Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Spatial variation and seasonal dynamics of leaf-area index in the arctic tundra-implications for linking ground observations and satellite images

Spatial variation and seasonal dynamics of leaf-area index in the arctic tundra-implications for... Vegetation in the arctic tundra typically consists of a small-scale mosaic of plant communities, with species differing in growth forms, seasonality, and biogeochemical properties. Characterization of this variation is essential for understanding and modeling the functioning of the arctic tundra in global carbon cycling, as well as for evaluating the resolution requirements for remote sensing. Our objective was to quantify the seasonal development of the leaf-area index (LAI) and its variation among plant communities in the arctic tundra near Tiksi, coastal Siberia, consisting of graminoid, dwarf shrub, moss, and lichen vegetation. We measured the LAI in the field and used two very-high-spatial resolution multispectral satellite images (QuickBird and WorldView-2), acquired at different phenological stages, to predict landscape-scale patterns. We used the empirical relationships between the plant community-specific LAI and degree-day accumulation (0 °C threshold) and quantified the relationship between the LAI and satellite NDVI (normalized difference vegetation index). Due to the temporal difference between the field data and satellite images, the LAI was approximated for the imagery dates, using the empirical model. LAI explained variation in the NDVI values well (R 0.42–0.92). Of the plant adj. functional types, the graminoid LAI showed the largest seasonal amplitudes and was the main cause of the varying spatial patterns of the NDVI and the related LAI between the two images. Our results illustrate how the short growing season, rapid development of the LAI, yearly climatic variation, and timing of the satellite data should be accounted for in matching imagery and field verification data in the Arctic region. observations of vegetation are equally essential when Introduction vegetation parameters, such as the leaf-area index Vegetation monitoring is a tool for detecting the (LAI), are included in ecosystem models (e.g. Cramer impacts of climate change on the composition and et al 2001, Melton et al 2013). Many key properties of phenology of arctic ecosystems. For example, satellite vegetation can be reasonably well inferred from image series spanning several decades have already spectral reflectance and thus mapped for large areas, revealed the large-scale greening of the arctic, a using remotely sensed data (Laidler and Treitz 2003, consequence of the increased plant growth and spatial Laidler et al 2008, Ustin and Gamon 2010). However, expansion of shrubs and trees (Stow et al 2004, Forbes the temporal and spatial scales of currently available et al 2010, Frost and Epstein 2014). Spatially extensive remote-sensing products may pose a challenge during © 2017 IOP Publishing Ltd Environ. Res. Lett. 12 (2017) 095002 20°0′0″E 50°0′0″E 90°0′0″E 130°0′0″E 160°0′0″E Bare Water Bog Flux tower Graminoid tundra Survey plots Dry fen Wet fen / half water Lichen tundra Shrub tundra Figure 1. Location of Tiksi (left) and the land-cover map of the study area with survey plot locations (right). The graminoid tundra and flood meadow were grouped in the land-cover classification. mapping of spatially heterogeneous landscapes, such small time difference between the image and the as the northern tundra (Virtanen and Ek 2014). ground-truth data can result in wide variations in The pixel size of the commonly used satellite terms of growth stage. imageries, e.g. those obtained from the Landsat and In this study, therefore, our objective was to Moderate Resolution Imaging Spectroradiometer quantify the spatial and seasonal variation in LAI (MODIS) satellites, ranges from tens to hundreds of among the dominant plant communities in the arctic meters and cannot reveal fine-scale heterogeneity in tundra and to evaluate how the seasonality of the vegetation and ecosystem properties (Laidler et al vegetation affects the interpretation of vegetation 2008, Virtanen and Ek 2014, Mora et al 2015, Bratsch structure from satellite images. We measured the et al 2016). This complicates the examination of plant seasonal development and spatial pattern of the LAI in growth responses to warming, which may vary among the coastal arctic tundra near Tiksi, NE Russia, in the neighboring communities (e.g. McManus et al 2012, summer of 2014 and examined how the normalized Bratsch et al 2016). To resolve this problem, images of difference vegetation index (NDVI), derived from the very high spatial resolution (VHSR, 0.32 m pixel reflectance data, varied among the plant communities size) have become available in recent years. Their and between the VHSR multispectral satellite images usage, however, is hampered by the high price, limited of two different growing seasons. To quantify the temporal availability, which is caused by the relatively dependence of the seasonal LAI development on the long revisit periods of high-resolution image sensor weather and for reconstructing the LAI values, we satellites in the same location, and by the frequent developed regression models between the LAI, degree- cloud cover and low solar angle in the Arctic (Hope days (DD) accumulation, and satellite-based NDVI. and Stow 1996, Rees et al 2002, Stow et al 2004, Based on these models, we then evaluated the impacts Westergaard-Nielsen et al 2013). Therefore, the best- that the temporal mismatch between the satellite quality image often does not temporally match the imagery and field data may have on the interpretation ground-truth data. of the NDVI and LAI distributions in the landscape. This temporal mismatch is a source of uncertainty in the end product, because