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

Learn More →

Effect of biometeorological variables on the onset of phenophases derived from MODIS data and visual observations

Effect of biometeorological variables on the onset of phenophases derived from MODIS data and... In this study we analyzed the effect of selected biometeorological variables on the onset of phenophases in three beech stands in different climatic areas (warm, moderately warm and cold). We have focused on two phenophases ­ leaf unfolding and leaf colouring. Timing of both phenophases was identified visually and using series of MODIS satellite images. The data were collected during a 13-year period (2000­2012). For the spring period, we found a significant dependence between temperature and precipitation-based biometeorological variables and leaf unfolding in both datasets ­ those based on visual and remote sensing-based observations. The average air temperature in the period from February­April was the most significant factor which initiated the onset of beginning of leaf unfolding in all three investigated stands. The evapotranspiration-based biometeorological variables (climatic water balance, actual evapotranspiration, dryness index) had no effect on the onset of the beginning of leaf unfolding observed using both methods. The high precipitation totals in April caused the later onset of leaf unfolding in all stands. The relationship between the first autumn phenophase ­ leaf colouring and biometeorological variables was found significant in beech stand in the warm climatic area only. Keywords: Fagus sylvatica; phenology; biometeorological variables; MODIS Abstrakt V práci sme analyzovali vplyv biometeorologických premenných na nástup fenologických fáz v troch bukových porastoch nachádzajúcich sa v rozdielnych klimatických oblastiach (teplej, mierne teplej a studenej). Sledovanými fenofázami boli zaciatok zalistenia a zaciatok zltnutia listov. Ich nástup bol sledovaný vizuálne a s pouzitím satelitných údajov MODIS. Pozorovania prebiehali pocas 13-rocného obdobia (2000­2012). V jarnom období sme zistili významný vplyv biometeorologických premenných zalozených na teplote a zrázkach na zaciatok zalistenia bukových porastov zisteného obomi ­ vizuálnym aj satelitným sledovaním. Najvýznamnejsou z týchto premenných bola priemerná teplota vzduchu od februára do apríla, ktorá vyvolávala skorsí nástup zaciatku zalistenia na vsetkých troch bukových stanovistiach. Vyssie úhrny zrázok v mesiaci apríl spôsobovali na sledovaných stanovistiach neskorsí nástup tejto fenofázy. Naopak, premenné zalozené na evapotranspirácií (klimatická vodná bilancia, aktuálna evapotranspirácia, index sucha) nemali ziaden vplyv na zaciatok zalistenia zistený oboma metódami. Vzah medzi prvou jesennou fenofázou ­ zaciatkom zltnutia listov a biometeorologickými premennými bol významný iba na bukovom stanovisti nachádzajúcom sa v teplej klimatickej oblasti. Kúcové slová: Fagus sylvatica; fenológia; biometeorologické premenné; MODIS 1. Introduction The reactions of vegetation to climate change conditions have been widely discussed in recent studies on phenology, climatology and remote sensing. The phenological phases responding to the fluctuation of meteorological elements are sensitive and easy-to-observe reactions of vegetation to climate change (Badeck et al. 2004). In the middle latitudes, the onset of the phenophases like bud bursting, leaf unfolding, flowering, etc. is effected by the air temperature (Ahas et al. 2000; Sparks et al. 2000; Defila & Clot 2000; Menzel 2003). It was proved that higher temperatures during the spring period evoke earlier onset of tree species spring phenophases (Chmielewski & Rötzer 2001; Menzel et al. 2006; Kramer 1996; Pálesová 2012). But there are other meteoro- logical elements that could be limiting environmental factors for tree species in the climate change conditions: the lack of rainfall amount and uprising evapotranspiration during the growing season (Skvarenina et al. 2004). In Slovakia, there is the assumption that changed bioclimatic conditions especially from the 1st to the 3rd (partially the 4th) vegetation belt disturb existing plant (primarily beech) communities and their species compositions. However, favourable temperature conditions with sufficient water balance in the higher vegetation tiers (from the 5th) may represent climatic conditions appropriating the presence of deciduous tree species like beech (Skvarenina et al. 2004). In the past, the onset of phenophases could be observed only visually. During the last decades, the methods of phenological observations using remote sensing data (Kang et *Corresponding author. Veronika Lukasová, e-mail: veronika.brandysova@gmail.com T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 al. 2003; Zhang et al. 2003; Fisher & Mustard 2007), and simple or hemispherical digital photography (Nagai et al. 2010; Mozný et al. 2012; Felts et al. 2011) have been developed. The remote sensing data have significant potential in monitoring of vegetation dynamics on a regional to global scale. The first studies from the field of remote sensing phenology utilized data from the radiometer AVHRR. The dates of the beginning and end of the growing season were derived and the length of the growing season was calculated. Later, after the Terra and Aqua spacecrafts with the spectroradiometer MODIS were launched into orbit, the phenology was monitored more precisely (Zhang et al. 2003; Fisher et al. 2006; Soudani et al. 2012). Vegetation indices were proposed to be the biophysical indicators of the changing amount and quality of green vegetation, e.g. NDVI ­ Normalized Difference Vegetation Index (Beck et al. 2006; Narasihman & Stow 2010; Soudani et al. 2012), EVI ­ Enhanced Vegetation Index (Zhang et al. 2003; Ganguly et al. 2010), PVI ­ Perpendicular Vegetation Index (Guyon et al. 2011), WDRVI ­ Wide Dynamic Range Vegetation Index (Jönsson et al. 2010). The vegetation index commonly used in remote sensing phenology is the ratio index NDVI, which is calculated from the reflectance of vegetation in the red and near infrared band. Healthy (green) vegetation absorbs the incoming visible radiation and reflects the near infrared radiation (Weier & Herring 2011). This study was focused on the forests stands with dominant occurrence of European beech ­ Fagus sylvatica L. (subspecies F. sylvatica ssp. sylvatica), which is continually distributed across Europe. Beech is a competing proficient tree species, which does not grow only in very shallow soils with low water capacity, on wet or flooded soils with permanently or periodically low air capacity, and in the harsh mountain climate. The lower border of the European beech extension belongs to the warm climate area, dry to moderately moist precinct. The general occurrence of beech forests is localized in moderately warm to cold climate area, moderately to very moist precinct within the 2nd to 6th altitudinal vegetation belt. The upper border of sporadic occurrence of beech is localized on the transition between the moderately cold and cold mountain precincts in the cold climate area (Gömöry et al. 2011). In this study, we examined what are the limiting meteorological variables that had the most significant effect on the onset of vegetative phenophases of European beech. and cold (Table 1). The three beech stands were selected from the phenology network of the Slovak Hydrometeorological Institute. The homogeneity of beech stands in surrounding areas was verified on the basis of the Forest Management Plan database and Landsat-based tree species composition map of the Slovak forests as well (Bucha 1999). The species composition map derived from Landsat was aggregated from 30 m to 250 m pixels corresponding to MODIS images. The 250 m pixels with more than 60% of beech were classified as a "beech" class and used for further satellite based phenology analyses. The pixels on the boundary between beech forests and other land cover classes or forest types were excluded due to possible spectral contamination (Wolfe et al. 2002). 2.2. Visual phenological observations In each stand, phenological observations were performed on a sample of ten representative overstory beeches. The method by Braslavská & Kamenský (1996) was used during the 13-year period (2000­2012). In spring, one phenological stage of leaf unfolding was observed: first leaf (VIS_LU_10), when 10% of leaves of monitored trees were unfolded. In autumn, one stage of the phenophase leaf colouring was observed: beginning leaf colouring (VIS_LC_10), when 10% of leaves of monitored trees changed their colour to yellow or brown. 2.3. Phenology metrics from MODIS data The MOD09GQ (MYD09GQ) daily surface reflectance products from the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard Terra and Aqua satellites were used to calculate the seasonal course of NDVI. The MOD09GQ (MYD09GQ) consists of the red spectral band (RED 620­670 nm) and near infrared spectral band (NIR 841­876 nm) with 250 m spatial resolution. We checked the recorded reflectance at pixel level in each individual image for possible clouds; fog, aerosols, etc. according to the flags in the quality control data (MOD09GA, MYD09GA) (Vermote et al. 2011). The image pixels that did not match the quality criteria in the quality analysis, were eliminated from further analyses. Based on the MOD09GQ and MYD09GQ products we calculated daily NDVI for each of the stands in the period 2000­2012. NDVI = NIR ­ RED NIR + RED [1] 2. Material and methods 2.1. Investigated forest stands The beech stands analyzed in this study were established in three contrasting climatic areas ­ warm, moderately warm Table 1. Description of investigated beech stands. Stand Zelezná Studienka Zvolen Poana Altitude [m a.s.l.] 304 566 1051 Aspect NW SW SE Phenological metrics. The normalized difference vegetation index (NDVI) was considered as a suitable indicator of changing amount, quality and structure of European beech assi- Tree species composition [%] beech coniferous deciduous 60 20 20 65 -- 35 80 15 5 Climatic area Warm Moderately warm Cold T [°C] 9.4 7.6 4.9 P [mm] 645 745 924 T ­ average annual air temperature during period 1961­1990; P ­ average annual precipitation totals during period 1961­1990. T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 milatory organs (Brandýsová & Bucha 2012). NDVI temporal profiles were modelled with sigmoid logistic function proposed by Fisher et al. (2006) using the software program Phenological profile (Bucha & Kore 2009): stands were selected on the basis of minimal horizontal and vertical distance (Fig. 2). Table 2. Meteorological stations assigned to the phenological stands and distance between them. Phenological stand Zelezná Studienka Zvolen Poana Meteorological station Bratislava ­ Koliba Banská Stiavnica Lom nad Rimavicou Horizontal distance [km] 3 21 12 Vertical distance [m] 17 9 33 v(t) = vmin + vamp 1 