1 Remote Sensing Laboratories, Department of Geography, University of Zurich, Zurich, Switzerland, 2 Institute for Applied
Remote Sensing, EURAC, Bolzano, Italy, 3 Central Institute for Meteorology and Geodynamics, Vienna, Austria
Abstract Snowcover impacts alpinelandsurfacephenology in various ways, but our knowledge about the effect of snowcover on alpinelandsurfacephenology is still limited. We studied this relationship in the European Alps using satellite-derived metrics of snowcoverphenology (SCP), namely, ﬁrst snow fall, last snow day, and snowcover duration (SCD), in combination with landsurfacephenology (LSP), namely, start of season (SOS), end of season, and length of season (LOS) for the period of 2003–2014. We tested the dependency of interannual differences (Δ) of SCP and LSP metrics with altitude (up to 3000 m above sea level) for seven natural vegetation types, four main climatic subregions, and four terrain expositions. We found that 25.3% of all pixels showed signiﬁcant (p < 0.05) correlation between ΔSCD and ΔSOS and 15.3% between ΔSCD and ΔLOS across the entire study area. Correlations between ΔSCD and ΔSOS as well as ΔSCD and ΔLOS are more pronounced in the northern subregions of the Alps, at high altitudes, and on north and west facing terrain—or more generally, in regions with longer SCD. We conclude that snowcover has a greater effect on alpinephenology at higher than at lower altitudes, which may be attributed to the coupled inﬂuence of snowcover with underground conditions and air temperature. Alpine ecosystems may therefore be particularly sensitive to future change of snowcover at high altitudes under climate warming scenarios.
In this respect, we found ﬁrst SOS to be inﬂuenced primarily by SCD and second by SWE m across elevations (Figure 5a). In general, both the snowmelt timing and snow depth have important effects on plant phenology and growth, but the snowmelt timing has stronger implications than snow depth (Wipf et al., 2009). Shorter duration of snowcover is mainly caused by earlier melt of snowcover in spring (Laternser & Schneebeli, 2003). Moreover, the timing and growth of phenological events is highly correlated to the date of snowmelt (Julitta et al., 2014; Rammig et al., 2010; Steltzer et al., 2009). A delayed snowmelt can compress the length of growing seasons and thus may decrease vegetation productivity (Morgner et al., 2010; Wipf & Rixen, 2010). Thus, these arguments support our ﬁnding that SCD has a stronger inﬂuence and a higher relative weight effect on alpinephenology compared to SWE m . In addition, the snow melt date and snow depth are often strictly linked (Hejcman et al., 2006). Together with springtime temperatures, the depth of the winter snow- pack determines the timing of snowmelt (Richardson et al., 2013). Therefore, the timing and accumulation of snow may have synergistic effects with spring temperature in the determination of the alpinephenology. Second, we found LOS be equally inﬂuenced by SCD and SWE m across elevations (Figure 5b). In alpine eco- systems, snow may inﬂuence LOS through its effect on vegetation greenness (Trujillo et al., 2012). A deeper snowpack raises winter soil temperatures and may increase soil moisture and nutrient availability and lead to higher rates of litter decomposition (Chen et al., 2005; Wahren et al., 2005). In contrast, thin and early melting snow may result in plants being exposed to cold air temperatures that cause frost damage or inhi- bit rates of development (Wipf et al., 2006). For instance, Hiller et al. (2005) reported that low temperature, saturation of soil with water during snowmelt, and occasional drought may hamper plant activity during the growing season in the alpine tundra. In contrast, deep and late melted snow provides frost protection (Desai et al., 2016; Hu et al., 2010) for the plants until air temperatures are suitable for growth (Richardson et al., 2013). In addition, abundant snow will also result in increased N mineralization, N 2 O ﬂux and net nitriﬁcation (Williams et al., 1998). These aspects may demonstrate the importance of both SCD and SWE m for LOS.
