observations — 31km Hourly 1979–present 􏽶�������������� � 􏽐 sin􏼐θ − θ 􏼑sin􏼐θ − θ 􏼑 r,i r e,i e N i�1 1 2 􏽱�������������������������� R � . dir (8) (2) 2 2 RMSE � 􏽘 􏼐x − x 􏼑 , N spd e,i r,i 􏽐 sin 􏼐θ − θ 􏼑sin 􏼐θ − θ 􏼑 r,i r e,i e N i�1 i�1 Please, refer to the study of Jammalamadaka and Sen- gupta [68] for more information about circular statistics. +e bias � 􏽘􏼐x − x 􏼑, (3) spd e,i r,i bias and RMSE formulas for wind direction data are the i�1 same as for wind speed data. However, the minor angle difference (θ ) was derived as follows: x − x x − x 􏽐 􏼐 􏼑􏼐 􏼑 i�1 e,i e r,i r 􏽱������������� �􏽱������������� � ° ° R � , (4) spd ⎧ ⎪ ⎫ ⎪ θ − θ , if − 180 <􏼐θ − θ 􏼑 <180 2 2 ⎪ e,i r,i e,i r,i ⎪ N N ⎪ ⎪ ⎪ ⎪ 􏽐 􏼐x − x 􏼑 􏽐 􏼐x − x 􏼑 i�1 e,i e i�1 r,i r ⎨ ⎬ θ � θ − θ − 360, if 􏼐θ − θ 􏼑 >180 . d,i e,i r,i e,i r,i ⎪ ⎪ 􏽶������������ � ⎪ ⎪ ⎪ ⎪ 􏽴 ⎪ ⎪ ⎩ ° ⎭ θ − θ + 360, if 􏼐θ − θ 􏼑 < − 180 e,i r,i e,i r,i (5) σ � 􏽘 x − x􏼁 , spd i (9) i�1 +erefore, RMSE and bias for direction can be read as where x refers to the wind speed values at instant i, r refers to follows: 􏽶���������� the in situ buoy data used as a reference, and e refers to the data estimated by the alternative sources of wind data. 1 (10) RMSE � 􏽘 􏼐θ 􏼑 , dir d,i Regarding wind direction data, the circular statistics was i�1 applied. +e mean direction (θ) is estimated as follows: − 1 ⎧ ⎪ ⎫ ⎪ N ⎪ tan , if C >0, S >0 ⎪ ⎪ ⎪ ⎪ ⎪ bias � 􏽘 􏼐θ 􏼑. (11) ⎪ ⎪ dir d,i ⎪ ⎪ ⎪ N ⎪ ⎪ i�1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎬ −1 +e comparisons considered the buoys geographic co- θ � . (6) tan + 180 , if C <0 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ordinates and the respective grid point or pixel of the other ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ datasets. Each buoy is identified by its name (PNBOIA, shelf ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ waters) or geographic coordinates (PIRATA, oceanic wa- −1 ⎩ ⎭ tan + 360 , if C >0, S <0 ters), as adopted by the respective monitoring programs. An interval of 30 minutes was considered around the mea- Considering each angle (θ) in an instant i as a vector, surement time of the in situ PNBOIA data, and of 5 minutes N N C � 􏽐 cos(θ ) and S � 􏽐 sin(θ ) are the respective x- i�1 i i�1 i in the case of PIRATA buoys, according to the respective and y-axis representations of the resultant vector length (r). temporal resolution of the in situ data recording (hourly or +e circular standard deviation (σ ) is estimated by the dir 10 minutes, respectively). Two approaches were taken for following equation: data comparison, prioritizing different criteria: (i) the first 􏽰������ approach compared data that were obtained simultaneously (7) σ � 2 ln(r), dir by all the five datasets—ASCAT, CCMP, ERA5, BRAMS, √������ 2 2 where r � (r/N) � ( C + S /N) would be the mean re- buoys—(ii) a second approach compared all available paired data separately for each buoy. When comparing sultant length. +e circular correlation (R ) is defined as dir 6 Advances in Meteorology −1 simultaneous data (i), the Taylor diagrams [69] were pro- −0.48m·s for BRAMS and ERA5. +e highest correlations duced, and ASCAT-A, ASCAT-B, and ASCAT-CDR data (R >0.87) were obtained for ASCATproducts considering spd were considered separately, due to the difficulty of matching all and SW buoys, or CCMP considering OW buoys. them over time. Wind roses diagrams were also generated. In general, relatively higher differences were observed in relation to wind direction, especially for shelf water buoys. Best results were obtained for OW with RMSE values dir 3. Results ° ° ° <23.2 , bias between −0.09 and 0.19 , and correlations dir >0.83, except for the BRAMS product (RMSE �31.2 , 3.1. In Situ Buoy Data. Wind roses of oceanic water buoys dir bias � −1.40 , R �0.72, in the best cases). Considering all (PIRATA) show a low variability in the wind direction, dir dir buoys, the values of RMSE and bias varied between 46.8 dir dir except for the buoys moored in the north-south oscillation ° ° ° and 53.4 , and between −1.06 and 4.88 , respectively. Also, region of the Intertropical Convergence Zone [44, 70] the best correlations (R >0.87) were obtained for the (Figure 2). +e northernmost buoy (8N38W) has dominant dir CCMP product. On the contrary, high RMSE values NE winds and the buoys 4N23W and 4N38W have two dir ° ° ° >73.5 , bias between −14.34 and −0.09 and lower cor- predominant wind directions: NNE–SSE and NE–SE, re- dir relations (<0.35) were obtained for SW. spectively. +ese buoys also have relatively higher standard ° ° +e comparisons of all available paired data for each deviations in wind direction (52.6 and 82.1 , Table 3). +e oceanic water buoy did not differ much with respect to wind 0N35W and 08S30W buoys have dominant winds from ESE speed (Figure 6 and Table S2 in Supplemental File). +e and the 0N23W buoy from SE. Buoys 14S32W and 19S34W −1 lower RMSE values varied between 0.77m·s (ASCAT-A, have dominant E-SE and E-NE winds, respectively. +e spd −1 0N23W) and 1.13m·s (ASCAT-B, 19S34W), except for higher mean wind speeds were observed for buoys 8S30W −1 −1 −1 −1 BRAMS (1.04m·s to 1.90m·s ), with bias between spd (7.36m·s ) and 14S32W (6.92m·s ). −1 −1 −0.73m·s (ERA5, 4N23W) and 0.43m·s (ASCAT-B, For shelf water buoys PNBOIA, the higher mean wind −1 8N38W), and correlation coefficients between 0.61 (BRAMS, speeds were observed in Fortaleza (8.91m·s ), Rio Grande −1 −1 −1 4N38W) and 0.92 (ASCAT-A, 8N38W). Regarding wind (8.34m·s ), Itaja´ı (7.68m·s ), and Cabo Frio 2 (7.52m·s ) direction, the lower RMSE values varied between 13.7 (Figure 2, Table 3). Fortaleza, Recife, and Cabo Frio 2 buoys dir (CCMP, 8S30W) and 32.5 (ASCAT-B and ASCAT-CDR, showed a lower variability in wind direction, with domi- 4N23W), with bias between −9.18 (ASCAT-A, 4N38W) nance from the SE-E and NE quadrants, respectively. dir and 19.5 (BRAMS, 19S34W), and correlation coefficients However, in Cabo Frio 2, the