1+e m -n t 1 1 1 1+e m -n t 2 2 [2] Where v(t) represents the value of the vegetation index NDVI. We used a four-parameter method (4D). Parameters vmin and vamp were predefined (Bucha et al. 2011). They represent the minimal value (vmin = 0.4295) and amplitude of NDVI (vamp = 0.4971) of beech stands in Slovakia. Phenological metrics was taken from the study Brandýsová (2013). The satellite-derived phenological markers were identified by inflection point of that sigmoid logistic function (Fig. 1) as follows: ­ Inflection point in the spring period ­ SAT_LU_1, as a time when the assimilatory organs of overstory beeches started unfolding; ­ Inflection point in the autumn period ­ SAT_LC_2, as a time when the assimilatory organs of overstory beeches were almost recoloured. We used only the first phenological marker in spring: SAT_ LU_1. This marker corresponds with the visually observed phenophase beginning of leaf unfolding (Brandýsová 2013). As the dependence between satellite derived and visually observed onset of autumn phenophases was not revealed in our previous research (Brandýsová 2013), only the effect of selected biometeorological variables on visually observed leaf coloring phase was investigated. The differences between monthly average air temperatures and precipitation totals from the 30-year normal were calculated. The earliest and latest days of the onset of phenophases of the period 2000­2012 were considered in terms of these differences. We defined biometeorological variables based on the air temperature, precipitation and evapotranspiration to find out which of them affected most significantly the onset of visually observed and satellite derived phenological phases. i) Average air temperature (AATp) of defined period: where Taverage is average daily air temperature, n is number of days. We calculated: AATp = STaverage n [3] ­ average air temperature of the period February­April: AATII­IV, ­ average air temperature of the period March­April: AATIII­IV, ­ average air temperature of February: AATII, ­ average air temperature of April: AATIV, ­ average air temperature of the period August­September: AATVIII­IX, ­ average air temperature of September: AATIX. ii) Number of days below a prescribed temperature threshold: ­ number of chilling days (nCD; Tmin < 0): number of days with minimum daily air temperature below 0 °C, counted from the end of the growing season in the previous year to the onset day of the beginning of leaf unfolding, ­ number of frozen days (nFD; Tmax < 0): number of days with maximum daily air temperature below 0 °C, counted from the end of the growing season in the previous year to the onset day of the beginning of leaf unfolding. iii) Climatic water balance (CWB): ­ climatic water balance of the period January­April: CWBI­IV, ­ climatic water balance of the period April­September: CWBIV­IX, ­ climatic water balance of the period July­September: CWBVII­IX, CWB = P ­ PET [4] where P is precipitation total (mm) and PET is potential evapotranspiration. PET was calculated according to a modified method of Tornthwait and Mather (1955) (Novák 1995): 41 Fig. 1. Inter-annual course of normalized differentiated vegetation index (NDVI) in selected beech stand in Slovakia fitted by the sigmoid logistic function. The inflection points of this function were used for the evaluation of forest stands phenology. Two phenophases were marked ­ beginning of leaf unfolding (SAT_LU_1) and leaf coloring (SAT_LC_2). On the x axis: day of year, on the y axis: NDVI. 2.4. Biometeorological variables The data from meteorological stations (Table 2) were obtained from the Slovak Hydrometeorological Institute. The data were used to analyze the effect of local meteorological conditions (especially temperature, precipitation and evapotranspiration-based) on the onset of phenological phases in beech stands. The meteorological stations from which the data were transferred to the positions of investigated forest T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 2. Location of 3 beech test stands ­ marked with a black circle, and meteorological stations ­ marked with a square. PET = 0.535 . f 10 . Tm I [5] where PET (mm) is potential evapotranspiration of months, Tm (°C) average monthly air temperature, f is correction factor depending of the length of the month and latitude: f = k . so [6] where k is number of days coefficient, so is maximal duration of sunshine during the day (h). I is temperature index calculated as a sum of twelve monthly indices: 12 [7] Tm 1.514 ­ 1.1<DI2.3 ­ semi-humid (moderately insufficient moisture; savannah), ­ 2.3<DI3.4 ­ semi-arid (insufficient moisture; semidesert), ­ 3.4<DI10 ­ arid (very insufficient moisture; desert), ­ 10<DI ­ hyper-arid (extremely insufficient moisture; desert) (Gao & Giorgi 2008; Skvarenina et al. 2009). vi) Precipitation total of April: PIV I= Si; j j=1 ij = 3. Results 3.2. Climatic variability and onset of phenophases in investigated stands Average values of biometeorological variables derived from meteorological data recorded during the period 2000­2012 are noted in table 3. The between-stand differences in biometeorological variables and onset dates of the phenophases corresponded to climatic areas, where the test stands are located. Stand 1 ­ Zelezná Studienka The average onset day of the visually observed phenophase VIS_LU_10 in the stand Zelezná Studienka was in the period 2000­2012 DOY 108. The earliest onset was recorded on DOY 102 in the years 2007 and 2009; the latest onset day was on DOY 114 in theyears 2004 and 2006. The average onset day of satellite-derived beginning of leaf unfolding was the same as that from visual observations. The earliest onset was on DOY 101 in 2007 and the latest was on DOY 113 in 2004. The year 2007, when both methods recorded the earliest onset of LU_10, had a distinctive positive temperature deviation from normal during the winter and spring months compared to other years. On the contrary, in the year 2004, when the latest onset day of LU_10 was observed, the absolute temperature differences from normal were especially during the first three months the lowest in comparison with other years with some negative values (Fig. 3). and a is exponent calculated using a = (0.0675 . I3 ­ 7.71 . I2 + 1792 . I + 47239) . 10­5 iv) Actual evapotranspiration Ea of Turc (1961) (Novák 1995): ­ actual evapotranspiration of the period January­April: Ea,I­IV, ­ actual evapotranspiration of the period April­September: Ea,IV­IX, ­ actual evapotranspiration of the period July­September: Ea,VII­IX, Ea = P . [0.9 + (P2 . L­2)]­1/2 [8] where P is precipitation total (mm), and L = 300 + 2.5T + 0.05T, where T is average air temperature of the period. v) Dryness index (DI) of Budyko (1958): ­ dryness index of the period January­April: DII­IV, ­ dryness index of the period July­September: DIVII­IX, DI = PET / P [9] where PET [mm] is potential evapotranspiration of the period, P is precipitation total [mm]. Thresholds for different climate regimes were defined as: ­ 0<DI1.1 ­ humid (surplus moisture regime; steppe to forest vegetation), T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Table 3. Average onset days of observed phenophases and average values of investigated biometeorological variables during the period 2000­2012 in forest stands. Onset day of the phenophase VIS_LU_10 SAT_LU_10 VIS_LC_10 Biometeorological variables AATII­IV [°C] AATIII­IV [°C] AATII [°C] AATIV [°C] AATVIII­IX [°C] AATIX [°C] nCD (-) nFD (-) CWBI­IV [mm] CWBIV­IX [mm] CWBVII­IX [mm] Ea, I­IV [mm] Ea, IV­IX [mm] Ea, VII­IX [mm] DII­IV (-) DIVII­IX (-) PIV [mm] Zelezná Studienka average st. dev. 108 4 108 4 270 3 6.2 8.8 1.0 11.8 18.3 15.7 73 24 106 -226 -118 166 336 220 0.48 1.86 191 1.2 1.1 2.6 1.7 1.1 1.7 22 9 52 135 79 33 58 61 0.16 0.87 51 Zvolen average 115 115 261 3.9 6.4 -1.0 9.4 15.9 13.4 99 31 160 -138 -88 184 325 210 0.32 1.47 228 st. dev. 4 4 9 1.2 1.0 2.4 1.4 1.1 1.7 16 12 67 157 86 35 46 62 0.14 0.57 64 average 119 123 252 1.3 3.7 -3.6 6.7 13.6 11.2 127 51 172 -99 -46 254 334 254 0.23 1.23 219 Poana st. dev. 4 5 4 1.1 1.0 2.1 1.4 1.1 1.8 20 14 65 207 103 66 33 66 0.09 0.51 63 The average onset day of the visually observed beginning of leaf colouring was on DOY 270. The earliest VIS_LC_10 occurred in 2003 on DOY 263, the latest onset occurred in 2010 on DOY 276. The year 2003 differed from the others with the temperature considerably above normal in June (+4.7 °C) and August (+4.8 °C), while the precipitation totals were below the normal nearly all year. In the year 2010, the precipitation totals had the highest positive deviation from normal, especially during the growing season (Fig. 3). In this stand, 70% of months of the period 2000­2012 had positive temperature deviations from normal, 28% of months had negative temperature deviations from normal and 2% of months were equal to normal. From the aspect of precipitation, 51% of months were below, 48% of months were above and 1% of months were equal to normal. Stand 2 ­ Zvolen The average visually observed onset day of beginning of leaf unfolding in the stand Zvolen was 115. The earliest onset was recorded on DOY 110 in 2007 and 2009, the latest on DOY 122 in 2005. The average MODIS-derived onset of the beginning of leaf unfolding was DOY 116. The earliest onset of SAT_LU_1 occurred on DOY 110 in 2009 and on DOY 111 in 2007. The average monthly air temperatures were considerably above the normal in the year 2007 in this stand too. The first Fig. 3. Differences between average monthly air temperatures and precipitation totals in the period 2000­2012 and in the reference period 1961­1990 in the stand Zelezná Studienka. T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 4. Differences between average monthly air temperatures and precipitation totals in the period 2000­2012 and in the reference period 1961­1990 in the stand Zvolen. half of this year was the warmest against the other years. In the year 2009, the highest average April air temperature was recorded. The years 2004 and 2005, when the latest onset of LU_10 were observed, were characterized by normal to below-normal average monthly air temperatures especially in the first half of the years (Fig. 4). The average onset day of beginning of leaf colouring was DOY 261. The leaves began yellowing at first on DOY 244 in 2003, when the precipitation totals of the growing season but especially in the summer were significantly below normal, and August was considerably hot. The latest onset was observed on DOY 275 in 2007, when the average monthly air temperatures were above normal during the whole growing season, and the precipitation totals in the period of August and September were normal. In this stand, 67% of months of the period 2000­2012 had positive temperature deviations from normal, 32% of months had negative temperature deviations from normal and 1% of months were equal to normal. From the aspect of precipitation, 53% of months were below and 47% of months were above normal. Stand 3 ­ Poana The visually observed average day of the onset of VIS_LU_10 was DOY 119. The earliest leaf unfolding started on DOY 111 in 2008, the latest on DOY 128 in 2003. The year 2008 (same as 2007 and 2002) was ranked among the years with considerably above-normal temperature, especially in the first half of the year. On the contrary, the year 2003 belonged to years with the coldest first quarter in the period 2000­2012. The MODIS-derived average onset day of SAT_LU_1 was DOY 123. The earliest beginning of leaf unfolding was recorded on DOY 111 in 2007. The latest onset of SAT_LU_1 was on DOY 131 in 2005. The air temperature in the first quarter of this year was below normal to normal (Fig. 5). In the stand Poana, the leaves started yellowing on average on DOY 252. The earliest visually observed VIS_LC_10 was recorded on DOY 245 in 2001 and 2002, the latest on DOY 257 in 2006 and 2010. The years 2001 and 2002 were characterized by sufficient to excess of precipitation amount during summer and hot temperatures. The average monthly air temperatures in the years 2006 and 2010 were above-normal during the growing seasons (Fig. 5). In the stand Poana, 73% of months of the period 2000­ 2012 had positive temperature deviations from the normal (1961­1990), 25% of months had negative temperature deviations from normal and 2% of months were equal to normal. From the aspect of precipitation, 60% of months were below normal, and 40% of months were above normal. 