and thermal time to peak (TTP), in the subsequent growing season. Then we evaluated the role of terrain features in shaping the relationships between snowcover and pasture phenology using exact multinomial tests for equivalence. Results revealed a positive relationship between snow covered dates (SCD) and PH occurred in over 1,664 km 2 at p <0.01 and 5,793 km 2 at p<0.05, which account for more than 8% of 68,881 km 2 of pasturelands analyzed in Kyrgyzstan. Also, more negative than positive correlations were found between snowcover onset and PH, and more positive correlations were observed between snowmelt timing and PH. Thus, a longer snow season can positively influence PH. Significant negative correlations between TTP and SCD appeared in 1,840 km 2 at p<0.01 and 6,208 km 2 at p <0.05, and a comparable but smaller area showed negative correlations between TTP and snowmelt date (1,538 km 2 at p<0.01 and 5,188 km 2 at p <0.05). Furthermore, terrain had a stronger influence on the timing of snowmelt than on the number of snow covered dates, with slope being more important than aspect, and the strongest effect appearing from the interaction of aspect with steeper slopes. In this study, we characterized the snow-phenology interactions in highland pastures and revealed strong dependencies of pasture phenology on timing of snowmelt and snow onset and snowcover duration and. Under changing climatic conditions toward earlier spring warming, decreased duration of snowcover may lead to lower pasture productivity threatening the sustainability of montane agropastoralism.
4 Dipartimento Interateneo di Fisica “M. Merlin”, Università degli Studi di Bari e Politecnico di Bari, 70126 Bari, Italy; email@example.com
* Correspondence: firstname.lastname@example.org; Tel.: +49-8153281230
Received: 1 September 2018; Accepted: 4 November 2018; Published: 7 November 2018 Abstract: Alpine ecosystems are particularly sensitive to climate change, and therefore it is of significant interest to understand the relationships between phenology and its seasonal drivers in mountain areas. However, no alpine-wide assessment on the relationship between landsurfacephenology (LSP) patterns and its climatic drivers including snow exists. Here, an assessment of the influence of snowcover variations on vegetation phenology is presented, which is based on a 17-year time-series of MODIS data. From this data snowcover duration (SCD) and phenology metrics based on the Normalized Difference Vegetation Index (NDVI) have been extracted at 250 m resolution for the entire European Alps. The combined influence of additional climate drivers on phenology are shown on a regional scale for the Italian province of South Tyrol using reanalyzed climate data. The relationship between vegetation and snow metrics strongly depended on altitude. Temporal trends towards an earlier onset of vegetation growth, increasing monthly mean NDVI in spring and late summer, as well as shorter SCD were observed, but they were mostly non-significant and the magnitude of these tendencies differed by altitude. Significant negative correlations between monthly mean NDVI and SCD were observed for 15–55% of all vegetated pixels, especially from December to April and in altitudes from 1000–2000 m. On the regional scale of South Tyrol, the seasonality of NDVI and SCD achieved the highest share of correlating pixels above 1500 m, while at lower elevations mean temperature correlated best. Examining the combined effect of climate variables, for average altitude and exposition, SCD had the highest effect on NDVI, followed by mean temperature and radiation. The presented analysis allows to assess the spatiotemporal patterns of earth-observation based snow and vegetation metrics over the Alps, as well as to understand the relative importance of snow as phenological driver with respect to other climate variables.