standard deviation of the mean ° between 0.57 (BRAMS, 0N35W) and 0.92 (CCMP, 4N38W). wind direction was 75.0 (Table 3), while for Fortaleza and ° ° In general, the lower RMSE and bias values and the higher R Recife, the standard deviations were lower (23.1 and 33.3 , values, both for wind speed and direction, were obtained for respectively). the ASCAT and CCMP products. +e comparisons of all available paired data for each 3.2. Ocean Surface Wind Statistical Comparisons. shelf water buoy did not differ much with respect to wind +roughout the statistical comparisons presented below, a speed (Figure 7 and Table S3 in Supplemental File). +e −1 negative systematic bias was observed for the BRAMS wind lower RMSE values varied between 0.78m·s (ASCAT-A, spd −1 speed data, which was corrected by applying the scaled Cabo Frio 2) and 1.09m·s (ASCAT-B, Itaja´ı), with bias spd −1 −1 distribution mapping method [71]. For oceanic waters (OW) between −1.21m·s (ERA5, Fortaleza) and −0.004m·s buoys comparisons, the correction of the systematic bias (ASCAT-B, Rio Grande), and R between 0.51 (BRAMS, spd −1 resulted in lower RMSE Recife) and 0.96 (ASCAT-A, Fortaleza, and ASCAT-B, Cabo values (a decrease of 1.8m·s in spd the best case), while for shelf waters (SW) buoys, there was Frio 2). Regarding wind direction, the lower RMSE values dir ° ° no relevant change with slightly higher RMSE values for varied between 12.7 (ASCAT-B, Fortaleza) and 79.6 spd two buoys (Figure S1 in Supplemental File). +erefore, (CCMP, Santos), with bias between −28.5 (BRAMS, dir BRAMS wind speed data with systematic bias correction Santos) and 40.5 (BRAMS, Rio Grande), and circular were used only for comparisons with OW buoys. correlation coefficients between 0.58 (CCMP and ERA5, +e simultaneous comparisons of the ocean surface wind Itaja´ı) and 0.94 (ASCAT-B, Fortaleza), except for compar- products with in situ measured buoys data did not differ isons with the Santos buoy (≤0.34). much with respect to wind speed (Figures 3–5 and Table S1 in Supplemental File). Considering the comparison of all buoys together, the RMSE values ranged between 3.3. Error Dependence on Measured Wind Speed. spd −1 −1 Scatterometers limitations in representing low and high 0.98m·s and 1.82m·s . Considering the OW and SW wind speeds are well documented in the literature [72, 73]. buoys separately, the RMSE values were relatively lower spd −1 −1 (0.91m·s and 1.02m·s in the best cases, respectively). +e To analyze if wind speeds from the alternative wind data sources show any variation or dependence with the in situ highest values of RMSE were obtained for the product spd BRAMS, even after removing the systematic bias. For all buoy data, four different bins were considered for statistical −1 −1 −1 −1 −1 comparison: 3–6m·s ; 6–9m·s ; 9–12m·s and >12m·s buoys, the bias varied between ±0.30m·s . For OW spd −1 buoys, the bias varied between ±0.19m·s . For SW, the (Figures 8 and 9 and Table S4 in Supplemental File). In spd −1 −1 −1 general, the lowest RMSE (0.76m·s ) and bias bias varied between −0.18m·s and −0.25m·s for spd spd spd −1 −1 ASCAT and CCMP products and between −0.70m·s and (−0.05m·s ), and highest R values (0.70) were observed spd Advances in Meteorology 7 N N N 38 % 25 % 20 % NW NE NW NE NW NE 30 20 22 15 15 10 8 5 W E W EE W SW SE SW SE SW SE S S S N N N 38 % 40 % 50 % NW NE NW NE NW NE 30 40 22 30 15 20 8 10 W E W EE W SW SE SW SE SW SE S S S N N N 38 % 20 % 40 % NW NE NW NE NW NE 15 30 10 20 5 10 W E W EE W SW SE SW SE SW SE S S S N N N 30 % 9 % 30 % NW NE NW NE NW NE 6 20 3 10 W E W EE W SW SE SW SE SW SE S S S N N N 12 % 9 % 9 % NW NE NW NE NW NE 6 6 3 3 W EE W W E SW SE SW SE SW SE S S S –1 Wind Speed (m.s ) 0 - 3 6 - 9 12 - 15 3 - 6 9 - 12 > 15 Figure 2: Wind roses of oceanic (PIRATA) and shelf (PNBOIA) water buoys. Santos Recife 14S32W 0N35W 8N38W 25.44°S; 45.04°W 8.15°S; 34.56°W 14°S; 32°W 0°N; 23°W 8°N; 38°W Itajaí Porto Seguro 19S34W 0N23W 4N38W 28.50°S; 47.39°W 15.99°S; 37.94°W 19°S; 34°W 0°N; 35°W 4°N; 23°W Rio Grande Cabo Frio 2 Fortaleza 8S30W 4N23W 31.57° S; 49.86°W 2 23.63° S; 42.20°W 3.21°S; 38.40°W 8°S; 30°W 4°N; 38°W 1.0 1.0 1.0 0.99 0.99 0.99 0.95 0.95 0.95 0.9 0.9 0.9 0.8 0.8 0.8 Correlation Correlation Correlation 0.7 0.7 0.7 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 8 Advances in Meteorology Table 3: Mean (±standard deviation) and maximum and minimum wind speed and circular mean (±circular standard deviation) wind direction from the in situ measurement buoys. Mean Max Min Dataset Buoy −1 −1 −1 Speed (ms ) Dir. ( ) Speed (ms ) Speed (ms ) 8N38W 6.78 ±2.60 244 ±60.0 20.6 0.00 4N23W 5.38 ±2.17 307 ±82.1 32.3 0.00 4N38W 6.31 ±2.21 274 ±52.6 17.9 0.00 0N23W 5.67 ±1.89 306 ±35.5 19.0 0.00 PIRATA (OW) 0N35W 6.27 ±2.07 288 ±38.2 17.1 0.00 8S30W 7.36 ±1.70 290 ±20.5 16.3 0.00 14S32W 6.92 ±1.90 276 ±28.1 17.1 0.00 19S34W 6.15 ±2.17 254 ±52.3 16.5 0.00 Fortaleza 8.91 ±2.19 287 ±23.1 16.0 0.43 Recife 6.53 ±1.76 298 ±33.3 47.4 0.11 Porto Seguro 5.53 ±2.42 280 ±111 16.1 0.11 PNBOIA (SW) Cabo Frio 2 7.52 ±3.19 264 ±75.0 16.3 0.11 Santos 6.88 ±3.07 290 ±102 20.1 0.11 Itaja´ı 7.68 ±3.41 242 ±146 36.3 0.11 Rio Grande 8.34 ±3.58 218 ±113 48.9 0.11 OW: oceanic waters; SW: shelf waters. 0.0 0.0 0.0 0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0 6 12 18 24 30 36 42 48 54 Standard deviation Standard deviation Standard deviation PIRATA Bias PIRATA Bias PIRATA Bias CCMP 0.08 CCMP 0.10 CCMP 1.12 ERA5 -0.08 ERA5 0.54 ERA5 -0.09 BRAMS 0.02 BRAMS 0.04 BRAMS 7.92 ASCAT 0.17 ASCAT 0.29 ASCAT 0.19 RMSE RMSE RMSE (a) (b) (c) Figure 3: Continued. 