3.2. Dependence between onset of phenophases and biometeorological variables Spring period The significant dependence between temperature-based biometeorological variables and leaves unfolding, visually observed (VIS_LU_10) as well as satellite-derived (SAT_ LU_1), were discovered in the spring period. The higher AAT of the previous periods shifted the time of leaf unfolding to earlier days (Fig. 6). The average air temperature of the period from February to April was the most significant factor which evoked the onset of beginning of leaf unfolding in all three stands (Table 4). When considering only the VIS_LU_10, the effect of the number of chilling days was significant in Zelezná Studienka and Zvolen and number of frozen days in the stand Poana. When considering the SAT_LU_1, the number of chilling days was the most significant element evoking the beginning of leaf unfolding in Zelezná Studienka and the number of frozen days in Poana (Table 4). The increasing number of these days, when minimum or maximum daily air temperature did not exceed the temperature of 0°, delayed the onset of leaf unfolding. The evapotranspiration-based biometeorological variables had no effect on the onset of beginning of leaf unfolding observed with both methods. But we discovered that the high precipitation total of April (PIV), when the beginning of leaf unfolding usually begins, was the biometeorological element which caused the later leaf unfolding onset in all stands (Table 4). T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 5. Differences between average monthly air temperatures and precipitation totals in the period 2000­2012 and in the reference period 1961­1990 in the stand Poana. Table 4. Correlation between phenological phases (beginning of leaf unfolding and leaf coloring) and biometeorological variables in three investigated beech stands. Response variables Spring period VIS_LU_10 Explanatory variables AATII­IV AATIII­IV AATII AATIV nCD nFD CWBI­IV Ea DII­IV PIV AATII­IV AATIII­IV AATII AATIV nCD nFD CWBI­IV Ea, I­IV DII­IV PIV n Zelezná Studienka ryx b -0.72 -0.81 -0.34 -0.51 0.61 0.39 0.28 0.07 -0.24 0.56 -0.82 -0.77 -0.53 -0.59 0.67 0.55 0.31 0.04 -0.20 0.59 -0.49 0.05 0.77 0.50 0.73 0.45 -0.60 -2.4 -3.1 ns -1.2 0.11 ns ns ns ns 0.08 -2.6 -2.8 -0.8 -1.4 0.12 0.23 ns ns ns 0.08 1.3 ns 0.02 0.02 0.04 ns -2.0 Zvolen ryx -0.65 -0.45 -0.59 -0.38 0.56 0.36 0.16 0.03 -0.23 0.76 -0,81 -0.63 -0.69 -0.48 0.63 0.48 0.13 -0.01 -0.30 0.72 0.18 0.01 -0.14 0.08 -0.10 0.09 0.07 b -2.1 ns -1.0 ns 0.13 ns ns ns ns 0.09 -2.7 -2.3 -1.3 -1.3 0.15 0.15 ns ns ns 0.09 ns ns ns ns ns ns ns Poana ryx -0.79 -0.75 -0.57 -0.66 0.16 0.69 0.01 0.00 -0.20 0.46 -0.66 -0.56 -0.55 -0.51 0.54 0.65 0.12 0.01 -0.14 0.65 0.00 0.39 -0.15 0.04 -0.26 -0.05 0.11 b -2.6 -2.7 -1.1 -1.7 ns 0.23 ns ns ns 0.07 -2.9 -2.7 -1.3 -1.8 -0.30 0.19 ns ns ns 0.12 ns ns ns ns ns ns ns SAT_LU_1 apparent as in the case of leaf unfolding. The Climatic Water Balance of the period from July to September and also of the period from April to September had significant effect on visually observed beginning of leaf colouring only in the stand Zelezná Studienka ­ the lower the negative values of CWB, the earlier onset of leaf colouring (Fig. 7a). Also the Actual Evapotranspiration of the growing season (April­ September) was a significant factor on the VIS_LC_10 only in stand Zelezná Studienka ­ the lower the negative values of Ea, the earlier onset of leaf colouring (Fig. 7a). The higher Average Air Temperatures of August­September evoked the earlier onset of leaf colouring in the stand Zelezná Studienka (Fig. 7b). In this stand also the higher values of Dryness Index of the summer (July­September) affected the onset of VIS_LC_10 (Table 4). The low negative values of CWB and Ea together with high AAT of the months before onset of VIS_LC_10 caused the earlier onset of this phenophase only in the stand Zelezná Studienka at 304 m a.s.l. in the warm climatic area. In the other stands (in moderately warm and cold climatic areas) no statistically significant relationship with biometeorological variables and VIS_LC_10 was discovered. 4. Discussion Temperature effect on phenology in the spring period Some authors reported that the overcoming of the winter dormancy of European beech is effected by the previous freezing period, and that the effect of the increasing temperature and photoperiod is minimal (Falusi & Calamassi 1990; Kramer 1994; Caffarra & Donnelly 2011). However most authors consider just increasing temperature to be a driving force of the onset of spring phenophases (Braslavská 2000; Rötzer & Chmielewski 2001; Badeck et al. 2004; Menzel et al. 2006; Pálesová 2012 etc.). According to the previous studies, the onset day of spring phenophases depends most significantly on the temperature of the previous 2­3 months (Piao et al. 2006). The analyses of phenological and meteorological data from the International Phenological Gardens revealed that the beginning of the growing season was affected mainly by the average air temperature of March, 45 Autumn period VIS_LC_10 AATVIII­IX AATIX CWBIV­IX CWBVII­IX Ea, IV­IX Ea, VII­IX DIVII­IX AAT ­ average air temperature, nCD ­ number of chilling days, nFD ­ number of frozen days, CWB ­ climatic water balance, Ea ­ actual evapotranspiration, DI ­ dryness index, P ­ precipitation total , n ­ number of analyzed years, ryx ­ correlation coefficient, b ­ regression coefficient, significant values are highlighted by bolt, ns ­ non significance. Autumn period The relationship between investigated biometeorological variables and the first autumn phenophase was not as T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 6. Relationships between visually observed and MODIS-derived onset days of the beginning of leaf unfolding (LU_10) and average air temperatures calculated for the periods February­April (AAT_II­IV), March­April (AAT_III­IV), February (AAT_II), and April (AAT_IV). 46 T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 7. Relationship between investigated biometeorological variables and beginning of leaf colouring (VIS_LC_10) in the stand Zelezná Studienka: a) CWB_IV­IX ­ Climatic Water Balance during the period April­September; CWB_VII­IX ­ Climatic Water Balance during the period July­September; EA_IV­IX ­ Actual Evapotranspiration during the period April­September; b) AAT_VIII­IX ­ Average Air Temperature during the period August­September. but also of February and partially of April ­ the higher the temperature in late winter, the earlier onset of leaf unfolding (Chmielewski & Rötzer 2001). Menzel et al. (2006) found a clear reaction of the plant phenological phases to the increasing temperature. Most of the phenophases correlated with the average temperature of the month when the phenophase began and of the previous month. The trend in leaf unfolding of European beech agreed with the increasing temperature in March (r = -0.86). Our results confirmed that the beginning of leaf unfolding of European beech depended on the air temperature of the previous 2­3 months. We revealed the most significant relationship between average air temperature of the period February­April and the beginning of leaf unfolding in all three stands. But the other temperature based biometeorological variables had a significant effect. Previous studies from Slovakia revealed a significant relationship between Average Air Temperature of the period March­April (PTVIII­IV) and LU_10: i) Barna et al. (2009) ­ r = -0.61, ii) Schieber et al. (2009) ­ r = -0.86, iii) Braslavská (2000) on the localities below 500 m a.s.l. ­ r = -0.83, and on the localities above 500 m a.s.l. ­ r = -0.76. In this study similar results were discovered with data from both the satellite and visual phenological methods. In the stand Zelezná Studienka (below 500 m a.s.l.) we revealed ­ r = -0.81 and -0.77, for visual and satellite method respectively. In the stands Zvolen and Poana (above 500 m a.s.l.) we revealed - r = ­0.45 and ­0.63 for Zvolen and ­ r = -0.75 and -0.56 for Poana, for visual and satellite method respectively (Table 4). Higher temperatures of the months before the onset of spring phenological phases caused the shift to the earlier onset. The bud bursting of European beech advanced by 3.2 days·°C-1 (Rötzer & Chmielewski 2001) to 3.6 days·°C-1 (Kramer 1996). On the contrary Menzel et al. (2001) pointed out a weaker sensitivity of beech to increasing winter and spring temperatures. Also Vitasse et al. (2009a) reported, that beech showed weaker sensitivity to higher spring temperatures with the advance of bud bursting by 2.1 days·°C-1 and advance of leaf unfolding by 1.9 days·°C-1. Our analyses revealed the advance of the beginning of leaf unfolding by 2.1­2.9 days·°C-1 of average temperature of the period February­April, 2.3­3.1 days·°C-1 of average temperature of the period March­April, 0.8­1.3 days·°C-1 of average air temperature of February and 1.2­1.8 days·°C-1 of average air temperature of April for particular stands and methods (Table 4). A much bigger advance of the beginning of leaf unfolding by 2.5 days·°C-1 of average temperature of April found by Pálesová (2012). Besides average temperature, beginning of leaf unfolding significantly depended on the number of chilling and frozen days as well. The increasing number of chilling days delayed the onset of VIS_LU_10 and SAT_LU_1 by 1.5­ 2.3 days·10 days-1 and increasing number of frozen days