the reflectance is Methods dependent on the amount of biomass, plant species composition, and the water, nutrient, and pigment Study site contents of plant tissues, all of which are affected by the growth stage of the vegetation (e.g. Ustin and The study site is located about 500 m from the coast of the Arctic Ocean near the Hydrometeorological Gamon 2010). Therefore, the spectral responses to the seasonal changes in vegetation properties should be Observatory of Tiksi in NE Russia (71.5936°N, 128.8850°E, figure 1). The climate at Tiksi is arctic, better understood (Garrigues et al 2008, Ustin and Gamon 2010, Rautiainen et al 2011, Westergaard- with very cold and windy winters, short but relatively Nielsen et al 2013). Since the growing season is short warm summers, and short shoulder seasons between and the changes in the LAI and biomass of the these. The mean annual temperature was 12.7 °C vegetation are rapid in the Arctic, it is likely that even a and the mean annual precipitation 323 mm in 60°0′0″N Environ. Res. Lett. 12 (2017) 095002 Sphagnum mosses (Sphagnum L.) and feathermosses a Dwarf shrub (e.g. Pleurozium schreberi (Willd. ex Brid.) Mitt) with Graminoid 1.0 shrubs were abundant in the dry fens, while the moss Salix spp. B. nana cover was sparse, due to aboveground water in the wet 0.8 Herb fens. The bogs showed typical microtopographic 0.6 variation, and their vegetation was characterized by 0.4 the presence of dwarf shrubs, dwarf birch (Betula nana 0.2 L.), Sphagnum, and feathermosses. The vegetation of the flood meadows along the stream and drier 0.0 graminoid tundra was dominated by graminoids Lichens 1.0 (sedges, grasses) and willows (Salix L. spp.). Abundant Other mosses Sphagnum spp. feathermoss coverage on the ground layer and dwarf 0.8 shrubs in the field layer characterized the shrub-moss 0.6 tundra. Lichen tundra patches alternated with stony 0.4 bare-ground surfaces. 0.2 Flow of the study 0.0 Several steps were needed to obtain time series of LAI, to model LAI for each land cover class (LC) for the years of satellite data, and to produce the LAI maps over the study area. The steps were, in brief, as follows, while the details are given in following paragraphs. Land cover class 1. Phenological dynamics of vegetation: Vegetation Figure 2. Mean leaf-area index (LAI) of (a) all vascular plants and (b) all non-vascular plants in the various land cover surveys (% cover and mean height of each plant classes (LC). In both cases, the LAI is divided into plant functional types) in sample of plots seven times functional types, while the LCs are shown in order of decreasing wetness from the left (the number of harvested over the study period in 2014. field plots in each type is given in parentheses). 2. Estimation of LAI on basis of % cover and height: Sample of plots were harvested after vegetation survey during the peak biomass and 1981–2010. Within this reference period, the average LAI of harvested material was measured at PFT growing season (0 °C threshold) lasted from 7 June to level. Data were used to develop regression 26 September, with DD of 668 (Arctic and Antarctic models to predict LAI. Research Institute AARI 2016). Meteorological data from the Hydrometeorological Observatory was used 3. Producing time series of total and vascular plant to calculate the DD for the examined periods in this LAI: A model using degree-day accumulation study. and chilling temperature accumulation as drivers The site represents a typical coastal tundra of was fitted for each land cover class and LAI was Eastern Siberia with alkaline bedrock and high plant modelled for years 2005, 2012, and 2014. species diversity. We focused on an area of approxi- 4. Mapping spatial distribution of LAI using satellite mately 1 km around the micrometeorological station, data: LAI was modeled over the study area using established in 2010 for eddy covariance (EC) relationship between satellite derived NDVI and measurements of the land-atmosphere exchange of LAI in the study plots. water, heat, carbon dioxide (CO ), and methane (CH ) (Uttal et al 2016, figure 1). The terrain around the EC mast was relatively flat; in addition to microtopographic variation there was a gentle slope Field data rising towards the north and a small stream running The field data on the vegetation were collected in the through the area. The vegetation and land-cover types summer of 2014, which had a warmer (871 DD) within the area were classified, using ground-based growing season than the long-term average. The 92 visual judgement during an expedition in 2012. These inventory plots with radius of 2.5 m were placed along land cover classes (LC) were characterized as (1) dry 16 compass points at regular distances of 25, 50, 75, fen, (2) wet fen, (3) bog, (4) lichen tundra, (5) shrub- 100, 150, and 250 m from the EC mast (figure 1). moss tundra, (6) graminoid tundra, (7) flood meadow, Several additional plots were monitored at distances of (8) bare ground, and (9) water. 