0.6 0.6 1.2 1.2 8 1.0 1.0 1.0 1.0 1.0 1.0 0.99 0.99 0.99 0.99 0.99 0.99 0.95 0.95 0.95 0.95 0.95 0.95 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 Correlation Correlation Correlation Correlation Correlation Correlation 0.7 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.2 Advances in Meteorology 9 0.0 0.0 0.0 0 6 12 182430364248 54 0.0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7 0 8 16 24 32 40 48 56 Standard deviation Standard deviation Standard deviation PIRATA Bias PIRATA Bias PIRATA Bias CCMP 0.19 CCMP 0.10 CCMP 1.78 ERA5 -0.30 ERA5 -0.07 ERA5 1.27 BRAMS 6.34 BRAMS 0.02 BRAMS 7.96 ASCAT 0.40 ASCAT 0.18 ASCAT 0.83 RMSE RMSE RMSE (d) (e) (f) Figure 3: Taylor diagrams for statistical comparisons between CCMP, BRAMS, ERA5, and ASCAT-A (a, b), or ASCAT-B (c, d), or ASCAT- CDR (e, f) that have coincident records with oceanic water buoy data. (a) Speed-ASCAT-A. (b) Direction-ASCAT-A. (c) Speed-ASCAT-B. (d) Direction-ASCAT-B. (e) Speed-ASCAT-CDR. (f) Direction-ASCAT-CDR. 0.0 0.0 0.0 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 0 15 30 45 60 75 90 105 120 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 Standard deviation Standard deviation Standard deviation PNBOIA Bias PNBOIA Bias PNBOIA Bias CCMP -0.24 CCMP -6.15 CCMP -0.25 ERA5 -0.69 ERA5 -6.87 ERA5 -0.70 -0.49 -1.37 -0.60 BRAMS BRAMS BRAMS ASCAT -0.23 ASCAT -8.42 ASCAT -0.18 RMSE RMSE RMSE (a) (b) (c) Figure 4: Continued. 0.6 3.0 3.0 1.0 1.0 1.2 2.0 2.0 6 1.0 1.0 1.0 1.0 1.0 1.0 0.99 0.99 0.99 0.99 0.99 0.99 0.95 0.95 0.95 0.95 0.95 0.95 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.8 0.8 Correlation Correlation Correlation Correlation Correlation Correlation 0.7 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.2 10 Advances in Meteorology 0.0 0.0 0.0 0 15 30 45 60 75 90 105 120 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 0 15 30 45 60 75 90 105 120 Standard deviation Standard deviation Standard deviation PNBOIA Bias PNBOIA Bias PNBOIA Bias CCMP -6.11 CCMP -0.18 CCMP -13.52 ERA5 -4.16 ERA5 -0.65 ERA5 -13.67 BRAMS -0.09 BRAMS -0.48 BRAMS -5.51 ASCAT -3.84 ASCAT -0.20 ASCAT -14.34 RMSE RMSE RMSE (d) (e) (f) Figure 4: Taylor diagrams for statistical comparisons between CCMP, BRAMS, ERA5, and ASCAT-A (a, b), or ASCAT-B (c, d), or ASCAT- CDR (e, f) that have coincident records with shelf water buoy data. (a) Speed-ASCAT-A. (b) Direction-ASCAT-A. (c) Speed-ASCAT-B. (d) Direction-ASCAT-B. (e) Speed-ASCAT-CDR. (f) Direction-ASCAT-CDR. 0.0 0.0 0.0 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 0 1020304050607080 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 Standard deviation Standard deviation Standard deviation PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias CCMP -0.01 CCMP -1.26 CCMP -0.04 ERA5 -0.28 ERA5 -1.89 ERA5 -0.30 BRAMS -0.13 BRAMS 4.88 BRAMS -0.19 ASCAT 0.04 ASCAT -2.56 ASCAT 0.06 RMSE RMSE RMSE (a) (b) (c) Figure 5: Continued. 3.0 2.4 2.4 0.8 0.8 1.0 1.6 1.6 2.0 42 1.0 1.0 1.0 0.99 0.99 0.99 0.95 0.95 0.95 0.9 0.9 0.9 0.8 0.8 0.8 Correlation Correlation Correlation 0.7 0.7 0.7 0.6 0.6 0.6 0.4 0.4 0.4 0.2 0.2 0.2 Advances in Meteorology 11 0.0 0.0 0.0 10 20 30 40 50 60 70 80 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 0 8 16 24 32 40 48 56 64 72 Standard deviation Standard deviation Standard deviation PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias PIRATA+PNBOIA Bias CCMP -1.99 CCMP 0.03 CCMP -2.39 ERA5 -1.64 ERA5 -0.23 ERA5 -2.80 BRAMS 4.12 BRAMS -0.11 BRAMS 4.29 ASCAT -1.06 ASCAT 0.07 ASCAT -3.30 RMSE RMSE RMSE (d) (e) (f) Figure 5: Taylor diagrams for statistical comparisons between CCMP, BRAMS, ERA5, and ASCAT-A (a, b), or ASCAT-B (c, d), or ASCAT- CDR (e, f) that have coincident records with all buoys data. (a) Speed-ASCAT-A. (b) Direction-ASCAT-A. (c) Speed-ASCAT-B. (d) Direction-ASCAT-B. (e) Speed-ASCAT-CDR. (f) Direction-ASCAT-CDR. −1 in the 6–9m·s interval, while higher differences were in BRAMS wind speed data, the highest RMSE con- spd −1 −1 −1 observed for winds >12m·s (1.11m·s <RMSE tinued to be obtained for this dataset (RMSE >1.56 m·s spd spd −1 −1 −1 >4.59m·s , −3.55m·s <bias > −0.59m·s ). For all in simultaneous comparisons). spd speed intervals and all datasets considered in the analysis, the Indeed, if analysis is restricted to OW buoys, best per- lower RMSE values and higher R values were obtained formance is observed for CCMP product (Figure 6 and spd spd for ASCAT-A and ASCAT-B products. Regarding wind Table S2), showing the lowest errors in terms of RMSE and dir −1 direction, higher differences were observed in the 3–6m·s the best values of R and R for 5 or more buoys among spd dir interval for OW buoys, considering all wind products. For the 8 OW buoys compared. Although the data from the SW buoys, higher differences were also observed for the PIRATA buoys are assimilated in the generation of the lower wind speed interval, considering ASCAT-A, ASCAT- CCMP dataset, these results agree with previous studies B, and ERA5 products. Best comparisons were obtained for that compared CCMP and PIRATA wind data [41–43]. −1 CCMP product with oceanic water buoys (RMSE <23.3 , Suzuki et al. [42] obtained an RMSE value of 1.04m·s , dir spd ° ° −0.01 <bias <0.61 , R >0.91), and for ASCAT-B with and Yan et al. [43] obtained RMSE , bias , and R dir dir spd spd spd −1 −1 ° ° ° shelf water buoys (RMSE <56.6 , −0.47 <bias <7.66 , values of 1.29m·s , 0.24 m·s , and 0.90, respectively. For dir dir R >0.59), in all speed intervals. wind direction, these authors [43] obtained dir ° ° RMSE �22.4 , bias �0.89 , and R �0.97. In the dir dir dir present study, ASCAT products showed the best results 4. Discussion after CCMP, with a performance close to other recent In the present study, data from six different ocean surface comparisons [44]. +e lowest errors in terms of RMSE spd wind products were compared with in situ measurements for 5 of the 8 OW buoys were obtained for ASCAT-A. from 15 moored buoys in oceanic and shelf waters of the Verhoef et al. [61] validated the ASCAT-A product opti- Brazilian margin. In general, there is a good correspon- mized for coastal regions using in situ buoy data from the dence between the wind speeds from the simultaneous tropical oceans and along North American and European records of wind products and buoy observations