led to delay of VIS_LU_10 and SAT_LU_1 by 1.1­3 days 10 days-1 (Table 4) for a particular stand. Pálesová (2012) also found the delay of bud swelling with increasing number of chilling and frozen days. The significant impact of the temperature also revealed the evaluation of timing of the earliest and the latest onset days of the beginning of leaf unfolding from deviations in temperature and precipitation from normal on beech stands. The earliest onset of the beginning of leaf unfolding occurred in the years when the temperature of months of the first quarter of the year were considerably above normal and higher than in the other years. On the contrary in the period 2000­2012, the latest onset days of beginning of leaf unfolding occurred in the years with the coldest first quarter. Temperature effect on phenology in autumn period Most of the studies focused on the onset of autumn phenophases (Chmielewski & Rötzer 2001; Menzel 2002; Sparks & Menzel 2002) showed a weak relationship between tem47 T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 perature and the onset. Our study revealed dual results. The average air temperature of the period August­September had a significant effect on the onset of leaf colouring, but only in the stand Zelezná Studienka where the VIS_LC_10 advanced by 1.3 days·°C-1. However in the other stands the effect of temperature was not significant (Table 4). Estrella & Menzel (2006) and Cufar et al. (2012) found the opposite relationship between the air temperature and leaf colouring (r = 0.56), when the increasing temperature delayed the onset of this phenophase. The reason may be the location of our stand Zelezná Studienka in a warm climatic area, where the high temperatures in summer could have a negative effect on the condition of trees and leaves. Vitasse et al. (2009b) considered just average temperature of the period August­ November to be the main factor affecting the general leaf colouring of beech forests in France. Climatic water balance effect on phenology The importance of temperature and day length in determining the end of the growing season varies among plant species with certain groups solely dependent on photoperiod (Howe et al. 1996). This dependence on the length of the photoperiod has a substantially positive impact on forests in the current global warming conditions: while temperatures may increase over time, day length remains unchanged, limiting the potential of woody plants controlled mainly by the length of the photoperiod to extend their growing season (Way 2011). In this study, we did not test the effect of the length of photoperiod on the onset of phenophases, but we used this parameter as an input for the calculation of potential evapotranspiration according to Tornthwait and Mater (1955). Skvarenina et al. (2004) considered the climatic water balance to be the appropriate variable for characterizing the climatic conditions of the area. The input values for calculation were just potential evapotranspiration and precipitation (Baumgartner & Liebscher 1990). Following this assumption, we tested the effect of climatic water balance, actual evapotranspiration, dryness index and precipitation on the onset of spring and autumn phenophases. ­ The study stands Zelezná Studienka and Zvolen were located in the 3rd vegetation belt (oak-beech). According to Walter's climadiagram, the precipitation prevails over the evapotranspiration. However, the climatic water balance climadiagram pointed out that in the period March­August potential evapotranspiration prevails over the precipitation and the lack of water is less than 100 mm. In the stand Zelezná Studienka, the CWB of the period April­September was on average ­226 mm, and CWB of the period July­September was ­118 mm (Table 3). The effect of these CWBs on the onset of leaf colouring was significant (Table 4). CWB of the period January­April was not significant; the CWB values were positive (on average 106 mm). This was caused by the low level of potential evapotranspiration before the growing season. The Actual Evapotranspiration in this period was also lower ­ on average 166 mm ­ than in the summer season (July­September), when it was on average 220 mm (Table 3). The Dryness Index for the summer season was also significant in this stand (Table 4). In the years 2004 and 2012 it exceeded the value 2.3, which indicated a semi-arid climate regime and in 2003 DI (3.8) indicated even an arid regime. However, the average DI (1.9, Table 3) of the period 2000­2012 characterized the a semi-humid regime. In the Zvolen stand, the effect of CWB, Ea, DI was not significant (Table 4), although this stand is located in the same belt as Zelezná Studienka. The CWB of the period April­September was on average ­138 mm, and CWB of the period July­ September was ­88 mm (Table 3). These values are similar to that published in Skvarenina et al. (2002). The dryness index of the summer season was in comparison with Zelezná Studienka smaller, with maximal value 2.2 in the year 2007. The average DI in the period 2000­2012 was 1.5 (Table 3). All these values classified the climate regime in the Zvolen stand as a semi-humid. ­ The Poana stand is located in the 5th vegetation belt (firbeech). The water balance in the general growing season is positive. The precipitation totals in the summer exceed the sum of 100 mm. The effect of CWB, Ea, DI on the beginning of leaf unfolding and leaf colouring was not significant. The maximal dryness index was 2 in the year 2009. The average summer DI in the period 2000­2012 was 1.2 (semi-humid) with 6 humid summers in this period. The precipitation totals of summer really exceeded 100 mm published for this vegetation belt by Skvarenina et al. (2002). However, the climatic water balance still had negative values in both analyzed periods (April­September and July­September) (Table 3). The timing of the onset of spring and autumn phenophases reflect the sensitivity of European beech to the amount of available water in the soil profile (Stelcová et al. 2008; Nielsen & Jorgensen 2003). Bagar et al. (2001) found out that increasing sums of effective air temperatures together with lower precipitation totals caused earlier termination of assimilation and thus earlier onset of the leaf colouring. We discovered a similar pattern in the stand Zelezná Studienka, where the high average air temperatures of the period August­September advanced the phenophase VIS_LC_10. Schieber et al. (2009) detected the dependence between cumulative precipitation total of the period May­ August and the leaf colouring in beech forests (r = 0.58). In this study, we tested this biometeorological variable only in relation to beginning of leaf unfolding and found a significant dependence between the precipitation total of April and VIS_LU_10 and SAT_LU_1. Spring rainfalls are usually accompanied by the drop of temperature. This could be the reason why the higher precipitation totals delayed the onset of the beginning of leaf unfolding. According to findings of CWB published in Minás et al. (2011), deterioration of conditions could be expected for beech up to the 4th vegetation belt and improvement of conditions for beech from the 5th to the 7th vegetation belts. We found a statistically significant effect of CWB only in the stand Zelezná Studienka (3rd veg. belt), where the low negative values of CWB caused earlier onset of leaf colouring. The significant impact of the high temperature and low amount of available water was also visible from the evaluaount unt tion of the earliest and latest onset days of the beginning of leaf colouring from deviations in temperature and precipitation from normal. This corresponded with results of Kramer (1995). He presented that higher average temperatures in the period of summer-autumn together with the previous dryness period forced the onset of autumn phenophases. T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 5. Conclusion It was found that air temperature is one of the most significant elements affecting the onset of spring phenophases in the tested area of beech stands. In this study, we calculated biometeorological elements of the beginning of leaf unfolding based on the minimum, maximum and average daily air temperatures: Average Air Temperature of the period February­April (AATIIIV), Average Air Temperature of the period March­April (AATIII­IV), Average Air Temperature of February (AATII), Average Air Temperature of April (AATIV), Number of chill and frozen days (nCD, nFD). We also tested the effect of evapotranspiration-based biometeorological variables: Climatic Water Balance of the period January­ April (CWBIV­IX), Actual Evapotranspiration (Ea), Dryness Index (DI) and precipitation total of April (PIV). The average air temperature of the period from February to April was the most significant factor which evoked the onset of beginning of leaf unfolding in all three stands. Compared with visual phenological observations, for satellite observation more significant correlations with biometeorological variables were found. This could result from the subjective error of an observer when he evaluates the phenological phases visually. This error is eliminated by the use of the digital image analyses of the satellite data. The dependence of the onset of autumn phenophases on the external biometeorological variables has been poorly studied and characterized. From the variables which may affect the onset of autumn phenophases we had chosen: Climatic Water Balance of the growing period (April­ September) (CWBIV­IX) and of the period July­September (CWBVII­IX), Actual Evapotranspiration (Ea), Dryness Index (DI), Average Air Temperature of the period August­September (AATVIII­IX) and Average Air Temperature of September (AATIX). The effect of these biometeorological variables on the beginning of leaf colouring was significant only in the stand Zelezná Studienka. The analysis of the relationship between biometeorological variables and onset of the phenophases of the period 2000­2012 pointed to the inappropriate climate conditions for beech in the 3rd vegetation belt in warm climate area. As suspected Skvarenina et al. (2004), here the unfavorable cli, matic conditions ­ high temperatures in August and September and low negative values of climatic water balance in the summer season ­ caused an earlier onset of leaf colouring and shortening of the growing season. On the contrary, the high temperatures in the early spring caused the advance of beginning of leaf unfolding and thereby increased the risk of damage by the late frosts. Therefore, the cultivation of beech in the 3rd vegetation belt requires increasing attention paid to the climate conditions. This kind of research is necessary to achieve further noticeable progress in phenological modelling by using the expanding area of remote sensing observation. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Forestry Journal de Gruyter