300, 350, and 400 m to balance the number of plots in The fen and bog were peat-forming environments, the various LCs. In each plot, the vegetation was while the other land cover classes showed no clearly inventoried in four subplots (45 cm  45 cm in area), discernible peat. Sedges (Carex L. spp.) characterized located 2 m from the plot midpoint in four main the vascular plant vegetation in the fens (figure 2). compass directions. Each plot was classified according Wet fen (6) Flood meadow (11) Graminoid tundra (10) Dry fen (16) Bog (21) Shrub tundra (11) Lichen tundra (19) Bare ground (3) Vascular LAI Non-vascular LAI Environ. Res. Lett. 12 (2017) 095002 Table 1. Regression models describing the dependence of the leaf-area index (LAI) of the various plant functional types on their areal cover (C, %) and height (H, cm) in the Tiksi field plots surveyed in 2014. All regressions and parameters were significant at p < 0.05. a 2 b Plant functional type df reg., res. R adj Salix spp. LAI = 0.0126  C 1, 63 0.88 Dwarf shrub LAI = 0.0192  (C þ 0.001) þ 0.0397  H 2, 63 0.93 Betula nana LAI = 0.0132  C 1, 36 0.91 Graminoids LAI = 0.0150  C 1, 85 0.88 Herbs LAI = 0.0098  C þ 0.0046  H 2, 84 0.93 Degrees of freedom for regression and residuals. Adjusted coefficient of determination. to the apriori LC scheme described above (dry fen, substitution, to estimate the LAI for the subplots that wet fen, etc.), and the plot midpoint was georefer- were monitored for seasonal dynamics. enced, using a Global Positioning System (GPS) device and a measuring tape to achieve a location Satellite image acquisition and processing accuracy of 1–3m. To examine the spatial and temporal variation in the The vegetation was surveyed to characterize the vegetation patterns within the study area, we acquired species composition and to quantify the LAI and its two VHSR multispectral satellite images from the seasonal development in the various LCs of the archive of DigitalGlobe (Westminster, CO, USA). To tundra. The vegetation was inventoried as plant enable the comparison of images taken under different functional types (PFTs), following the typification by atmospheric conditions, the images were corrected for Hugelius et al (2011), which is a modification of that atmospheric scattering and transformed into surface by Chapin et al (1996). The PFTs included: (1) reflectance values, using the dark-object subtraction Sphagnum mosses, (2) other mosses, (3) lichen (4) method (Chavez 1988). dwarf shrubs, (5) deciduous shrub Betula nana,(6) The images, QuickBird (QB, DigitalGlobe, 15 July deciduous shrub Salix spp., (7) herbaceous species, 2005) and WorldView-2 (WV2, DigitalGlobe, 12 and (8) graminoids. August 2012), were chosen because they were of good The projection cover percentage of each PFT was quality and showed the best temporal matching with visually estimated, and the mean height was measured the collection of the peak season field verification data in each of the subplots during the main survey, 23–24 (23–24 July 2014) in terms of calendar days. The July 2014. These values were then averaged for the growing season of the QB image was shorter and main 2.5 m radius plot to relate the vegetation and the cooler (10 June–30 September 2005, 646 DD) than the satellite image-based spectral reflectance. The seasonal growing seasons of the WV2 image (22 May–30 development of the vegetation in one of the subplots September 2012, 1071 DD) and the verification data was monitored by seven successive surveys performed (6 June–24 September 2014, 863 DD). Thus, the QB between 2 July and 15 August 2014. Another subplot image captured a period during which the vegetation was harvested immediately after the main survey to was in the fast-growing phase, while the WV2 image quantify the one-sided LAI. The vascular plant captured the peak LAI period (figure 3 (a)(c)). material harvested was scanned, using a Canon MP The mean reflectance values were extracted for Navigator EX scanner (Canon Inc., Tokyo, Japan) and circular plots with a 2.5 m radius. The reflectance data by calculating the green surface area (= LAI) of the were used to calculate the normalized difference scanned images, using GNU Image Manipulation vegetation index, NDVI = (NIR  VIS)/(NIR þ VIS), Program 2 (GIMP 2) software. The LAI of the mosses which describes the absorbance of the red portion of was estimated as a projection coverage, i.e. we visible (VIS) light and the reflectance of near-infrared determined that a 100% cover would represent an (NIR) radiation by green vegetation. Thus, the NDVI LAI of 1 (Riutta et al 2007). This approach under- is an indicator of the quantity and photosynthetic estimates the true multilayered leaf area of mosses, but capacity of green vegetation and has commonly been probably estimates reasonably well their light-captur- used in spatial extrapolations of LAI (Tucker 1979, ing and reflectance properties, due to the lower Laidler and Treitz 2003, Shaver et al 2013). Supervised pigment and nutrient contents in moss tissues land-cover classification was carried out, based on a (Tieszen and Johnson 1968, Moore et al 2006, Street WV2 image (12 August 2012) to visualize the spatial et al 2012). Using the data of the harvested subplots, distribution of vegetation in the area (figure 1). we then calculated for each PFT the relationship between the areal cover, plant height, and LAI and, Temporal modeling of the LAI using these relationships, estimated the LAI for each We examined the factors determining the seasonal subplot at the time of the main survey (table 1). These development of the LAI to estimate it for the specific relationships were also used, assuming space-for-time dates of the two satellite images. Since our field data on 4 Environ. Res. Lett. 12 (2017) 095002 800 2012 2014 b 2005 d f Dry fen Wet fen Bog 1.0 Lichen tundra Shrub tundra Graminoid tundra Flood meadow 0.5 0.0 1.0 0.5 0.0 Aug Jun Jul Aug Jun Jul Aug Jun Jul Month Figure 3. Accumulation of (a)(c) the temperature sum above 0 °C (degree-days, DD), (d)(f) vascular plant leaf-area index (LAI), and (g)(i) total LAI in the dominant plant community types for the years of the field data (2014), QuickBird (QB) image (2005), and WorldView-2 (WV2) image (2012). The vascular plant LAI was modeled based on seasonal accumulated temperature (DD) (table 2), while the non-vascular LAI was assumed to be constant and directly proportional to the areal cover. The vertical dotted lines indicate the timing of the satellite images and of the field data. LAI development originated from a single growing In this equation, t is the time, S is the DD season only, we used long-term (years 2010–2014) accumulated from all daily mean air temperatures over data on the