coasts. For buoys located more than 50km from the coast, −1 (Figures 3–5 and Table S1). Without considering the these authors obtained RMSE of 2.2 m·s and bias of spd spd −1 −1 BRAMS model, comparisons of simultaneous records are −0.29m·s . In the present study, RMSE <1.25m·s and spd −1 −1 even better. It can be noted that ASCAT and CCMP bias between −0.33m·s and 0.25 m·s were computed spd products had slightly better results than ERA5. +e spatial for ASCAT-A in oceanic waters. In the ASCAT-B vali- −1 resolution of the databases seems to influence their per- dation report [62], bias values of 0.02 m s (CMOD5.n) −1 formance, with the finest resolution products (satellite- and 0.04 m s (CMOD7) were reported, whereas in the −1 derived databases) showing better results than the coarser present study, bias values varied between −0.24m·s and −1 ERA5 database. Despite the removal of the systematic bias 0.26 m·s . In the ASCAT-CDR validation report [23], bias 2.4 0.8 1.6 4 12 Advances in Meteorology 2.0 1.5 1.0 0.5 0.0 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 0.4 0.0 –0.4 –0.8 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 1.00 0.75 0.50 0.25 0.00 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W –10 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W 1.00 0.75 0.50 0.25 0.00 8n38w 4N38W 4N23W 0N35W 0N23W 8S30W 14S32W 19S34W OW buoys surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 Figure 6: Barplots representing the statistical metrics computed for each oceanic water (OW) buoy. RMSE and RMSE : mean square spd dir error for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s correlation coefficient for spd dir spd dir wind speed and direction. –1 –1 R bias (°) RMSE (°) R bias (m s ) RMSE (m s ) dir dir dir spd spd spd Advances in Meteorology 13 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Rio Grande Itajaí 0.0 –0.4 –0.8 –1.2 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande 1.00 0.75 0.50 0.25 0.00 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Rio Grande Itajaí Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande –20 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande 1.00 0.75 0.50 0.25 0.00 Fortaleza Recife Porto Seguro Cabo Frio 2 Santos Itajaí Rio Grande SW buoys surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 Figure 7: Barplots representing the statistical metrics computed for each shelf waters (SW) buoy. RMSE and RMSE : mean square error spd dir for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s correlation coefficient for wind spd dir spd dir speed and direction. –1 –1 R bias (°) RMSE (°) R bias (m s ) RMSE (m s ) dir dir dir spd spd spd 14 Advances in Meteorology 0 0 0.5 7.5 0.0 5.0 –0.5 2.5 –1.0 –1.5 0.0 –2.0 –2.5 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 3-6 6-9 9-12 >12 3-6 6-9 9-12 >12 wind speed intervals surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 −1 Figure 8: Barplots representing the statistical metrics computed for all buoys in four wind speed bins (m·s ). RMSE and RMSE : mean spd dir square error for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s correlation spd dir spd dir coefficient for wind speed and direction. −1 −1 values ranged between −0.3 m·s and 0.2m·s , whereas in shows the differences between the wind roses of Santos buoy −1 the present study, bias values varied between −0.37m·s and ASCAT-A paired series, which are also observed for the −1 other datasets considered in this study (Figures S2–S16 in and 0.15m·s . In terms of wind direction, relatively higher differences Supplemental File). It is worth remembering that PNBOIA data (SW) are hourly averages, while PIRATA data (OW) are were observed in the present study, even without consid- ering BRAMS comparisons. Most oceanic water buoys are 10 minutes averages, which can also contribute to the rel- located in the Tropical Atlantic region where winds are more atively greater differences observed for southern shelf water stable [50, 51]. However, highest statistical differences were buoy comparisons [75]. For satellite-derived wind products, observed for shelf water buoys (Figure 2 and Table S1). Wind land masking limitations can be an issue [45]. In addition, direction comparisons were considered more satisfactory more coastal winds are very dependent on the surrounding only for SW buoys located in the northern part of the topography, land-ocean gradients, and discontinuities. Brazilian continental margin (Figure 7 and Table S3). For Previous studies reported overestimations of wind speed Santos, Itaja´ı, and Rio Grande buoys located in the southern by coastal ASCAT-A and ASCAT-B products in the Iberian, Irish, and Japanese continental shelf regions [20, 45, 46]. part of the study region, wind direction comparisons resulted in relatively higher differences. According to Lin Takeyama et al. [46] proposed that this overestimation is et al. [74], when comparing the wind vectors classified as inversely proportional to the distance from the coast due to variable or stable, the estimated error variances of the increased backscatter by land contamination. However, variable winds are much higher than those of the stable Verhoef et al. [61] observed that the ASCAT-A product winds. +e southern Brazilian continental margin is well optimized for the coastal region underestimated buoy wind known in the literature for frequent incursions of frontal speed measurements in the northern hemisphere. In the systems and relevant cyclogenetic activity, which increases present study, not only ASCAT products but all wind the variability of the wind field in this region [52–56]. +is products underestimated buoy wind speed for shelf waters. variability pattern for these 3 buoys can be depicted from the Payan [76] observed that the ASCAT-A product over- wind roses shown in Figure 2. As an example, Figure 10 estimated buoy wind speeds in the northern hemisphere and –1 –1 R bias (m s ) RMSE (m s ) spd spd spd R bias (°) RMSE (°) dir dir dir Advances in Meteorology 15 oceanic waters shelf waters –1 –2 –3 –4 0.75 0.50 0.25 0.00 –0.25 –5 –10 –15 1.00 0.75 0.50 0.25 0.00 3-6 6-9 9-12 >12 3-6 6-9 9-12 >12 wind speed intervals surface ocean wind products ASCAT-A ASCAT-CDR CCMP ASCAT-B BRAMS ERA5 −1 Figure 9: Barplots representing the statistical metrics computed for oceanic and shelf water buoys in four wind speed bins (m·s ). RMSE spd and RMSE : mean square error for wind speed and direction; bias and bias : bias for wind speed and direction; R and R : Pierson’s dir spd dir spd dir correlation coefficient for wind speed and direction. –1 –1 R bias (°) RMSE (°) R bias (m s ) RMSE (m s ) dir dir dir spd spd spd 16 Advances in Meteorology N N NW NE NW NE 15 15 10 10 5 5 W E W E SW SE SW SE S S –1 –1 Wind Speed (m.s ) Wind Speed (m.s ) 0 - 3 6 - 9 12 - 15 0 - 3 6 - 9 12 - 15 3 - 6 9 - 12 > 15 3 - 6 9 - 12 > 15 (a) (b) Figure 10: Wind roses of the Santos buoy and ASCAT-A paired series (Please see Figures S2–S16 for all buoys and wind datasets). underestimated speeds measured in the southern hemi- 5. Conclusions sphere. In general, the comparisons of wind speed obtained +is study evaluates and compares satellite-derived ocean in the present study for shelf waters agree with previous surface wind data (ASCAT-A and ASCAT-B scatter- studies. +e best performances are distributed between ASCATproducts (Figure 7, Table S3). CCMP and ERA5 also ometers), reanalysis (ERA5), BRAMS-modelled surface winds, and a satellite-reanalysis-buoy-blended product show good performance for wind speed, with exception for Recife buoy, which is the closest to the coast (35km). (Cross-Calibrated Multi-Platform ocean wind vectors), to analyze the performance of these datasets in representing the Regarding the comparisons by wind speed intervals (Figures 8 and 9 and Table S4), ASCAT-A and ASCAT-B in situ measurements obtained by moored buoys along the Brazilian margin. +e results presented here can be of great showed a better statistical performance in terms of RMSE value for climate and oceanographic studies, among others, and R, considering all datasets. In general, the lowest especially where in situ measured wind data are not available RMSE and higher R values were observed in the spd spd −1 or are insufficient, therefore requiring alternative sources of 6–9m·s interval. +e poor wind speed accuracy for interval −1 wind data to be considered. >12m·s for OW buoys can be related to the relatively low Ocean surface wind estimations obtained with ASCAT amount of data recorded for this wind speed interval, as seen scatterometer optimized for coastal regions and with a in the wind roses shown in Figure 2. Almost all wind −1 −1 spatial resolution of 12.5km are the ones that best represent products overestimated speeds between 3m·s and 6m·s −1 and underestimated speeds >6m·s , so that as wind speed the in situ wind speed, with the lower RMSE spd −1 −1 (<1.59m·s ), lower bias (±0.68m·s ), and higher cor- increases, the negative bias increases proportionally. spd According to Carvalho et al. [19], weaker winds produce low relation (>0.75) in relation to the other products considered in the study. However, CCMP product also showed a good amounts of backscatter, which scatterometers have diffi- culties in measuring. Stronger winds can only produce statistical comparison with in situ buoy data −1 −1 (RMSE <1.87m·s , bias ±0.96m·s , R >0.82), as well as proportional backscatter in the ocean surface until a certain −1 −1 ERA5 (RMSE <1.84m·s , bias ±1.21m·s , threshold, after which no more backscatter is produced even spd spd R >0.76). Among all the compared ocean surface wind if the wind speed keeps increasing. Regarding wind direc- spd products, BRAMS is the least able to represent the wind field tion, best comparisons with OW buoys were obtained for −1 (RMSE >1.04m·s , bias ±2.89, R <0.87). In regions CCMP product. High values of RMSE and low correla- dir spd spd spd with more variable winds, as in the southern Brazilian tions for SW buoys can be related to subcell wind variability in southern Brazilian shelf, as mentioned above [74]. In continental margin, the estimation of wind direction is more susceptible to errors (RMSE >42.4 ). general, the statistical performance for wind direction in- dir creases towards the higher speed intervals. According to Considering the relatively higher temporal resolution of ERA5 and CCMP products (1 and 6 hours, respectively), and Chakraborty et al. [37], the scatterometer errors associated with wind direction are highly random and are dependent the extension of the available time series (>30 years), it is suggested to use these datasets to build a regional clima- on wind speed. In general, the lower the wind speeds, the tology and to analyze the space-time variability of wind fields higher the wind direction errors. Santos (%) ASCAT-A (%) Advances in Meteorology 17 [3] C. Aguirre, S. Garc´ıa-Loyola, G. Testa, D. Silva, L. Farias, and in the study region. Nevertheless, there are seven orbital J. E. Tremblay, “Insight into anthropogenic forcing on coastal scatterometers operating currently [77–81]. +is number of upwelling off south-central Chile,” Elementa Science of the operational scatterometers has been increasing in recent Anthropocene, vol. 6, no. 59, 2018. years and in near future. Once the scatterometers can give [4] H. Demarcq, “Trends in primary production, sea surface winds within 6 hours over the regional ocean, it may be good temperature and wind in upwelling systems (1998–2007),” to use combined winds from multiple scatterometers. +e Progress in Oceanography, vol. 83, no. 1–4, pp. 376–385, 2009. derived information for such satellite constellation and al- [5] M. G. Jacox, E. L. Hazen, and S. J. Bograd, “Optimal envi- ternative sources could be useful for several applications, as ronmental conditions and anomalous ecosystem responses: on