Effect of biometeorological variables on the onset of phenophases derived from MODIS data and visual observations

Download PDF
13 pages

Loading next page...
 
/lp/de-gruyter/effect-of-biometeorological-variables-on-the-onset-of-phenophases-li3KpB80kl

References (57)

Publisher
de Gruyter
Copyright
Copyright © 2014 by the
ISSN
0323-1046
eISSN
0323-1046
DOI
10.2478/forj-2014-0004
Publisher site
See Article on Publisher Site

Abstract

In this study we analyzed the effect of selected biometeorological variables on the onset of phenophases in three beech stands in different climatic areas (warm, moderately warm and cold). We have focused on two phenophases ­ leaf unfolding and leaf colouring. Timing of both phenophases was identified visually and using series of MODIS satellite images. The data were collected during a 13-year period (2000­2012). For the spring period, we found a significant dependence between temperature and precipitation-based biometeorological variables and leaf unfolding in both datasets ­ those based on visual and remote sensing-based observations. The average air temperature in the period from February­April was the most significant factor which initiated the onset of beginning of leaf unfolding in all three investigated stands. The evapotranspiration-based biometeorological variables (climatic water balance, actual evapotranspiration, dryness index) had no effect on the onset of the beginning of leaf unfolding observed using both methods. The high precipitation totals in April caused the later onset of leaf unfolding in all stands. The relationship between the first autumn phenophase ­ leaf colouring and biometeorological variables was found significant in beech stand in the warm climatic area only. Keywords: Fagus sylvatica; phenology; biometeorological variables; MODIS Abstrakt V práci sme analyzovali vplyv biometeorologických premenných na nástup fenologických fáz v troch bukových porastoch nachádzajúcich sa v rozdielnych klimatických oblastiach (teplej, mierne teplej a studenej). Sledovanými fenofázami boli zaciatok zalistenia a zaciatok zltnutia listov. Ich nástup bol sledovaný vizuálne a s pouzitím satelitných údajov MODIS. Pozorovania prebiehali pocas 13-rocného obdobia (2000­2012). V jarnom období sme zistili významný vplyv biometeorologických premenných zalozených na teplote a zrázkach na zaciatok zalistenia bukových porastov zisteného obomi ­ vizuálnym aj satelitným sledovaním. Najvýznamnejsou z týchto premenných bola priemerná teplota vzduchu od februára do apríla, ktorá vyvolávala skorsí nástup zaciatku zalistenia na vsetkých troch bukových stanovistiach. Vyssie úhrny zrázok v mesiaci apríl spôsobovali na sledovaných stanovistiach neskorsí nástup tejto fenofázy. Naopak, premenné zalozené na evapotranspirácií (klimatická vodná bilancia, aktuálna evapotranspirácia, index sucha) nemali ziaden vplyv na zaciatok zalistenia zistený oboma metódami. Vzah medzi prvou jesennou fenofázou ­ zaciatkom zltnutia listov a biometeorologickými premennými bol významný iba na bukovom stanovisti nachádzajúcom sa v teplej klimatickej oblasti. Kúcové slová: Fagus sylvatica; fenológia; biometeorologické premenné; MODIS 1. Introduction The reactions of vegetation to climate change conditions have been widely discussed in recent studies on phenology, climatology and remote sensing. The phenological phases responding to the fluctuation of meteorological elements are sensitive and easy-to-observe reactions of vegetation to climate change (Badeck et al. 2004). In the middle latitudes, the onset of the phenophases like bud bursting, leaf unfolding, flowering, etc. is effected by the air temperature (Ahas et al. 2000; Sparks et al. 2000; Defila & Clot 2000; Menzel 2003). It was proved that higher temperatures during the spring period evoke earlier onset of tree species spring phenophases (Chmielewski & Rötzer 2001; Menzel et al. 2006; Kramer 1996; Pálesová 2012). But there are other meteoro- logical elements that could be limiting environmental factors for tree species in the climate change conditions: the lack of rainfall amount and uprising evapotranspiration during the growing season (Skvarenina et al. 2004). In Slovakia, there is the assumption that changed bioclimatic conditions especially from the 1st to the 3rd (partially the 4th) vegetation belt disturb existing plant (primarily beech) communities and their species compositions. However, favourable temperature conditions with sufficient water balance in the higher vegetation tiers (from the 5th) may represent climatic conditions appropriating the presence of deciduous tree species like beech (Skvarenina et al. 2004). In the past, the onset of phenophases could be observed only visually. During the last decades, the methods of phenological observations using remote sensing data (Kang et *Corresponding author. Veronika Lukasová, e-mail: veronika.brandysova@gmail.com T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 al. 2003; Zhang et al. 2003; Fisher & Mustard 2007), and simple or hemispherical digital photography (Nagai et al. 2010; Mozný et al. 2012; Felts et al. 2011) have been developed. The remote sensing data have significant potential in monitoring of vegetation dynamics on a regional to global scale. The first studies from the field of remote sensing phenology utilized data from the radiometer AVHRR. The dates of the beginning and end of the growing season were derived and the length of the growing season was calculated. Later, after the Terra and Aqua spacecrafts with the spectroradiometer MODIS were launched into orbit, the phenology was monitored more precisely (Zhang et al. 2003; Fisher et al. 2006; Soudani et al. 2012). Vegetation indices were proposed to be the biophysical indicators of the changing amount and quality of green vegetation, e.g. NDVI ­ Normalized Difference Vegetation Index (Beck et al. 2006; Narasihman & Stow 2010; Soudani et al. 2012), EVI ­ Enhanced Vegetation Index (Zhang et al. 2003; Ganguly et al. 2010), PVI ­ Perpendicular Vegetation Index (Guyon et al. 2011), WDRVI ­ Wide Dynamic Range Vegetation Index (Jönsson et al. 2010). The vegetation index commonly used in remote sensing phenology is the ratio index NDVI, which is calculated from the reflectance of vegetation in the red and near infrared band. Healthy (green) vegetation absorbs the incoming visible radiation and reflects the near infrared radiation (Weier & Herring 2011). This study was focused on the forests stands with dominant occurrence of European beech ­ Fagus sylvatica L. (subspecies F. sylvatica ssp. sylvatica), which is continually distributed across Europe. Beech is a competing proficient tree species, which does not grow only in very shallow soils with low water capacity, on wet or flooded soils with permanently or periodically low air capacity, and in the harsh mountain climate. The lower border of the European beech extension belongs to the warm climate area, dry to moderately moist precinct. The general occurrence of beech forests is localized in moderately warm to cold climate area, moderately to very moist precinct within the 2nd to 6th altitudinal vegetation belt. The upper border of sporadic occurrence of beech is localized on the transition between the moderately cold and cold mountain precincts in the cold climate area (Gömöry et al. 2011). In this study, we examined what are the limiting meteorological variables that had the most significant effect on the onset of vegetative phenophases of European beech. and cold (Table 1). The three beech stands were selected from the phenology network of the Slovak Hydrometeorological Institute. The homogeneity of beech stands in surrounding areas was verified on the basis of the Forest Management Plan database and Landsat-based tree species composition map of the Slovak forests as well (Bucha 1999). The species composition map derived from Landsat was aggregated from 30 m to 250 m pixels corresponding to MODIS images. The 250 m pixels with more than 60% of beech were classified as a "beech" class and used for further satellite based phenology analyses. The pixels on the boundary between beech forests and other land cover classes or forest types were excluded due to possible spectral contamination (Wolfe et al. 2002). 2.2. Visual phenological observations In each stand, phenological observations were performed on a sample of ten representative overstory beeches. The method by Braslavská & Kamenský (1996) was used during the 13-year period (2000­2012). In spring, one phenological stage of leaf unfolding was observed: first leaf (VIS_LU_10), when 10% of leaves of monitored trees were unfolded. In autumn, one stage of the phenophase leaf colouring was observed: beginning leaf colouring (VIS_LC_10), when 10% of leaves of monitored trees changed their colour to yellow or brown. 2.3. Phenology metrics from MODIS data The MOD09GQ (MYD09GQ) daily surface reflectance products from the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard Terra and Aqua satellites were used to calculate the seasonal course of NDVI. The MOD09GQ (MYD09GQ) consists of the red spectral band (RED 620­670 nm) and near infrared spectral band (NIR 841­876 nm) with 250 m spatial resolution. We checked the recorded reflectance at pixel level in each individual image for possible clouds; fog, aerosols, etc. according to the flags in the quality control data (MOD09GA, MYD09GA) (Vermote et al. 2011). The image pixels that did not match the quality criteria in the quality analysis, were eliminated from further analyses. Based on the MOD09GQ and MYD09GQ products we calculated daily NDVI for each of the stands in the period 2000­2012. NDVI = NIR ­ RED NIR + RED [1] 2. Material and methods 2.1. Investigated forest stands The beech stands analyzed in this study were established in three contrasting climatic areas ­ warm, moderately warm Table 1. Description of investigated beech stands. Stand Zelezná Studienka Zvolen Poana Altitude [m a.s.l.] 304 566 1051 Aspect NW SW SE Phenological metrics. The normalized difference vegetation index (NDVI) was considered as a suitable indicator of changing amount, quality and structure of European beech assi- Tree species composition [%] beech coniferous deciduous 60 20 20 65 -- 35 80 15 5 Climatic area Warm Moderately warm Cold T [°C] 9.4 7.6 4.9 P [mm] 645 745 924 T ­ average annual air temperature during period 1961­1990; P ­ average annual precipitation totals during period 1961­1990. T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 milatory organs (Brandýsová & Bucha 2012). NDVI temporal profiles were modelled with sigmoid logistic function proposed by Fisher et al. (2006) using the software program Phenological profile (Bucha & Kore 2009): stands were selected on the basis of minimal horizontal and vertical distance (Fig. 2). Table 2. Meteorological stations assigned to the phenological stands and distance between them. Phenological stand Zelezná Studienka Zvolen Poana Meteorological station Bratislava ­ Koliba Banská