The combination of environmental monitoring and modeling plays an important role when investigating current and fu- ture climate and their control of diverse phenomena of the cryosphere. Measurements are widely used for model val- idation and calibration. However, the problem of compar- ing model simulations made at one scale to measurements taken at another scale has no simple solution. The rele- vance of this issue increases when modeling phenomena such as snowcover or permafrost in highly variable terrain such as the Swiss Alps, since variations occur at smaller scales than in more homogeneous terrain. The difficulties that arise from scaling issues can be large: in contrast to measure- ments, spatially-distributed models are often grid-based and represent areas of several square meters to square kilome- ters. Since the physical processes that influence the pattern of variation of a phenomena operate and interact at different spatial scales, spatial variation can simultaneously occur on scales of different orders of magnitude (Oliver and Webster, 1986). Therefore, the extrapolation of results (including cali- brated model outputs) based on point measurements requires caution, especially in highly variable terrain (Nelson et al., 1998). A specific statement concerning this issue was made by Gupta et al. (2005):
For non-forested vegetated land covers, we found strong relationships between air temperature and albedo (all bands), which suggests that tgrowing season phenology was sufficiently synchronized with monthly mean air temperatures. With the exception of croplands (“CRO”) and pastures (“PAS”), SW albedo (black-sky) decreased with increasing air temperature, suggesting that it was driven by increased vegetation masking of a higher albedo surface during the growing season. However, for cropland, pastures, and all forest endmembers, the SW albedo increased with increasing air temperature, suggesting either a larger role played by understory vegetation or by increased canopy masking of a lower albedo surface. During the snow season, we found that air temperature and surface albedo were negatively correlated. This is a relationship that is presumably driven by the influence of air temperature on snow metamorphosis and snow physical state [ 46 , 87 ]. When applied outside the training region, we found absolute normalized median errors to be ≤ 10% for most non-forest endmember models (SW black-sky). The non-forest endmember models on average performed best during spring (MAM) and summer (JJA) where the proportion of the total number of predictions agreeing within 10% of the MCD43A3 retrievals was around 35% and 45%, respectively (Figure S7, Supporting Information). These shares were weighed downward by the persistent positive error for forested peatbogs (“PB-f”) and persistent negative error for non-vegetated open areas (“O-nv”) whose proportions of the total number of predictions agreeing within 10% of the MCD43A3 retrievals were only 20% and 25% within the validation region, respectively (Figure S7). In general, these two landcover (endmember) types exhibit the largest variation in both geological attributes (exposed mineral composition) and vegetation attributes (vegetation cover fraction). These are two physical attributes important for the surface albedo which were not captured by the models. Furthermore, the model training region contained a disproportionately low share of “O-nv” relative to the validation region (Figure S1), and given the large variation in surface attributes within “O-nv”, a larger “O-nv” sample during model fitting would likely have resulted in an improved performance by
the Alpine region. The majority of the drainage areas corre- spond to elevations where snow dynamics are fast and sev- eral snow accumulation–melt cycles may happen over the year. This makes snow mapping under clouds a challenge. A spatio-temporal combination of MODIS images for cloud reduction was tested by Parajka and Blöschl (2008) over the whole of Austria, a territory similar to our case. Testing some backward filters (2, 5 and 7 days), they obtained a minimum accordance with ground observation of about 92 % in relation to a 7-day temporal filter. The latter was applied after merg- ing Terra and Aqua images. However, temporal filters leave a percentage of cloud cover depending on the allowed tem- poral window from which the procedure can derive informa- tion to add to the cloudy day. Moreover, the accuracy of this method was dependent on the season (Parajka and Blöschl, 2008). During the winter snow processes in mountains are slower and in midsummer snow maps tend to a steady layout with snow over glaciers and snowfields. In spring and autumn ground conditions change more quickly, due to snowfalls and snowmelt which determine several land/snow micro-cycles on the transition altitudes. On the other hand, spatial filters exist which use information of the same daily images to map snow in areas hidden by clouds. Some of them estimate landcover pixel by pixel, filling the gaps with data found in the nearest neighbours (Parajka and Blöschl, 2008; Gafurov and Bárdossy, 2009; Tong et al., 2009). Others are based on the concept of snow transition elevation, looking for the altitude above which all pixels within a region, a basin or a slope can be supposed to be snow covered (Parajka et al., 2010). Paudel and Andersen (2011) pointed out that snow processes are highly affected by local conditions such as slope, aspect and land use, and the limit of spatial filters falls with the adoption of data recorded in other areas for estimating landcover of each individual cell.