daily maximum ecosystem photosynthesis 0 °C, and S is the DD accumulated during the latter (GP )to define a functional form that describes the part of the growing season (after 15 July) from the max seasonal growth of vegetation from soil thawing to daily mean air temperatures above 0 °C, but below maximum activity and further to senescence. This 10 °C and a, b , b , c , and c are the parameters to be 1 2 1 2 approach is justified by the close relationship between estimated. The equation was fitted to the total vascular the LAI and GP (e.g. Laurila et al 2001, Street et al plant LAI data for each LC (table 2). The fits obtained max 2007). The daily GP was derived from the were favorable for the period covered by field max continuous CO flux data measured with the EC observations in 2014, and the values outside the method at the site. The GP was determined from measurement period (2 July–15 August) were not used max the eddy covariance flux as the night-day difference in in any further analysis. net ecosystem CO exchange (NEE). The daily GP 2 max Data analysis was obtained as the difference of the 7 day running A paired sample Wilcoxon signed-rank test was used mean of the nighttime NEE (photosynthetic photon 2 1 to determine whether the vegetation classes showed flux density < 20 mmol m s ) and 3 day running significant seasonal differences in their NDVI signals; mean of the daytime NEE (PPFD > 600 mmol m i.e. the NDVI (15 July 2005) and NDVI s ). A function including two temperature-depen- QB WV2 (12 August 2012) were paired for each PCT. To dent sigmoid terms operating during different phases further illustrate the seasonal changes in the various of the growing season proved suitable for modeling the plant communities, we calculated the difference in GP cycle (supplement 1 available at stacks.iop.org/ max NDVI between the late-season and early-season image ERL/12/095002/mmedia). It was fitted to the LAI data (NDVI NDVI ). The relationships between the WV2 QB 2 3 LAI (either measured in the 2014 field survey or 1 1 estimated for the actual dates of the satellite images) 4 5 LAIðÞ t ¼a and the NDVI derived from the QB and WV2 images S ðtÞb S ðtÞb 1 1 2 2 1þexp  1þexp c c 1 2 were examined, using regression analysis. This ð1Þ relationship is commonly used to extrapolate the LAI LAI DD (°C) tot vascular Environ. Res. Lett. 12 (2017) 095002 Table 2. Fit statistics and parameter values of the vascular plant leaf-area index (LAI) model (equation (1)) for each land cover class. The model was fitted to the community mean LAI of 7 measurement days. 2 a b Land cover class R RMSE ab c b c adj 1 1 2 2 Wet fen 0.98 0.055 1.09 227 74 153 9 Flood meadow 0.96 0.077 0.90 212 70 200 20 Dry fen 0.97 0.032 0.52 220 74 150 9 Graminoid tundra 0.96 0.063 0.83 241 126 200 9 Bog 0.95 0.032 0.53 138 113 200 9 Shrub tundra 0.76 0.045 0.69 83 68 200 9 Lichen tundra 0.63 0.032 0.26 58 43 130 9 Adjusted coefficient of determination. Root-mean-squared error. LAI over a landscape, using the NDVI as input (e.g. Shaver et al 2007, Williams et al 2008, Shaver et al 0.8 a 2013, Marushchak et al 2013). The regression relationships for both the total and vascular LAI were 0.6 applied to calculate the LAI maps, using the QB and WV2 images. We used IBM SPSS Statistics Version 22 0.4 (IBM Corp., Armonk, NY, USA) and JMP Pro 10.02 software (SAS Institute Inc., Cary, NC, USA) for statistical modeling and testing. 0.2 0.8 Results 0.6 At harvest time, the average vascular plant LAI was 0.55 across all the harvested plots. The highest values 0.4 were found in the wet fen and flood meadow, both dominated by graminoids, while the lichen and bare- 0.2 ground tundra showed the lowest values (figure 2(a)). 0.4 The moss LAI was highest in the dry fen and bog (figure 2(b)). The total LAI did not differ greatly among the LCs, excluding the lichen and bare-ground 0.2 tundra, due to the contrasting distribution of mosses and vascular plants in the communities. 0.0 The seasonal amplitudes of the vascular and total LAIs was largest in those communities with abundant graminoid vegetation, i.e. in the wet fen, graminoid tundra and flood meadow (figures 3(d)(i). Our modeled LAI showed that the development of the vegetation was delayed in the cool growing season of 2005 (figures 2(e) and (h)), whereas in 2012 the LAI Land cover class developed rapidly, due to the early start of the growing Figure 4. Boxplots of the normalized difference vegetation season and higher DD accumulation (figures 3(c), (f) index (NDVI) derived from the (a) QuickBird (QB) and (b) and (i)). The yearly differences were pronounced in WorldView-2 (WV2) reflectance, and (c) the difference the graminoid-dominated communities and small in between these for the various land cover classes (shown in order of decreasing wetness from the left). The horizontal the moss- and evergreen-dominated communities line, box, and whisker ends indicate the median, 25th and (3(d)(f)). The varying meteorological conditions 75th percentiles, and the 10th and 90th percentiles, respectively, and the data points outside this range are shown during the years of the satellite images led to by dots. The Z- and p-statistics of the Wilcoxon signed-rank differences in the LAI that were larger than would test of the difference between the early (QB image) and late- be expected, based solely on the difference between the season (WV2 image) NDVI in the various land cover classes are shown in panel (c) p < 0.05, p < 0.01, p < 0.001. dates of the images. The differences in the LC-specific NDVI values between the two satellite images were in line with the differences in vegetation development illustrated by the WV2 image, which represents the maximum LAI, the LAI. In the QB image, which represents the LAI in the NDVI maxima were found in the wet fen, flood the growth phase (figure 4(a)), the NDVI values were meadow, and