the estimation of the potential wind power generation in constraining bottom-up controls of phytoplankton biomass in the Brazilian margin region. the California current system,” Scientific Reports, vol. 6, Ar- ticle ID 27612, 2016. [6] Y. Amitai, Y. Ashkenazy, and H. Gildor, “+e effect of wind- Data Availability stress over the eastern Mediterranean on deep-water for- mation in the Adriatic sea,” Deep Sea Research Part II: Topical In situ buoy wind data can be accessed at https:// Studies in Oceanography, vol. 164, pp. 5–13, 2019. www.marinha.mil.br/chm/dados-do-goos-brasil/pnboia- [7] J. L. Bullister, M. Rhein, and C. Mauritzen, “Deepwater mapa and https://www.pmel.noaa.gov/tao/drupal/disdel/; formation,” in Ocean Circulation and Climate—A 21st Cen- scatterometer wind data at projects.knmi.nl/scatterometer/; tury Perspective, vol. 103, pp. 227–253, Academic Press, blended CCMP data at http://www.remss.com/measure- Cambridge, MA, USA, 2013. ments/ccmp/; and ERA5 reanalysis data at www.ecmwf.int/ [8] W. G. Large and S. Pond, “Open ocean momentum flux en/forecasts/datasets/archive-datasets/reanalysis-datasets/ measurements in moderate to strong winds,” Journal of era5. +e authors agree to make data and materials sup- Physical Oceanography, vol. 11, no. 3, pp. 324–336, 1981. porting the results presented in the manuscript available [9] D. B. Chelton, M. G. Schlax, M. H. Freilich, and R. F. Milliff, upon reasonable request. “Satellite measurements reveal persistent small-scale features in ocean winds,” Science, vol. 303, no. 5660, pp. 978–983, Conflicts of Interest [10] C. Fairall, B. Barnier, and D. I. Berry, “Observations to quantify air-sea fluxes and their role in climate variability and +e authors declare that there are no conflicts of interest predictability,” in Proceedings of 2009 OceanObs: Sustained regarding the publication of this paper. Ocean Observations and Information for Society, vol. 2, Venice, Italy, September 2009. Acknowledgments [11] J. P. Machado, F. Justino, and L. P. Pezzi, “Impacts of wind stress changes on the global heat transport, baroclinic in- +e authors would like to thank the Brazilian Navy Hy- stability, and the thermohaline circulation,” Advances in drography Centre, INPE’s Weather and Climate Forecast Meteorology, vol. 2016, Article ID 2089418, 15 pages, 2016. Centre, the Pacific Marine Environmental Laboratory, the [12] W. T. Liu, W. Tang, and X. Xie, “Wind power distribution European Organization for the Exploitation of Meteoro- over the ocean,” Geophysical Research Letters, vol. 35, no. 13, logical Satellites, the Royal Netherlands Meteorological Institute, and the European Centre for Medium Range [13] J. Vogelzang and A. Stoffelen, “Scatterometer wind vector Weather Forecasts for the data used in this study. +is work products for application in meteorology and oceanography,” Journal of Sea Research, vol. 74, pp. 16–25, 2012. was supported by the Coordenação de Aperfeiçoamento de [14] C. M. Risien and D. B. Chelton, “A satellite-derived clima- Pessoal de N´ıvel Superior (CAPES), Brazil (Grant Finance tology of global ocean winds,” Remote Sensing of Environ- Code 001). VGP was supported by CAPES (PROEX 0487/ ment, vol. 105, no. 3, pp. 221–236, 2006. 1681024). [15] D. Dee, J. Fasullo, D. Shea, J. Walsh, and NCAR, >e Climate Data Guide: Atmospheric Reanalysis: Overview & Comparison Supplementary Materials Tables, National Center for Atmospheric Research, Boulder, CO, USA, 2016, https://climatedataguide.ucar.edu/climate- +e supporting information consists of detailed tables and data/atmospheric-reanalysis-overview-comparison-tables. figures of our results that are mentioned and discussed in the [16] A. J. Wallcraft, A. B. Kara, C. N. Barron, E. J. Metzger, text of the Results and Discussion sections. (Supplementary R. L. Pauley, and M. A. Bourassa, “Comparisons of monthly Materials) mean 10m wind speeds from satellites and NWP products over the global ocean,” Journal of Geophysical Research: At- mospheres, vol. 114, no. 16, 2009. References [17] S. R. Smith, P. J. Hughes, and M. A. Bourassa, “A comparison of nine monthly air-sea flux products,” International Journal [1] M. Bourassa, A. Stoffelen, H. Bonekamp et al., “Remotely sensed winds and wind stresses for marine forecasting and of Climatology, vol. 31, no. 7, pp. 1002–1027, 2011. [18] E. Kalnay, Atmospheric Modeling, Data Assimilation and ocean modeling,” in Proceedings of 2009 OceanObs: Sustained Ocean Observations and Information for Society, vol. 2, Predictability, Cambridge University Press, New York, NY, USA, 2003. Venice, Italy, September 2009. [2] J. Kampf ¨ and P. Chapman, “Seasonal wind-driven coastal [19] D. Carvalho, A. Rocha, M. Go´mez-Gesteira, and C. Silva Santos, “Comparison of reanalyzed, analyzed, satellite-retrieved upwelling systems,” in Upwelling Systems of the World, Chapter 8, pp. 315–361, Springer, Berlin, Germany, 2016. and NWP modelled winds with buoy data along the Iberian 18 Advances in Meteorology Peninsula coast,” Remote Sensing of Environment, vol. 152, [36] F. Pimenta, W. Kempton, and R. Garvine, “Combining me- pp. 480–492, 2014. teorological stations and satellite data to evaluate the offshore [20] D. Carvalho, A. Rocha, M. Gomez-Gesteira, ´ and C. Silva wind power resource of southeastern Brazil,” Renewable Santos, “Offshore winds and wind energy production esti- Energy, vol. 33, no. 11, pp. 2375–2387, 2008. mates derived from ASCAT, OSCAT, numerical weather [37] A. Chakraborty, R. Kumar, and A. Stoffelen, “Validation of prediction models and buoys—a comparative study for the ocean surface winds from the OCEANSAT-2 scatterometer using triple collocation,” Remote Sensing Letters, vol. 4, no. 1, Iberian Peninsula Atlantic coast,” Renewable Energy, vol. 102, pp. 433–444, 2017. pp. 84–93, 2013. [38] D. S. Dukhovskoy, M. A. Bourassa, G. N. Petersen, and [21] M. A. Bourassa, A. Stoffelen, H. Bonekamp et al., “Remotely sensed winds and wind