The Arctic net ecosystem exchange (NEE) of CO 2 between the landsurface and the
atmosphere is influenced by the timing of snow onset and melt. The objective of this study was to examine whether uncertainty in model estimates of NEE could be reduced by representing the influence of snow on NEE using remote sensing observations of snowcover area (SCA). Observations of NEE and time-lapse images of SCA were collected over four locations at a low Arctic site (Daring Lake, NWT) in May–June 2010. Analysis of these observations indicated that SCA influences NEE, and that good agreement exists between SCA derived from time-lapse images, Landsat and MODIS. MODIS SCA was therefore incorporated into the vegetation photosynthesis respiration model (VPRM). VPRM was calibrated using observations collected in 2005 at Daring Lake. Estimates of NEE were then generated over Daring Lake and Ivotuk, Alaska (2004–2007) using VPRM formulations with and without explicit representations of the influence of SCA on respiration and/or photosynthesis. Model performance was assessed by comparing VPRM output against unfilled eddy covariance observations from Daring Lake and Ivotuk (2004–2007). The uncertainty in VPRM estimates of NEE was reduced when respiration was estimated as a function of air temperature when SCA ≤ 50% and as a function of soil temperature when SCA > 50%.
urbanized and urbanizing landscapes have resulted significant impacts on the local and global ecosystems. Among tremendous environmental issues associated with human activities, the LST effect has been one of the increasing concentrations of urban problems. The distribution of thermal phenomenon was commonly referred to as temperature differences between different land use types, with higher air temperatures in densely built cities and lower temperatures in surrounding rural regions.
Urban heat island (UHI) is considered as one of the main causes of urban micro-climate warming . It is defined as an environmental phenomenon where air and landsurface temperatures (LST) of urban areas are higher than those of its surrounding areas . UHI is associated with a number of local problems such as biophysical hazards (e.g., heat stress), air pollution and associated public health problems. As a result, the development of strategies to mitigate UHI is now a key policy challenge in order to reduce urban micro-climate warming and to enhance local livability, public health, and well-being . This research focuses on the LST component of the UHI effect, and estimates LST in a simulated environment. This is due to the fact that UHI is related to the spatial distribution of LST . As a result, LST and UHI are often used interchangeably in this paper. Being an important contributor to the UHI effect, it is, therefore, critical to obtain LST as a first and key step to the UHI analysis; and then to simulate future LST so that policies can be undertaken to reduce the UHI effect.
To improve the classification results, a two-stage classification strategy was implemented: (1) a majority filter was applied to remove misclassified pixels; and (2) more accurate water and sandbar data (acquired by other methods) was integrated into the filtered image. In particular, water areas were determined by using the normalized difference water index (NDWI; Xu, 2006) whereas sandbars were manually fixed. The accuracy of each classification was assessed by uploading 350 points taken from each classified image to Google Earth Pro to compare their similarity. The "view historical imagery" tool in Google Earth Pro was used to find the best possible referenced image for each year. Based on this, the overall accuracy (the percentage of correctly classified pixels out of all pixels sampled for all classes), producer’s accuracy (the percentage of a particular LULC type on the ground is correctly classified in the map), user’s accuracy (the percentage of a class on the map that matches the corresponding class on the ground) and kappa index (the degree of matching between reference data set and classification) were calculated to evaluate the accuracy of the classification. The classified images were then compiled by using the overlay tool in ArcGIS to assess land use change from 2003 to 2015.
Nowadays, more than 40% of the population lives in Chinese cities. The rapid urbanization process brought about many eco- environmental problems, such as the drastic change of land use and development of urban heat island. Three Landsat TM and ETM+ images data of Beijing acquired on April 9, 1995 and April 30, 2000 were selected to this research. The landsurface temperature (LST) and land use and landcover (LULC) classes were retrieved and extracted. The temperature-vegetation index (TVX) space was constructed to investigate the influence of land changes over LST. The result showed that the land use change was an important driver for LST increase, the temporal trajectory of pixels in the TVX space migrated from the dense-vegetation- low temperature condition to the sparse vegetation-high temperature condition.