Stiavnica Lom nad Rimavicou Horizontal distance [km] 3 21 12 Vertical distance [m] 17 9 33 v(t) = vmin + vamp 1 1+e m -n t 1 1 1 1+e m -n t 2 2 [2] Where v(t) represents the value of the vegetation index NDVI. We used a four-parameter method (4D). Parameters vmin and vamp were predefined (Bucha et al. 2011). They represent the minimal value (vmin = 0.4295) and amplitude of NDVI (vamp = 0.4971) of beech stands in Slovakia. Phenological metrics was taken from the study Brandýsová (2013). The satellite-derived phenological markers were identified by inflection point of that sigmoid logistic function (Fig. 1) as follows: ­ Inflection point in the spring period ­ SAT_LU_1, as a time when the assimilatory organs of overstory beeches started unfolding; ­ Inflection point in the autumn period ­ SAT_LC_2, as a time when the assimilatory organs of overstory beeches were almost recoloured. We used only the first phenological marker in spring: SAT_ LU_1. This marker corresponds with the visually observed phenophase beginning of leaf unfolding (Brandýsová 2013). As the dependence between satellite derived and visually observed onset of autumn phenophases was not revealed in our previous research (Brandýsová 2013), only the effect of selected biometeorological variables on visually observed leaf coloring phase was investigated. The differences between monthly average air temperatures and precipitation totals from the 30-year normal were calculated. The earliest and latest days of the onset of phenophases of the period 2000­2012 were considered in terms of these differences. We defined biometeorological variables based on the air temperature, precipitation and evapotranspiration to find out which of them affected most significantly the onset of visually observed and satellite derived phenological phases. i) Average air temperature (AATp) of defined period: where Taverage is average daily air temperature, n is number of days. We calculated: AATp = STaverage n [3] ­ average air temperature of the period February­April: AATII­IV, ­ average air temperature of the period March­April: AATIII­IV, ­ average air temperature of February: AATII, ­ average air temperature of April: AATIV, ­ average air temperature of the period August­September: AATVIII­IX, ­ average air temperature of September: AATIX. ii) Number of days below a prescribed temperature threshold: ­ number of chilling days (nCD; Tmin < 0): number of days with minimum daily air temperature below 0 °C, counted from the end of the growing season in the previous year to the onset day of the beginning of leaf unfolding, ­ number of frozen days (nFD; Tmax < 0): number of days with maximum daily air temperature below 0 °C, counted from the end of the growing season in the previous year to the onset day of the beginning of leaf unfolding. iii) Climatic water balance (CWB): ­ climatic water balance of the period January­April: CWBI­IV, ­ climatic water balance of the period April­September: CWBIV­IX, ­ climatic water balance of the period July­September: CWBVII­IX, CWB = P ­ PET [4] where P is precipitation total (mm) and PET is potential evapotranspiration. PET was calculated according to a modified method of Tornthwait and Mather (1955) (Novák 1995): 41 Fig. 1. Inter-annual course of normalized differentiated vegetation index (NDVI) in selected beech stand in Slovakia fitted by the sigmoid logistic function. The inflection points of this function were used for the evaluation of forest stands phenology. Two phenophases were marked ­ beginning of leaf unfolding (SAT_LU_1) and leaf coloring (SAT_LC_2). On the x axis: day of year, on the y axis: NDVI. 2.4. Biometeorological variables The data from meteorological stations (Table 2) were obtained from the Slovak Hydrometeorological Institute. The data were used to analyze the effect of local meteorological conditions (especially temperature, precipitation and evapotranspiration-based) on the onset of phenological phases in beech stands. The meteorological stations from which the data were transferred to the positions of investigated forest T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 2. Location of 3 beech test stands ­ marked with a black circle, and meteorological stations ­ marked with a square. PET = 0.535 . f 10 . Tm I [5] where PET (mm) is potential evapotranspiration of months, Tm (°C) average monthly air temperature, f is correction factor depending of the length of the month and latitude: f = k . so [6] where k is number of days coefficient, so is maximal duration of sunshine during the day (h). I is temperature index calculated as a sum of twelve monthly indices: 12 [7] Tm 1.514 ­ 1.1<DI2.3 ­ semi-humid (moderately insufficient moisture; savannah), ­ 2.3<DI3.4 ­ semi-arid (insufficient moisture; semidesert), ­ 3.4<DI10 ­ arid (very insufficient moisture; desert), ­ 10<DI ­ hyper-arid (extremely insufficient moisture; desert) (Gao & Giorgi 2008; Skvarenina et al. 2009). vi) Precipitation total of April: PIV I= Si; j j=1 ij = 3. Results 3.2. Climatic variability and onset of phenophases in investigated stands Average values of biometeorological variables derived from meteorological data recorded during the period 2000­2012 are noted in table 3. The between-stand differences in biometeorological variables and onset dates of the phenophases corresponded to climatic areas, where the test stands are located. Stand 1 ­ Zelezná Studienka The average onset day of the visually observed phenophase VIS_LU_10 in the stand Zelezná Studienka was in the period 2000­2012 DOY 108. The earliest onset was recorded on DOY 102 in the years 2007 and 2009; the latest onset day was on DOY 114 in theyears 2004 and 2006. The average onset day of satellite-derived beginning of leaf unfolding was the same as that from visual observations. The earliest onset was on DOY 101 in 2007 and the latest was on DOY 113 in 2004. The year 2007, when both methods recorded the earliest onset of LU_10, had a distinctive positive temperature deviation from normal during the winter and spring months compared to other years. On the contrary, in the year 2004, when the latest onset day of LU_10 was observed, the absolute temperature differences from normal were especially during the first three months the lowest in comparison with other years with some negative values (Fig. 3). and a is exponent calculated using a = (0.0675 . I3 ­ 7.71 . I2 + 1792 . I + 47239) . 10­5 iv) Actual evapotranspiration Ea of Turc (1961) (Novák 1995): ­ actual evapotranspiration of the period January­April: Ea,I­IV, ­ actual evapotranspiration of the period April­September: Ea,IV­IX, ­ actual evapotranspiration of the period July­September: Ea,VII­IX, Ea = P . [0.9 + (P2 . L­2)]­1/2 [8] where P is precipitation total (mm), and L = 300 + 2.5T + 0.05T, where T is average air temperature of the period. v) Dryness index (DI) of Budyko (1958): ­ dryness index of the period January­April: DII­IV, ­ dryness index of the period July­September: DIVII­IX, DI = PET / P [9] where PET [mm] is potential evapotranspiration of the period, P is precipitation total [mm]. Thresholds for different climate regimes were defined as: ­ 0<DI1.1 ­ humid (surplus moisture regime; steppe to forest vegetation), T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Table 3. Average onset days of observed phenophases and average values of investigated biometeorological variables during the period 2000­2012 in forest stands. Onset day of the phenophase VIS_LU_10 SAT_LU_10 VIS_LC_10 Biometeorological variables AATII­IV [°C] AATIII­IV [°C] AATII [°C] AATIV [°C] AATVIII­IX [°C] AATIX [°C] nCD (-) nFD (-) CWBI­IV [mm] CWBIV­IX [mm] CWBVII­IX [mm] Ea, I­IV [mm] Ea, IV­IX [mm] Ea, VII­IX [mm] DII­IV (-) DIVII­IX (-) PIV [mm] Zelezná Studienka average st. dev. 108 4 108 4 270 3 6.2 8.8 1.0 11.8 18.3 15.7 73 24 106 -226 -118 166 336 220 0.48 1.86 191 1.2 1.1 2.6 1.7 1.1 1.7 22 9 52 135 79 33 58 61 0.16 0.87 51 Zvolen average 115 115 261 3.9 6.4 -1.0 9.4 15.9 13.4 99 31 160 -138 -88 184 325 210 0.32 1.47 228 st. dev. 4 4 9 1.2 1.0 2.4 1.4 1.1 1.7 16 12 67 157 86 35 46 62 0.14 0.57 64 average 119 123 252 1.3 3.7 -3.6 6.7 13.6 11.2 127 51 172 -99 -46 254 334 254 0.23 1.23 219 Poana st. dev. 4 5 4 1.1 1.0 2.1 1.4 1.1 1.8 20 14 65 207 103 66 33 66 0.09 0.51 63 The average onset day of the visually observed beginning of leaf colouring was on DOY 270. The earliest VIS_LC_10 occurred in 2003 on DOY 263, the latest onset occurred in 2010 on DOY 276. The year 2003 differed from the others with the temperature considerably above normal in June (+4.7 °C) and August (+4.8 °C), while the precipitation totals were below the normal nearly all year. In the year 2010, the precipitation totals had the highest positive deviation from normal, especially during the growing season (Fig. 3). In this stand, 70% of months of the period 2000­2012 had positive temperature deviations from normal, 28% of months had negative temperature deviations from normal and 2% of months were equal to normal. From the aspect of precipitation, 51% of months were below, 48% of months were above and 1% of months were equal to normal. Stand 2 ­ Zvolen The average visually observed onset day of beginning of leaf unfolding in the stand Zvolen was 115. The earliest onset was recorded on DOY 110 in 2007 and 2009, the latest on DOY 122 in 2005. The average MODIS-derived onset of the beginning of leaf unfolding was DOY 116. The earliest onset of SAT_LU_1 occurred on DOY 110 in 2009 and on DOY 111 in 2007. The average monthly air temperatures were considerably above the normal in the year 2007 in this stand too. The first Fig. 3. Differences between average monthly air temperatures and precipitation totals in the period 2000­2012 and in the reference period 1961­1990 in the stand Zelezná Studienka. T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 4. Differences between average monthly air temperatures and precipitation totals in the period 2000­2012 and in the reference period 1961­1990 in the stand Zvolen. half of this year was the warmest against the other years. In the year 2009, the highest average April air temperature was recorded. The years 2004 and 2005, when the latest onset of LU_10 were observed, were characterized by normal to below-normal average monthly air temperatures especially in the first half of the years (Fig. 4). The average onset day of beginning of leaf colouring was DOY 261. The leaves began yellowing at first on DOY 244 in 2003, when the precipitation totals of the growing season but especially in the summer were significantly below normal, and August was considerably hot. The latest onset was observed on DOY 275 in 2007, when the average monthly air temperatures were above normal during the whole growing season, and the precipitation totals in the period of August and September were normal. In this stand, 67% of months of the period 2000­2012 had positive temperature deviations from normal, 32% of months had negative temperature deviations from normal and 1% of months were equal to normal. From the aspect of precipitation, 53% of months were below and 47% of months were above normal. Stand 3 ­ Poana The visually observed average day of the onset of VIS_LU_10 was DOY 119. The earliest leaf unfolding started on DOY 111 in 2008, the latest on DOY 128 in 2003. The year 2008 (same as 2007 and 2002) was ranked among the years with considerably above-normal temperature, especially in the first half of the year. On the contrary, the year 2003 belonged to years with the coldest first quarter in the period 2000­2012. The MODIS-derived average onset day of SAT_LU_1 was DOY 123. The earliest beginning of leaf unfolding was recorded on DOY 111 in 2007. The latest onset of SAT_LU_1 was on DOY 131 in 2005. The air temperature in the first quarter of this year was below normal to normal (Fig. 5). In the stand Poana, the leaves started yellowing on average on DOY 252. The earliest visually observed VIS_LC_10 was recorded on DOY 245 in 2001 and 2002, the latest on DOY 257 in 2006 and 2010. The years 2001 and 2002 were characterized by sufficient to excess of precipitation amount during summer and hot temperatures. The average monthly air temperatures in the years 2006 and 2010 were above-normal during the growing seasons (Fig. 5). In the stand Poana, 73% of months of the period 2000­ 2012 had positive temperature deviations from the normal (1961­1990), 25% of months had negative temperature deviations from normal and 2% of months were equal to normal. From the aspect of precipitation, 60% of months were below normal, and 40% of months were above normal. 