graminoid tundra (figure 4(b)). The median difference in the NDVI and NDVI largest in the shrub-moss tundra and bog, whereas in QB WV2 Wet fen Flood meadow Graminoid tundra Dry fen Bog Shrub tundra Lichen tundra Bare ground NDVI - NDVI NDVI NDVI WV2 QB WV2 QB z=36* z=55** z=153*** z=78** z=210*** z=66** z=8 z=42 Environ. Res. Lett. 12 (2017) 095002 a QB WV2 this study Dry fen 2 Shaver et al 2007 Wet fen Bog Lichen tundra Shrub tundra Graminoid tundra Flood meadow 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 NDVI NDVI Figure 5. Relationships between normalized difference vegetation index (NDVI) and the field data of (a) the total leaf-area index (LAI) and (b) vascular plant LAI, with the NDVI derived from the QuickBird (QB) and WorldView-2 (WV2) satellite images, i.e. comparing LAI observations from 2014 with NDVI from 2005 and 2012. The open and closed circular data points denote individual study plots, while the land cover class means are indicated by the gray triangle and square symbols. For the regressions, the community 2 2 means were used as dependent variables: LAItot = 0.06 e(5.34NDVI ), R = 0.81, LAItot = 0.12 e(3.24NDVI , R = 0.86, QB adj. wv2 adj. 2 2 LAIvasc = 0.1 e(3.48NDVI ), R = 0.42, and LAIvasc = 0.05 e(3.36NDVI ), R = 0.92. The relationships between the QB adj. wv2 adj. NDVI and (c) total LAI and (d) vascular plant LAI, modeled to temporally match the QB and WV2 images. The data points represent the land cover class means and the regressions show are: LAItot = 0.1 e(2.83NDVI), R = 0.65 and LAIvasc. = 0.02 e adj. 2 8.0783NDVI (5.18NDVI), R = 0.84. LAI = modeled LAI, LAI = observed LAI. Equation (LAI = 0.0026e ) presented in adj. mod obs Shaver et al (2007 and 2013) also shown in the paned (d). values was significantly different from zero for all the Discussion plant communities, except for the lichen tundra and bare ground, which showed equally low NDVI values In this study, we documented the spatial and temporal in both images (figure 4(b) and (c)). The NDVI patterns of the LAI in the low-growth vegetation of the differences were larger among the LCs in the WV2 coastal arctic tundra and generalized these observa- image than in the QB image. tions for the entire study area, using VHSR satellite The relationship between the NDVI and LAI images. Our study illustrates the impact of the spatial showed a typical exponential shape with reasonably and temporal variation in the seasonality of vegetation, favorable adjusted coefficients of determination (R adj measured as the LAI, on the inference of remote- = 0.42–0.92, p > 0.05). The regression parameters, sensing products in the tundra. Interestingly, both the however, varied depending on the satellite image used plant communities and the LAI showed substantial due to the different timing of the images in relation to small-scale spatial variation in our study area, which the field data and growth stage (figure 5(a) and (b)). appears to be typical of tundra landscapes (Marush- The variation in total and vascular plant LAI was better chak et al 2013, Virtanen and Ek 2014). The temporal explained by the NDVI than the NDVI , due to dynamics of the LAI also differed among the plant WV2 QB the growth stage in the WV2 image better resembling communities. The graminoid-dominated vegetation the growth stage in the field data (figure 3). To showed intensive growth within the short arctic illustrate the spatial variation of the LAI in the study summer, leading to a large seasonal amplitude in area at the time of the satellite images, we calculated the LAI. In the Carex-dominated wet fens and flood the NDVI-LAI regressions, using values adjusted with meadows, the sparse moss cover further amplified the phenological model (figure 5(c) and (d)). For these seasonal differences in total greenness. The vascular LAI the relationship was roughly agreeable mixed vegetation of the evergreen and deciduous with a spatially more representative transfer function dwarf shrubs showed a smaller vascular plant LAI and in Shaver et al (2007, 2013) based on multiple arctic a more abundant moss cover and thus less seasonal sites (figure 5(d)). In the scale of whole focus area variation in its NDVI (figures 2 and 3). These particularly the vascular LAI differed between the differences in the LAI and NDVI dynamics were driven images. In the WV2 image, the average vascular LAI by the variation in the overwintering green biomass was 30% higher and the graminoid LAI in the and leaf production dynamics among the various graminoid communities was 2.3–2.9 f old compared to PFTs. Graminoids have little overwintering green their respective values in the QB image (figure 6). biomass, while the evergreen dwarf shrubs show rapid LAI LAI obs. mod. Environ. Res. Lett. 12 (2017) 095002 Vascular LAI, Vascular LAI, early season later season 0 - 0.074 0 - 0.139 0.075 - 0.198 0.14 - 0.376 0.199 - 0.309 0.377 - 0.585 0.31 - 0.396 0.586 - 0.739 0.397 - 0.47 0.74 - 0.892 0.471 - 0.544 0.893 - 1.073 0.545 - 0.631 1.074 - 1.296 0.632 - 0.779 1.297 - 1.575 0.78 - 1.113 1.576 - 2.411 1.114 - 3.153 2.412 - 3.554 Figure 6. Spatial extrapolation of the vascular plant leaf-area index (LAI), based on the normalized difference vegetation index (NDVI) from the QuickBird (QB) and WorldView-2 (WV2) images representing stages of a) increasing and b) peak LAI. The images were taken on 15 July 2005 and 12 August 2012, respectively. See figure 5(c) for the equation used to calculate the LAI and figure 1 for the distribution of the vegetation types. green-up in the spring and gradual leaf turnover the effects of chilling temperatures in the latter part of during the late summer (Johnson and Tieszen 1976, the growing season. The phenological dynamics are Saarinen 1998, Street et al 2007, Maanavilja et al 2010). also affected by the timing of thawing, thaw depth, Mosses, in turn, show little