stresses for marine forecasting and J. Steffen, “Comparison of the ocean surface vector winds ocean modeling,” Frontiers in Marine Science, vol. 6443 pages, from atmospheric reanalysis and scatterometer-based wind 2019. products over the Nordic seas and the northern north Atlantic [22] F. J. Wentz, L. Ricciardulli, E. Rodriguez et al., “Evaluating and their application for ocean modeling,” Journal of Geo- and extending the ocean wind climate data record,” IEEE physical Research: Oceans, vol.122, no. 3, pp.1943–1973, 2017. Journal of Selected Topics in Applied Earth Observations and [39] J. Yang and J. Zhang, “Comparison of Oceansat-2 scatter- ometer wind data with global moored buoys and ASCAT Remote Sensing, vol. 10, no. 5, pp. 2165–2185, 2017. [23] A. Verhoef, J. Vogelzang, and A. Stoffelen, ASCAT L2 Winds observation,” Advances in Meteorology, vol. 2019, Article ID Data Record Validation Report, Royal Netherlands Meteo- 1651267, 9 pages, 2019. rological Institute, De Bilt, Netherlands, 2016. [40] Z. Wang, A. Stoffelen, J. Zou et al., “Validation of new sea [24] N. Ebuchi, H. C. Graber, and M. J. Caruso, “Evaluation of surface wind products from scatterometers onboard the HY- wind vectors observed by QuikSCAT/SeaWinds using ocean 2B and MetOp-C satellites,” IEEE Transactions on Geoscience buoy data,” Journal of Atmospheric and Oceanic Technology, and Remote Sensing, vol. 58, no. 6, pp. 4387–4394, 2020. vol. 19, no. 12, pp. 2049–2062, 2002. [41] A. Okuro and M. Kubota, “Inter-comparison of daily mean [25] F. M. Pimenta, A. R. Silva, A. T. Assireu, V. d. S. e. Almeida, sea surface wind speeds in various global high-resolution and O. R. Saavedra, “Brazil offshore wind resources and at- products,” Journal of >e Remote Sensing Society of Japan, mospheric surface layer stability,” Energies, vol. 12, no. 21, vol. 34, no. 3, pp. 177–196, 2014. [42] N. Suzuki, M. Konda, K. Kutsuwada, and K. Utsunomiya, p. 4195, 2019. [26] A. B. Villas Boas, ˆ F. Ardhuin, A. Ayet et al., “Integrated “Comparison of the accuracy of various global wind speed observations of global surface winds, currents, and waves: datasets obtained from satellites and reanalyses,” Journal of requirements and challenges for the next decade,” Frontiers in Advanced Marine Science and Technology Society, vol. 24, Marine Science, vol. 6, no. 425, 2019. no. 2, pp. 31–37, 2018. [27] D. B. Chelton, M. H. Freilich, J. M. Sienkiewicz, and J. M. Von [43] Q. Yan, J. Zhang, J. Meng, J. Wang, and C. Fan, “Use of an Ahn, “On the use of QuikSCAT scatterometer measurements optimum interpolation method to construct a high-resolution of surface winds for marine weather prediction,” Monthly global ocean surface vector wind dataset from active scat- Weather Review, vol. 134, no. 8, pp. 2055–2071, 2006. terometers and passive radiometers,” International Journal of [28] D. Halpern, “Offshore Ekman transport and Ekman pumping Remote Sensing, vol. 38, no. 20, pp. 5569–5591, 2017. off Peru during the 1997–1998 El Nino,” Geophysical Research [44] A. Modi, S. K. Munaka, R. Harikumar, T. B. Nair, and Letters, vol. 29, no. 5, pp. 1–19, 2002. K. Srinivas, “Evaluation of winds from SCATSAT-1 and [29] D. Chelton and S.-P. Xie, “Coupled ocean-atmosphere in- ASCAT using buoys in the Indian ocean,” Journal of the teraction at oceanic mesoscales,” Oceanography, vol. 23, no. 4, Indian Society of Remote Sensing, vol. 49, pp.1915–1925, 2021. pp. 52–69, 2010. [45] T. Remmers, F. Cawkwell, C. Desmond, J. Murphy, and [30] A. Bentamy and D. C. Fillon, “Gridded surface wind fields E. Politi, “+e potential of advanced scatterometer (ASCAT) from Metop/ASCATmeasurements,” International Journal of 12.5km coastal observations for offshore wind farm site se- Remote Sensing, vol. 33, no. 6, pp. 1729–1754, 2012. lection in Irish waters,” Energies, vol. 12, no. 2, p. 206, 2019. [31] A. Bentamy, S. A. Grodsky, A. Elyouncha, B. Chapron, and [46] Y. Takeyama, T. Ohsawa, S. Shimada, K. Kozai, K. Kawaguchi, F. Desbiolles, “Homogenization of scatterometer wind re- and T. Kogaki, “Assessment of the offshore wind resource in trievals,” International Journal of Climatology, vol. 37, no. 2, Japan with the ASCAT microwave scatterometer,” Interna- pp. 870–889, 2017. tional Journal of Remote Sensing, vol. 40, no. 3, pp.1200–1216, [32] N. Ebuchi, “Intercomparison of four ocean vector wind 2019. products from OceanSat-2 scatterometer,” in Proceedings of [47] B. M. Castro, F. P. Brandini, M. Dottori, and J. F. Fortes, “A the 2013 IEEE International Geoscience and Remote Sensing Amazonia ˆ Azul: recursos e preservação,” Revista USP, Symposium—IGARSS 2013, Melbourne, Australia, 2013. vol. 113, pp. 7–26, 2017. [33] Z. Wang, A. Stoffelen, B. Zhang, Y. He, W. Lin, and X. Li, [48] M. S. Reboita, T. Ambrizzi, B. A. Silva, R. F. Pinheiro, and “Inconsistencies in scatterometer wind products based on R. P. Da Rocha, “+e south Atlantic subtropical anticyclone: ASCAT and OSCAT-2 collocations,” Remote Sensing of En- present and future climate,” Frontiers in Earth Science, vol. 7, vironment, vol. 225, pp. 207–216, 2019. p. 8, 2019. [34] M. C. Baptista, “An analysis of the surface wind field over the [49] X. Sun, K. H. Cook, and E. K. Vizy, “+e south Atlantic tropical and southern Atlantic ocean using data from the ERS subtropical high: climatology and interannual variability,” scatterometer,” Master’s thesis, INPE, San Jose dos Campos, Journal of Climate, vol. 30, no. 9, pp. 3279–3296, 2017. Brazil, 2000. [50] S. Hastenrath, Climate Dynamics of the Tropics, Vol. 8, [35] A. Bentamy, D. Croize-Fillon, and C. Perigaud, “Character- Springer Science & Business Media, Berlin, Germany, 1991. ization of ASCAT measurements based on buoy and [51] A. Melo, I. A. Cavalcanti, and P. Souza, “Zona de con- QuikSCAT wind vector observations,” Ocean Science, vol. 4, vergˆencia intertropical do Atlantico,” ˆ in Tempo e Clima no no. 4, pp. 265–274, 2008. BrasilOficina de Textos, Sao Paulo, Brazil, 