The lower boundary condition for momentum transfer is complex due to creep and dependent on instantaneous wind speed and turbulent motions near the surface. As a result, equilibrium conditions were never found in the field obser- vations reported here. Nonequilibrium saltation–wind inter- actions cannot be described with simple uniform trajecto- ries. The majority of particle trajectories in saltation con- sist of short hop lengths and times, resulting in high fre- quencies of particle collisions that break surface bond struc- tures and create dense quasi-fluidized bed characteristics. The complexity of conservation of mass, momentum, and ki- netic energy in blowing snow in natural environments, such as measured here, cannot be understated, especially when the large rimed, aggregate tumblons were present. In the alpine snowpacks investigated, variable particle restitution coeffi- cients contributed to this complexity. While high HHI wind- hardened surfaces exhibited similar behavior of slip veloc- ity and particle velocity gradients as rigid bed sand studies, complexities over a natural snowpack prevented conclusive bimodal “erodible” vs. “nonerodible” scale relations that ap- pear to be viable for sand (Ho et al., 2011). Blowing snow is a distinctive two-phase flow.
tors”). Spring models were more accurate than autumn, with median relative error values of 10 to 27 % (12 to 1 predic- tor), versus 26 to 60 % of autumn (14 to 1 predictor). Fig- ure 4 shows the pseudo-R 2 of the models as well as the rel- ative importance of each predictor. Spring models explained a percentage of the variance up to 81 % (Fig. 4a), whereas autumn explained up to 61 % (Fig. 4b). Cook et al. (2005), using a modelled based on GDD only, explained 63 % on the variance of onset date for mixed and boreal forest. Figure 5 shows the relative error in the prediction of different models after removing the least important predictor. Regarding the relative importance of the drivers, the same ranking in im- portance was observed within the different models of each phenophase, which reflected the stability in the RF impor- tance estimation, and a high reliability of the results (Fig. 4). To interpret the main weather drivers of the interannual vari- ation in phenology, simplified models with reduced num- ber of predictors were selected for spring and autumn (see Sect. 3.5), respectively. The spring model was composed of six predictors (pseudo-R 2 = 0.77 and median relative error of 10 %) and the autumn model of five predictors (pseudo- R 2 = 0.59 and median relative error of 28 %) (Fig. 6). Our re- sults suggest that interannual variation in the onset on green- ness (LSP) of temperate forest species is driven mainly by the daily temperature of the 30 days prior to onset (but not nec- essarily the GDD), with the most important driver being the minimum temperature. Photoperiod was also important; the most accurate empirical prediction was obtained by a com- bined temperature–radiation forcing, integrating the SIS of the previous 90 days. For senescence, temperature was sug- gested to be more important than photoperiod in controlling the senescence process (Archetti et al., 2013; Jeong and Med- vigy, 2014; Vitasse et al., 2009; Yang et al., 2012), with the most important drivers being the date of the first freeze and the accumulation of chilling temperatures. However, we did not observe a legacy effect of a much earlier or later spring onset on the date of senescence. Autumn models that in- cluded the interannual variation (z score values) in the onset of greenness did not outperform the remaining models (see Tables S2 and S3) and the relative importance was low in comparison with other drivers.
Fig. 1 : Normalized squared covariance (NSC, contours, in %) for the first MCA mode between observed SLP and Eurasian snowcover, for each month in the atmosphere. The lag is positive when the snowcover leads SLP. The gray shading provides the level of statistical significance for NSC. The plus symbols indicate the atmospheric month and time lag where the level of significance for the correlation (R) is below 5%.
The objectives of this study were (1) to identify fine- scale phenological patterns of brittlebush by empirically testing the effects of water availability and landcover type on the flowering phenology of brittlebush ( Encelia farinosa ) and (2) to investigate arthropod pollinator community associated with brittlebush in different landcover types and over time in the Phoenix metropolitan region, Arizona, United States. Brittlebush was chosen because it is a native Sonoran desert shrub, is commonly used in landscaping across the Phoenix metropolitan area, and appears to vary in flowering phenology across the city. Because previous studies have indicated that brittlebush is a winter-spring blooming plant that re- sponds to both water availability and temperature as cues for flowering (Bowers and Dimmitt 1994), we hy- pothesized that both water availability and landcover types (which strongly affect temperature) would affect the timing, length, and percent of plants in flower. Add- itionally, we hypothesized that pollinator abundance and richness would vary by landcover type and over time. Whereas general bee abundance and richness has been studied in cities worldwide, we wanted to understand how the entire pollinator community is synchronized with the flowering of brittlebush in different landcover types.