seasonal variation in previous year’s conditions, soil moisture variations, nutrient and pigment contents (Moore et al 2006, herbivory, and photoperiod (Arft et al 1999, March- Street et al 2012), but their water content affects their and et al 2004, Körner and Basler 2010, Oberbauer reflective properties (Vogelmann and Moss 1993). et al 2013). These factors may be only partly accounted We applied a simple regression model to estimate by the two-part DD model. the interannual variation in the seasonal LAI Our results show how the phenological stage of the development and to match the phenological phases vegetation can rapidly change in the short growing of the LAI illustrated in the two satellite images. The season of the Arctic and, therefore, that the timing of images were taken in different years and were 9 and satellite image acquisition really matters. We illustrate þ19 ordinal days apart from the date of the field data how the spatial pattern of the LAI in the two images of collection, which in turn represented the peak LAI the same area differ and how linking the vegetation season (figure 3). Both the NDVI values extracted patterns to the NDVI values of our QB image would for the survey plots and the modeled LAI values result in a biased relationship and affect the indicated that the earlier, i.e. 9 days, image interpretation of the remote-sensing products (Wil- represented a phase of vegetation development in liams et al 2008, Ustin and Gamon 2010). VHSR which the graminoid LAI had not yet reached its satellite imagery is a significant improvement over growing season maximum (figures 3 and 4). traditional imagery, such as Landsat, for mapping the Apparently, despite a larger deviation in the number vegetation and associated processes of heterogeneous of ordinal days, the later image was similar to the field landscapes, such as tundra (Laidler and Treitz 2003, data in terms of the growth phase. Virtanen and Ek 2014, Siewert et al 2015, Shrestha et al We acknowledge modeling the LAI values for years 2016). However, it is not unusual for these images to with no field measurements, but the continuous data be acquired in different years or dates than the field of daily GP that we have available for 2010–2014 data, because the availability of VHSR images of the max (S1) and the robust correlation between the LAI and Arctic is limited, due to infrequent satellite visits and GP (e.g. Laurila et al 2001, Street et al 2007), unfavorable cloud conditions and zenith angles in max however, provide strong support for the function that these areas (Hope and Stow 1996, Rees et al 2002, Stow we chose for modeling the seasonal LAI patterns. Our et al 2004, Westergaard-Nielsen et al 2013). Our data simple model with DD as an environmental driver showt that one must be cautious in interpreting seemed to be suitable for our site and for this time images without knowing the phenological stage of the period, but without cross-site validation it cannot be most abundant plant species and communities at the considered general, but rather a technical tool for local time of imaging. LAI estimation. Degree-day accumulation has found Moreover, these data add knowledge on arctic to be a reasonable driver of LAI in Arctic communities vegetation patterns and satellite derived NDVI in a less (e.g. Hollister et al 2005) and our model also included studied region in eastern Siberia representing a coastal 8 Environ. Res. Lett. 12 (2017) 095002 lowland middle Arctic tundra landscape. In order to our stay at the Tiksi Observatory and Yakutian Service spatially extrapolate LAI over the study area we applied for Hydrometeorology and Environmental Monitor- an empirical relationship between plot scale LAI ing for providing accommodation and access to the and VHRS NDVI with a typical exponential form observatory. (figure 5) observed across the Arctic (Van Wijk and Williams 2005, Stelzer and Welker 2006, Street et al 2007, Shaver et al 2007 and 2013, Williams et al 2008, References Stoy et al 2009, Stoy and Quaife 2015). We found the AARI 2016 Electronic archive AARI term meteorological and relationship roughly similar between our study with upper-air observations Hydrometeorological Observatory VHSR derived NDVI and, for example, study by AARI 2016 (station) Tiksi for 19322014 (www.aari.ru/ Shaver et al (2013) with multi-site data of field- main.php?sub=2&id=3) Arft A M et al 1999 Responses of tundra plants to experimental spectrometer derived NDVI (figure 5(d)). However, warming: meta-analysis of the international tundra similarities or differences in the NDVI-LAI relation- experiment Ecol. Monogr. 69 491–511 ship among sites and sensors deserve further attention, Bratsch S N, Epstein H E, Buchhorn M and Walker D A 2016 Differentiating among four arctic tundra plant because it is commonly used for large scale communities at Ivotuk, Alaska using field spectroscopy extrapolations, but are known to be affected by Rem. Sens. 8 51 changes in spatial scales (Spadavecchia et al 2008, Stoy Chapin F S III, Bret-Harte M S, Hobbie S E and Zhong H 1996 et al 2009, Stoy and Quaife 2015, Williams et al 2008), Plant functional types as predictors of transient responses of arctic vegetation to global change J. Veg. Sci. 7 347–58 between vegetation type or content (Stelzer and Chavez P S Jr 1988 An improved dark-object subtraction Welker 2006, Street et al 2007) and growing season technique for atmospheric scattering correction of (Street et al 2007). To our knowledge none of earlier multispectral data Rem. Sens. Environ. 