2009. Advances in Meteorology 19 [52] J. M. Wallace and P. V. Hobbs, Atmospheric Science: An [68] S. R. Jammalamadaka and A. Sengupta, Topics in Circular Statistics, Vol. 5, World Scientific, Singapore, 2001. Introductory Survey, Vol. 92, Elsevier, Amsterdam, Nether- lands, 2006. [69] K. E. Taylor, “Summarizing multiple aspects of model per- formance in a single diagram,” Journal of Geophysical Re- [53] I. A. Cavalcanti and V. E. Kousky, “Frentes frias sobre o Brasil,” in Tempo e Clima no Brasil, pp. 135–147, Oficina de search: Atmospheres, vol. 106, no. 7, pp. 7183–7192, 2001. [70] M. P. Byrne, A. G. Pendergrass, A. D. Rapp, and Textos, Sao Paulo, Brazil, 2009. [54] B. J. Hoskins and K. I. Hodges, “A new perspective on K. R. Wodzicki, “Response of the intertropical convergence zone to climate change: location, width, and strength,” southern hemisphere storm tracks,” Journal of Climate, Current Climate Change Reports, vol. 4, no. 4, pp. 355–370, vol. 18, no. 20, pp. 4108–4129, 2005. [55] M. S. Reboita, M. A. Gan, R. P. d. Rocha, and T. Ambrizzi, [71] M. B. Switanek, P. A. Troch, C. L. Castro et al., “Scaled “Regimes of precipitation in south America: a bibliographical distribution mapping: a bias correction method that preserves review,” Brazilian Journal of Meteorology, vol. 25, no. 2, raw climate model projected changes,” Hydrology and Earth pp. 185–204, 2010. System Sciences, vol. 21, no. 6, pp. 2649–2666, 2017. [56] L. M. V. Carvalho, C. Jones, and B. Liebmann, “+e south [72] C. Anderson, H. Bonekamp, C. Duff, J. Figa-Saldana, and Atlantic convergence zone: intensity, form, persistence, and J. J. W. Wilson, “Analysis of ASCAT ocean backscatter relationships with intraseasonal to interannual activity and measurement noise,” IEEE Transactions on Geoscience and extreme rainfall,” Journal of Climate, vol.17, no.1, pp. 88–108, Remote Sensing, vol. 50, no. 7, pp. 2449–2457, 2012. [73] L. Zeng and R. A. Brown, “Scatterometer observations at high [57] W. T. Liu, K. B. Katsaros, and J. A. Businger, “Bulk pa- wind speeds,” Journal of Applied Meteorology, vol. 37, no. 11, rameterization of air-sea exchanges of heat and water vapor pp. 1412–1420, 1998. including the molecular constraints at the interface,” Journal [74] W. Lin, M. Portabella, A. Stoffelen, J. Vogelzang, and of the Atmospheric Sciences, vol. 36, no. 9, pp. 1722–1735, A. Verhoef, “ASCAT wind quality under high subcell wind variability conditions,” Journal of Geophysical Research: [58] S. D. Smith, “Coefficients for sea surface wind stress, heat flux, Oceans, vol. 120, no. 8, pp. 5804–5819, 2015. and wind profiles as a function of wind speed and temper- [75] J. C. May and M. A. Bourassa, “Quantifying variance due to ature,” Journal of Geophysical Research, vol. 93, no. C12, temporal and spatial difference between ship and satellite pp. 15467–15472, 1988. winds,” Journal of Geophysical Research: Oceans, vol. 116, [59] H. Hersbach, A. Stoffelen, and S. De Haan, “+e improved C-band geophysical model function CMOD5,” in Proceedings [76] C. Payan, “Improvements in the use of scatterometer winds in of the 2005 Envisat & ERS Symposium, Salzburg, Austria, the operational NWP system at Met´ ´ eo-France,” in Proceedings of the 2010 International Winds Workshop, vol. 10, pp. 22–26, [60] A. Stoffelen, J. A. Verspeek, J. Vogelzang, and A. Verhoef, Tokyo, Japan, 2010. “+e CMOD7 geophysical model function for ASCAT and [77] A. Verhoef and A. Stoffelen, ASCAT Wind Validation Report, ERS wind retrievals,” IEEE Journal of Selected Topics in Ap- Royal Netherlands Meteorological Institute, De Bilt, Neth- plied Earth Observations and Remote Sensing, vol. 10, no. 5, erlands, 2018. pp. 2123–2134, 2017. [78] J. Verspeek, A. Verhoef, and A. Stoffelen, ASCAT-C Wind [61] A. Verhoef, M. Portabella, and A. Stoffelen, “High-resolution Product Calibration and Validation, Royal Netherlands Me- ASCAT scatterometer winds near the coast,” IEEE Transac- teorological Institute, De Bilt, Netherlands, 2019. tions on Geoscience and Remote Sensing, vol. 50, no. 7, [79] S. Singh, R. K. Tiwari, H. S. Gusain, and V. Sood, “Potential pp. 2481–2487, 2012. applications of SCATSAT-1 satellite sensor: a systematic [62] A. Verhoef and A. Stoffelen, ASCAT Coastal Winds Valida- review,” IEEE Sensors Journal, vol. 20, no. 21, tion Report, Royal Netherlands Meteorological Institute, De pp. 12459–12471, 2020. Bilt, Netherlands, 2013. [80] H. Wang, J. Zhu, M. Lin, Y. Zhang, and Y. Chang, “Evaluating [63] R. Atlas, R. N. Hoffman, J. Ardizzone et al., “A cross-calibrated, Chinese HY-2B HSCATocean wind products using buoys and multiplatform ocean surface wind velocity product for mete- other scatterometers,” IEEE Geoscience and Remote Sensing orological and oceanographic applications,” Bulletin of the Letters, vol. 17, no. 6, pp. 923–927, 2020. American Meteorological Society, vol. 92, no. 2, pp. 157–174, [81] R. E. N. Yueming, LM-4B Successfully Launched HY-2C Satellite, China Aerospace Science and Technology Corpo- [64] F. J. Wentz, J. Scott, R. Hoffman, M. Leidner, R. Atlas, and ration, Beijing, China, 2020. J. Ardizzone, “Remote sensing systems cross-calibrated multi- platform (CCMP) 6-hourly ocean vector wind analysis product on 0.25deg grid,” 2015, https://www.remss.com/ measurements/ccmp/. [65] R. Atlas, R. N. Hoffman, S. C. Bloom, J. C. Jusem, and J. Ardizzone, “A multiyear global surface wind velocity dataset using SSM/I wind observations,” Bulletin of the American Meteorological Society, vol. 77, no. 5, pp. 869–882, 1996. [66] R. N. Hoffman, “SASS wind ambiguity removal by direct minimization,” Monthly Weather Review, vol. 110, no. 5, pp. 434–445, 1982. [67] R. N. Hoffman, “SASS wind ambiguity removal by direct minimization. Part II: use of smoothness and dynamical constraints,” Monthly Weather Review, vol. 112, no. 9, pp. 1829–1852, 1984.

Journal

Advances in MeteorologyHindawi Publishing Corporation

Published: Sep 14, 2021

References