Acknowledgements. This study was funded through the nano- tera.ch project X-Sense. Model experiments were supported by the AAA/SWITCH funded Swiss Multi Science Computing Grid project (http://www.smscg.ch) with computational infrastructure and support. Customised libraries (gGEOtop and GC3Pie) and user support were kindly provided by GC3: Grid Computing Competence Centre (http://www.gc3.uzh.ch). The International Foundation High Altitude Research Stations Jungfraujoch and Gornergrat supported field work for rock temperatures near Jungfraujoch. We acknowledge MeteoSwiss for providing driv- ing climate time series at Corvatsch and Davos. Many people participated directly or indirectly in many development phases of GEOtop during the last decade. The keywords method actually used in GEOtop I/O had a first realisation with work of Emanuele Cordano; Glen Liston gave the code of his MicroMet model in FORTRAN from which derives the actual improved code used in GEOtop. Thomas Haiden offered the C code for estimating direct solar radiation and shadows, which was subsequently further tested, modified and embedded in GEOtop. Matteo Dall’Amico thanks the Monalisa project financed by the Autonomous Province of Bolzano that supported his work on GEOtop. Riccardo Rigon thanks the HydroAlp project of the Autonomous Province of Bolzano that supported his work. The authors thank Stefano Cozzini (Exact lab, Trieste, Italy), Gianfranco Gallizia and Angelo Leto for their help as computer scientists.
The advent of the Landsat series of sensors beginning with the launch of
Landsat-1 in 1972 heralded a new era of producing multispectral satellite images from which snow maps could be created at 80-m spatial resolution (Rango and Martinec, 1979). With Landsat data came the ability to create detailed basin- scale snow-cover maps when cloudcover permitted. Landsats-4 and -5 carried a Thematic Mapper (TM) sensor with 30-m resolution, and Landsat-7 carries an Enhanced Thematic Mapper Plus (ETM+) with spatial resolution of 30 m and the panchromatic band with a resolution of 15 m. Though the Landsat series has provided high-quality, scene-based snow maps, the 16- or 18-day repeat-pass interval of the Landsat satellites is not adequate for most snow-mapping
differ in their amounts of biomass and sequestered carbon (Vourlitis and da Rocha 2010), which renders them a useful proxy for carbon reporting in the context of REDD+ (MMA 2017). The Cerrado provides a range of ecosystem services (Lima et al. 2017), which are defined as “benefits people obtain from ecosystems” (MEA 2005). Due to the large extent of the Cerrado, ecosystem services and functions, such as climate regulation, carbon sequestration, or the provision of habitats, are of national and even international importance. Despite its ecological and societal value, the Cerrado has a very weak conservation status with only 2.2 % of the biome’s extent being considered in Brazils protected area network. Caused by its underrepresentation in national policies and conservation strategies, especially in comparison to the Amazon biome (Klink and Machado 2005). In contrast, as a response to a growing demand for agricultural products, governmental programs have aimed at the economic development of rural areas of Brazil. The ‘creation’ of Brazil’s new capital Brasília in the central Cerrado in 1960, was accompanied by the development of transportation systems and infrastructure, opening the Cerrado for settlement and other types of anthropogenic land use (Klink and Moreira 2002). Incentives for new farming activities were set by low-interest loans and advances in agricultural mechanization, while fertilizers enabled cultivation on the nutrient-poor soils of the Cerrado (Klink and Moreira 2002). The combination of these factors led to large-scale land conversion, with land clearing rates that exceeded those of the Amazon (Klink and Machado 2005) resulting in approximately 60 % of remaining natural vegetation (Sano et al. 2010). A growing international demand for agricultural products is considered a direct driver of land use change in Brazil (Lapola et al. 2013), and will have a direct effect on land use, which needs to be steered by appropriate policies to circumvent unsustainable developments (Strassburg et al. 2017).