24 459–79 studies has focused on similar Siberian calcareous Cramer W et al 2001 Global response of terrestrial ecosystem structure and function to CO and climate change: results tundra landscapes nor employed VHSR satellite 2 from six dynamic global vegetation models Glob. Change imagery to establish LAI-NDVI relationship there. Biol. 7 357–73 Data from multiple sites and including seasonal and Forbes B C Macias Fauria M and Zetterberg P I 2010 Russian multi-year variation would be valuable as verification Arctic warming and ‘greening’ are closely tracked by tundra shrub willows Glob. Change Biol. 16 1542–54 material for developing models to generate LAI maps Frost G V and Epstein H E 2014 Tall shrub and tree expansion of the Arctic based on satellite images in Siberian tundra ecotones since the 1960s Glob. Change Biol. 20 1264–77 Garrigues S et al 2008 Validation and intercomparison of global Conclusions leaf area index products derived from remote sensing data J. Geophys. Res. 113 G02028 Hollister R D, Webber P J and Bay C 2005 Plant response to Our objective was to determine the composition, temperature in northern Alaska: implications for predicting distribution, and seasonal dynamics of the LAI in the vegetation change Ecology 86 562–1570 plant communities of the arctic tundra and to assess Hope A S and Stow D A 1996 Shortwave reflectance properties of arctic tundra landscapes Landscape Function and how the seasonality of plant growth affects the Disturbance in Arctic Tundra ed J Reynolds and J interpretation of the vegetation signal derived from Tenhunen (Berlin Heidelberg: Springer) pp 155–64 satellite images. Our observations have significant Hugelius G, Virtanen T, Kaverin D, Pastukhov A, Rivkin F, implications for the evaluation and planning of optical Marchenko S, Romanovsky V and Kuhry P 2011 High- resolution mapping of ecosystem carbon storage and Earth observations and for compromising between potential effects of permafrost thaw in periglacial terrain, spatial and temporal resolution. Understanding the European Russian Arctic J. Geophys. Res. 116 G03024 small-scale spatial variation in plant communities, Johnson D A and Tieszen L L 1976 Aboveground biomass plant growth dynamics, and their constraints are allocation, leaf growth, and photosynthesis patterns in tundra plant forms in Arctic Alaska Oecologia 24 159–73 highly important in heterogeneous landscapes when Körner C and Basler D 2010 Phenology under global warming biological variables are interpreted, using VHSR Earth Science 327 1461–2 observation data. We conclude that the short growing Laidler G J and Treitz P 2003 Biophysical remote sensing of season of high latitudes, in association with climatic arctic environments Progr. Phys. Geogr. 27 44–68 Laidler G, Treitz P and Atkinson D 2008 Remote sensing of variation, sets special requirements for linking non- arctic vegetation: relations between the NDVI, spatial matching field data and satellite images in these areas. resolution and vegetation cover on Boothia Peninsula, Nunavut Arctic 61 1–13 Laurila T, Soegaard H, Lloyd C R, Aurela M, Tuovinen J-P and Nordstroem C 2001 Seasonal variations of net CO Acknowledgments exchange in European Arctic ecosystems Theor. Appl. Climatol. 70 183–201 We thank Emmi Vähä and Lauri Rosenius for field and Maanavilja L, Riutta T, Aurela M, Laurila T and Tuittila E-S 2010 laboratory assistance, and James Thompson for Spatial variation in CO exchange at a northern aapa mire Biogeochemistry 104 325–45 English revision. This research was supported by the Marchand F L, Nijs I, Heuer M, Mertens S, Kockelberh I, Academy of Finland (projects 269095 and 291736 for Pontailler J-Y, Impens I and Beyens L 2004 Climatic MA and TV). We thank G Chumachenko and warming responses of high arctic tundra Arct. Antarct. Alp. Res. 36 390–4 O Dmitrieva for kindly making arrangements for 9 Environ. Res. Lett. 12 (2017) 095002 Marushchak M E, Kiepe I, Biasi C, Elsakov V, Friborg T, Spadavecchia L, Williams M, Bell R, Stoy P C, Huntley B and Johansson T, Soegaard H, Virtanen T and Martikainen P Van Wijk M T 2008 Topographic controls on the leaf area 2013 Carbon dioxide balance of subarctic tundra from plot index and plant functional type of a tundra ecosystem J. to regional scales Biogeosciences 10 437–52 Ecol. 96 1238–51 McManusKM, Morton DC,Masek J G,WangD,SextonJO, Nagol Stelzer H and Welker J M 2006 Modeling the effect of J R, Ropars P and Boudreau S 2012 Satellite-based evidence for photosynthetic vegetation properties on the NDVI-LAI shrub and graminoid tundra expansion in northern Quebec relationship Ecology 87 2765–72 from 1986 to 2010 Glob. Change Biol. 18 2313–23 Stow D A et al 2004 Remote sensing of vegetation and land- Melton J R et al 2013 Present state of global wetland extent and cover change in arctic tundra ecosystems Rem. Sens. wetland methane modelling: conclusions from a model Environ. 89 281–308 inter-comparison project (WETCHIMP) Biogeosciences 10 Stoy P C and Quaife T 2015 Probabilistic downscaling of remote 753–88 sensing data with applications for multi-scale Moore T R, Lafleur P M, Poon D M, Heuman B W, Seaquist J biogeochemical flux modeling PLoS ONE 10 e0128935 W and Roulet N T 2006 Spring photosynthesis in a cool Stoy P C, Williams M, Spadavecchia L, Bell R A, Prieto-Blanco temperate bog Glob. Change Biol. 12 2323–35 A, Evans J G and Van Wijk M T 2009 Using information Mora C, Vieira G, Pina P, Lousada M and Christiansen H H theory to determine optimum pixel size and shape for 2015 Land cover classification csing high-resolution aerial ecological studies: aggregating land surface characteristics photography in Adventdalen, Svalbard Geogr. Ann. Ser. A: in arctic ecosystems Ecosystems 12 574–89 Phys. Geog. 97 473–88 Street L E, Shaver G R, Williams M and Van Wijk M T 2007 Oberbauer S F et al 2013 Phenological response of tundra plants What is the relationship between changes in canopy leaf to background climate variation tested using the area and changes in photosynthetic CO flux in arctic international tundra experiment Phil. Trans. R. Soc. B 368 ecosystems? J. Ecol. 95 139–50 20120481 Street L E, Stoy P C, Sommerkorn M, Fletcher B J, Sloan V L, Rautiainen M, Mõttus M, Heiskanen J, Akujärvi A, Majasalmi T Hill T C and Williams M 2012 Seasonal bryophyte and Stenberg P 2011 Seasonal reflectance dynamics of productivity in the sub-arctic: a comparison with vascular common understory types in a northern european boreal plants Funct. Ecol. 26 365–78 forest Rem. Sens. Environ. 