3.2. Dependence between onset of phenophases and biometeorological variables Spring period The significant dependence between temperature-based biometeorological variables and leaves unfolding, visually observed (VIS_LU_10) as well as satellite-derived (SAT_ LU_1), were discovered in the spring period. The higher AAT of the previous periods shifted the time of leaf unfolding to earlier days (Fig. 6). The average air temperature of the period from February to April was the most significant factor which evoked the onset of beginning of leaf unfolding in all three stands (Table 4). When considering only the VIS_LU_10, the effect of the number of chilling days was significant in Zelezná Studienka and Zvolen and number of frozen days in the stand Poana. When considering the SAT_LU_1, the number of chilling days was the most significant element evoking the beginning of leaf unfolding in Zelezná Studienka and the number of frozen days in Poana (Table 4). The increasing number of these days, when minimum or maximum daily air temperature did not exceed the temperature of 0°, delayed the onset of leaf unfolding. The evapotranspiration-based biometeorological variables had no effect on the onset of beginning of leaf unfolding observed with both methods. But we discovered that the high precipitation total of April (PIV), when the beginning of leaf unfolding usually begins, was the biometeorological element which caused the later leaf unfolding onset in all stands (Table 4). T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 5. Differences between average monthly air temperatures and precipitation totals in the period 2000­2012 and in the reference period 1961­1990 in the stand Poana. Table 4. Correlation between phenological phases (beginning of leaf unfolding and leaf coloring) and biometeorological variables in three investigated beech stands. Response variables Spring period VIS_LU_10 Explanatory variables AATII­IV AATIII­IV AATII AATIV nCD nFD CWBI­IV Ea DII­IV PIV AATII­IV AATIII­IV AATII AATIV nCD nFD CWBI­IV Ea, I­IV DII­IV PIV n Zelezná Studienka ryx b -0.72 -0.81 -0.34 -0.51 0.61 0.39 0.28 0.07 -0.24 0.56 -0.82 -0.77 -0.53 -0.59 0.67 0.55 0.31 0.04 -0.20 0.59 -0.49 0.05 0.77 0.50 0.73 0.45 -0.60 -2.4 -3.1 ns -1.2 0.11 ns ns ns ns 0.08 -2.6 -2.8 -0.8 -1.4 0.12 0.23 ns ns ns 0.08 1.3 ns 0.02 0.02 0.04 ns -2.0 Zvolen ryx -0.65 -0.45 -0.59 -0.38 0.56 0.36 0.16 0.03 -0.23 0.76 -0,81 -0.63 -0.69 -0.48 0.63 0.48 0.13 -0.01 -0.30 0.72 0.18 0.01 -0.14 0.08 -0.10 0.09 0.07 b -2.1 ns -1.0 ns 0.13 ns ns ns ns 0.09 -2.7 -2.3 -1.3 -1.3 0.15 0.15 ns ns ns 0.09 ns ns ns ns ns ns ns Poana ryx -0.79 -0.75 -0.57 -0.66 0.16 0.69 0.01 0.00 -0.20 0.46 -0.66 -0.56 -0.55 -0.51 0.54 0.65 0.12 0.01 -0.14 0.65 0.00 0.39 -0.15 0.04 -0.26 -0.05 0.11 b -2.6 -2.7 -1.1 -1.7 ns 0.23 ns ns ns 0.07 -2.9 -2.7 -1.3 -1.8 -0.30 0.19 ns ns ns 0.12 ns ns ns ns ns ns ns SAT_LU_1 apparent as in the case of leaf unfolding. The Climatic Water Balance of the period from July to September and also of the period from April to September had significant effect on visually observed beginning of leaf colouring only in the stand Zelezná Studienka ­ the lower the negative values of CWB, the earlier onset of leaf colouring (Fig. 7a). Also the Actual Evapotranspiration of the growing season (April­ September) was a significant factor on the VIS_LC_10 only in stand Zelezná Studienka ­ the lower the negative values of Ea, the earlier onset of leaf colouring (Fig. 7a). The higher Average Air Temperatures of August­September evoked the earlier onset of leaf colouring in the stand Zelezná Studienka (Fig. 7b). In this stand also the higher values of Dryness Index of the summer (July­September) affected the onset of VIS_LC_10 (Table 4). The low negative values of CWB and Ea together with high AAT of the months before onset of VIS_LC_10 caused the earlier onset of this phenophase only in the stand Zelezná Studienka at 304 m a.s.l. in the warm climatic area. In the other stands (in moderately warm and cold climatic areas) no statistically significant relationship with biometeorological variables and VIS_LC_10 was discovered. 4. Discussion Temperature effect on phenology in the spring period Some authors reported that the overcoming of the winter dormancy of European beech is effected by the previous freezing period, and that the effect of the increasing temperature and photoperiod is minimal (Falusi & Calamassi 1990; Kramer 1994; Caffarra & Donnelly 2011). However most authors consider just increasing temperature to be a driving force of the onset of spring phenophases (Braslavská 2000; Rötzer & Chmielewski 2001; Badeck et al. 2004; Menzel et al. 2006; Pálesová 2012 etc.). According to the previous studies, the onset day of spring phenophases depends most significantly on the temperature of the previous 2­3 months (Piao et al. 2006). The analyses of phenological and meteorological data from the International Phenological Gardens revealed that the beginning of the growing season was affected mainly by the average air temperature of March, 45 Autumn period VIS_LC_10 AATVIII­IX AATIX CWBIV­IX CWBVII­IX Ea, IV­IX Ea, VII­IX DIVII­IX AAT ­ average air temperature, nCD ­ number of chilling days, nFD ­ number of frozen days, CWB ­ climatic water balance, Ea ­ actual evapotranspiration, DI ­ dryness index, P ­ precipitation total , n ­ number of analyzed years, ryx ­ correlation coefficient, b ­ regression coefficient, significant values are highlighted by bolt, ns ­ non significance. Autumn period The relationship between investigated biometeorological variables and the first autumn phenophase was not as T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 6. Relationships between visually observed and MODIS-derived onset days of the beginning of leaf unfolding (LU_10) and average air temperatures calculated for the periods February­April (AAT_II­IV), March­April (AAT_III­IV), February (AAT_II), and April (AAT_IV). 46 T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 Fig. 7. Relationship between investigated biometeorological variables and beginning of leaf colouring (VIS_LC_10) in the stand Zelezná Studienka: a) CWB_IV­IX ­ Climatic Water Balance during the period April­September; CWB_VII­IX ­ Climatic Water Balance during the period July­September; EA_IV­IX ­ Actual Evapotranspiration during the period April­September; b) AAT_VIII­IX ­ Average Air Temperature during the period August­September. but also of February and partially of April ­ the higher the temperature in late winter, the earlier onset of leaf unfolding (Chmielewski & Rötzer 2001). Menzel et al. (2006) found a clear reaction of the plant phenological phases to the increasing temperature. Most of the phenophases correlated with the average temperature of the month when the phenophase began and of the previous month. The trend in leaf unfolding of European beech agreed with the increasing temperature in March (r = -0.86). Our results confirmed that the beginning of leaf unfolding of European beech depended on the air temperature of the previous 2­3 months. We revealed the most significant relationship between average air temperature of the period February­April and the beginning of leaf unfolding in all three stands. But the other temperature based biometeorological variables had a significant effect. Previous studies from Slovakia revealed a significant relationship between Average Air Temperature of the period March­April (PTVIII­IV) and LU_10: i) Barna et al. (2009) ­ r = -0.61, ii) Schieber et al. (2009) ­ r = -0.86, iii) Braslavská (2000) on the localities below 500 m a.s.l. ­ r = -0.83, and on the localities above 500 m a.s.l. ­ r = -0.76. In this study similar results were discovered with data from both the satellite and visual phenological methods. In the stand Zelezná Studienka (below 500 m a.s.l.) we revealed ­ r = -0.81 and -0.77, for visual and satellite method respectively. In the stands Zvolen and Poana (above 500 m a.s.l.) we revealed - r = ­0.45 and ­0.63 for Zvolen and ­ r = -0.75 and -0.56 for Poana, for visual and satellite method respectively (Table 4). Higher temperatures of the months before the onset of spring phenological phases caused the shift to the earlier onset. The bud bursting of European beech advanced by 3.2 days·°C-1 (Rötzer & Chmielewski 2001) to 3.6 days·°C-1 (Kramer 1996). On the contrary Menzel et al. (2001) pointed out a weaker sensitivity of beech to increasing winter and spring temperatures. Also Vitasse et al. (2009a) reported, that beech showed weaker sensitivity to higher spring temperatures with the advance of bud bursting by 2.1 days·°C-1 and advance of leaf unfolding by 1.9 days·°C-1. Our analyses revealed the advance of the beginning of leaf unfolding by 2.1­2.9 days·°C-1 of average temperature of the period February­April, 2.3­3.1 days·°C-1 of average temperature of the period March­April, 0.8­1.3 days·°C-1 of average air temperature of February and 1.2­1.8 days·°C-1 of average air temperature of April for particular stands and methods (Table 4). A much bigger advance of the beginning of leaf unfolding by 2.5 days·°C-1 of average temperature of April found by Pálesová (2012). Besides average temperature, beginning of leaf unfolding significantly depended on the number of chilling and frozen days as well. The increasing number of chilling days delayed the onset of VIS_LU_10 and SAT_LU_1 by 1.5­ 2.3 days·10 days-1 and increasing number of frozen days led to delay of VIS_LU_10 and SAT_LU_1 by 1.1­3 days 10 days-1 (Table 4) for a particular stand. Pálesová (2012) also found the delay of bud swelling with increasing number of chilling and frozen days. The significant impact of the temperature also revealed the evaluation of timing of the earliest and the latest onset days of the beginning of leaf unfolding from deviations in temperature and precipitation from normal on beech stands. The earliest onset of the beginning of leaf unfolding occurred in the years when the temperature of months of the first quarter of the year were considerably above normal and higher than in the other years. On the contrary in the period 2000­2012, the latest onset days of beginning of leaf unfolding occurred in the years with the coldest first quarter. Temperature effect on phenology in autumn period Most of the studies focused on the onset of autumn phenophases (Chmielewski & Rötzer 2001; Menzel 2002; Sparks & Menzel 2002) showed a weak relationship between tem47 T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 perature and the onset. Our study revealed dual results. The average air temperature of the period August­September had a significant effect on the onset of leaf colouring, but only in the stand Zelezná Studienka where the VIS_LC_10 advanced by 1.3 days·°C-1. However in the other stands the effect of temperature was not significant (Table 4). Estrella & Menzel (2006) and Cufar et al. (2012) found the opposite relationship between the air temperature and leaf colouring (r = 0.56), when the increasing temperature delayed the onset of this phenophase. The reason may be the location of our stand Zelezná Studienka in a warm climatic area, where the high temperatures in summer could have a negative effect on the condition of trees and leaves. Vitasse et al. (2009b) considered just average temperature of the period August­ November to be the main factor affecting the general leaf colouring of beech forests in France. Climatic water balance effect on phenology The importance of temperature and day length in determining the end of the growing season varies among plant species with certain groups solely dependent on photoperiod (Howe et al. 1996). This dependence on the length of the photoperiod has a substantially positive impact on forests in the current global warming conditions: while temperatures may increase over time, day length remains unchanged, limiting the potential of woody plants controlled mainly by the length of the photoperiod to extend their growing season (Way 2011). In this study, we did not test the effect of the length of photoperiod on the onset of phenophases, but we used this parameter as an input for the calculation of potential evapotranspiration according to Tornthwait and Mater (1955). Skvarenina et al. (2004) considered the climatic water balance to be the appropriate variable for characterizing the climatic conditions of the area. The input values for calculation were just potential evapotranspiration and precipitation (Baumgartner & Liebscher 1990). Following this assumption, we tested the effect of climatic water balance, actual evapotranspiration, dryness index and precipitation on the onset of spring and autumn phenophases. ­ The study stands Zelezná Studienka and Zvolen were located in the 3rd vegetation belt (oak-beech). According to Walter's climadiagram, the precipitation prevails over the evapotranspiration. However, the climatic water balance climadiagram pointed out that in the period March­August potential evapotranspiration prevails over the precipitation and the lack of water is less than 100 mm. In the stand Zelezná Studienka, the CWB of the period April­September was on average ­226 mm, and CWB of the period July­September was ­118 mm (Table 3). The effect of these CWBs on the onset of leaf colouring was significant (Table 4). CWB of the period January­April was not significant; the CWB values were positive (on average 106 mm). This was caused by the low level of potential evapotranspiration before the growing season. The Actual Evapotranspiration in this period was also lower ­ on average 166 mm ­ than in the summer season (July­September), when it was on average 220 mm (Table 3). The Dryness Index for the summer season was also significant in this stand (Table 4). In the years 2004 and 2012 it exceeded the value 2.3, which indicated a semi-arid climate regime and in 2003 DI (3.8) indicated even an arid regime. However, the average DI (1.9, Table 3) of the period 2000­2012 characterized the a semi-humid regime. In the Zvolen stand, the effect of CWB, Ea, DI was not significant (Table 4), although this stand is located in the same belt as Zelezná Studienka. The CWB of the period April­September was on average ­138 mm, and CWB of the period July­ September was ­88 mm (Table 3). These values are similar to that published in Skvarenina et al. (2002). The dryness index of the summer season was in comparison with Zelezná Studienka smaller, with maximal value 2.2 in the year 2007. The average DI in the period 2000­2012 was 1.5 (Table 3). All these values classified the climate regime in the Zvolen stand as a semi-humid. ­ The Poana stand is located in the 5th vegetation belt (firbeech). The water balance in the general growing season is positive. The precipitation totals in the summer exceed the sum of 100 mm. The effect of CWB, Ea, DI on the beginning of leaf unfolding and leaf colouring was not significant. The maximal dryness index was 2 in the year 2009. The average summer DI in the period 2000­2012 was 1.2 (semi-humid) with 6 humid summers in this period. The precipitation totals of summer really exceeded 100 mm published for this vegetation belt by Skvarenina et al. (2002). However, the climatic water balance still had negative values in both analyzed periods (April­September and July­September) (Table 3). The timing of the onset of spring and autumn phenophases reflect the sensitivity of European beech to the amount of available water in the soil profile (Stelcová et al. 2008; Nielsen & Jorgensen 2003). Bagar et al. (2001) found out that increasing sums of effective air temperatures together with lower precipitation totals caused earlier termination of assimilation and thus earlier onset of the leaf colouring. We discovered a similar pattern in the stand Zelezná Studienka, where the high average air temperatures of the period August­September advanced the phenophase VIS_LC_10. Schieber et al. (2009) detected the dependence between cumulative precipitation total of the period May­ August and the leaf colouring in beech forests (r = 0.58). In this study, we tested this biometeorological variable only in relation to beginning of leaf unfolding and found a significant dependence between the precipitation total of April and VIS_LU_10 and SAT_LU_1. Spring rainfalls are usually accompanied by the drop of temperature. This could be the reason why the higher precipitation totals delayed the onset of the beginning of leaf unfolding. According to findings of CWB published in Minás et al. (2011), deterioration of conditions could be expected for beech up to the 4th vegetation belt and improvement of conditions for beech from the 5th to the 7th vegetation belts. We found a statistically significant effect of CWB only in the stand Zelezná Studienka (3rd veg. belt), where the low negative values of CWB caused earlier onset of leaf colouring. The significant impact of the high temperature and low amount of available water was also visible from the evaluaount unt tion of the earliest and latest onset days of the beginning of leaf colouring from deviations in temperature and precipitation from normal. This corresponded with results of Kramer (1995). He presented that higher average temperatures in the period of summer-autumn together with the previous dryness period forced the onset of autumn phenophases. T. Hlásny et al. / Lesn. Cas. For. J. 60 (2014) 5­18 5. Conclusion It was found that air temperature is one of the most significant elements affecting the onset of spring phenophases in the tested area of beech stands. In this study, we calculated biometeorological elements of the beginning of leaf unfolding based on the minimum, maximum and average daily air temperatures: Average Air Temperature of the period February­April (AATIIIV), Average Air Temperature of the period March­April (AATIII­IV), Average Air Temperature of February (AATII), Average Air Temperature of April (AATIV), Number of chill and frozen days (nCD, nFD). We also tested the effect of evapotranspiration-based biometeorological variables: Climatic Water Balance of the period January­ April (CWBIV­IX), Actual Evapotranspiration (Ea), Dryness Index (DI) and precipitation total of April (PIV). The average air temperature of the period from February to April was the most significant factor which evoked the onset of beginning of leaf unfolding in all three stands. Compared with visual phenological observations, for satellite observation more significant correlations with biometeorological variables were found. This could result from the subjective error of an observer when he evaluates the phenological phases visually. This error is eliminated by the use of the digital image analyses of the satellite data. The dependence of the onset of autumn phenophases on the external biometeorological variables has been poorly studied and characterized. From the variables which may affect the onset of autumn phenophases we had chosen: Climatic Water Balance of the growing period (April­ September) (CWBIV­IX) and of the period July­September (CWBVII­IX), Actual Evapotranspiration (Ea), Dryness Index (DI), Average Air Temperature of the period August­September (AATVIII­IX) and Average Air Temperature of September (AATIX). The effect of these biometeorological variables on the beginning of leaf colouring was significant only in the stand Zelezná Studienka. The analysis of the relationship between biometeorological variables and onset of the phenophases of the period 2000­2012 pointed to the inappropriate climate conditions for beech in the 3rd vegetation belt in warm climate area. As suspected Skvarenina et al. (2004), here the unfavorable cli, matic conditions ­ high temperatures in August and September and low negative values of climatic water balance in the summer season ­ caused an earlier onset of leaf colouring and shortening of the growing season. On the contrary, the high temperatures in the early spring caused the advance of beginning of leaf unfolding and thereby increased the risk of damage by the late frosts. Therefore, the cultivation of beech in the 3rd vegetation belt requires increasing attention paid to the climate conditions. This kind of research is necessary to achieve further noticeable progress in phenological modelling by using the expanding area of remote sensing observation.

Journal

Forestry Journal – de Gruyter

Published: Mar 1, 2014

There are no references for this article.