115 3020–8 Tieszen L L and Johnson P L 1968 Pigment structure of some Rees G, Brown I, Mikkola K, Virtanen T and Werkman B 2002 arctic tundra communities Ecology 49 370–373 How can the dynamics of the tundra-taiga boundary be Tucker C J 1979 Red and photographic infrared linear remotely monitored? Ambio Spec. Rep. 12 56–62 combinations for monitoring vegetation Rem. Sens. Riutta T, Laine J and Tuittila E-S 2007 Sensitivity of CO Environ. 8 127–50 exchange of fen ecosystem components to water level Ustin S L and Gamon J A 2010 Remote sensing of plant variation Ecosystems 10 718–33 functional types New Phytol. 186 795–816 Saarinen T 1998 Demography of carex rostrata in a boreal Uttal T et al 2016 International arctic systems for observing the mesotrophic fen: shoot dynamics and biomass atmosphere (IASOA): an international polar year legacy development Ann. Bot. Fenn. 35 203–9 consortium Bull. Am. Meteorol. Soc. 97 1033–56 Shaver G R, Street L E, Rastetter E B, Van Wijk M T and Van Wijk M T and Williams M 2005 Optical instruments for Williams M 2007 Functional convergence in regulation of measuring leaf area index in low vegetation: application in net CO flux in heterogeneous tundra landscapes in Alaska Arctic ecosystems Ecol. Appl. 15 1462–70 and Sweden J. Ecol. 95 802–17 Virtanen T and Ek M 2014 The fragmented nature of tundra Shaver G R, Rastetter E B, Salmon V, Street L E, van de Weg M landscape Int. J. Appl. Earth Obs. Geoinf. 27 4–12 J, Rocha A, van Wijk M T and Williams M 2013 Pan-arctic Vogelmann J E and Moss D E 1993 Spectral reflectance modelling of net ecosystem exchange of CO Phil. Trans. measurements in the genus Sphagnum Rem. Sens. Environ. R. Soc. B 368 20120485 45 273–9 Shrestha R K, Arora V K and Melton J R 2016 The sensitivity of Westergaard-Nielsen A, Lund M, Hansen B U and Tamstorf M P simulated competition between different plant functional 2013 Camera derived vegetation greenness index as proxy types to subgrid-scale representation of vegetation in a for gross primary production in a low Arctic wetland area land surface model J. Geophys. Res.-Biogeosci. 121 809–28 ISPRS J. Photogramm. Rem. Sens. 86 89–99 Siewert M B, Hanisch J, Weiss N, Kuhry P, Maximov T C and Williams M, Bell R, Spadavecchia L, Street L E and van Wijk M Hugelius G 2015 Comparing carbon storage of siberian T 2008 Upscaling leaf area index in an Arctic landscape tundra and taiga permafrost ecosystems at very high spatial through multiscale observations Glob. Change Biol. 14 resolution J. Geophys. Res.-Biogeosci. 120 1973–94 1517–30 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Research Letters IOP Publishing

Spatial variation and seasonal dynamics of leaf-area index in the arctic tundra-implications for linking ground observations and satellite images

Loading next page...
 
/lp/iop-publishing/spatial-variation-and-seasonal-dynamics-of-leaf-area-index-in-the-NAxDTQm00B

References (52)

Copyright
Copyright © 2017 IOP Publishing Ltd
eISSN
1748-9326
DOI
10.1088/1748-9326/aa7f85
Publisher site
See Article on Publisher Site

Abstract

Vegetation in the arctic tundra typically consists of a small-scale mosaic of plant communities, with species differing in growth forms, seasonality, and biogeochemical properties. Characterization of this variation is essential for understanding and modeling the functioning of the arctic tundra in global carbon cycling, as well as for evaluating the resolution requirements for remote sensing. Our objective was to quantify the seasonal development of the leaf-area index (LAI) and its variation among plant communities in the arctic tundra near Tiksi, coastal Siberia, consisting of graminoid, dwarf shrub, moss, and lichen vegetation. We measured the LAI in the field and used two very-high-spatial resolution multispectral satellite images (QuickBird and WorldView-2), acquired at different phenological stages, to predict landscape-scale patterns. We used the empirical relationships between the plant community-specific LAI and degree-day accumulation (0 °C threshold) and quantified the relationship between the LAI and satellite NDVI (normalized difference vegetation index). Due to the temporal difference between the field data and satellite images, the LAI was approximated for the imagery dates, using the empirical model. LAI explained variation in the NDVI values well (R 0.42–0.92). Of the plant adj. functional types, the graminoid LAI showed the largest seasonal amplitudes and was the main cause of the varying spatial patterns of the NDVI and the related LAI between the two images. Our results illustrate how the short growing season, rapid development of the LAI, yearly climatic variation, and timing of the satellite data should be accounted for in matching imagery and field verification data in the Arctic region. observations of vegetation are equally essential when Introduction vegetation parameters, such as the leaf-area index Vegetation monitoring is a tool for detecting the (LAI), are included in ecosystem models (e.g. Cramer impacts of climate change on the composition and et al 2001, Melton et al 2013). Many key properties of phenology of arctic ecosystems. For example, satellite vegetation can be reasonably well inferred from image series spanning several decades have already spectral reflectance and thus mapped for large areas, revealed the large-scale greening of the arctic, a using remotely sensed data (Laidler and Treitz 2003, consequence of the increased plant growth and spatial Laidler et al 2008, Ustin and Gamon 2010). However, expansion of shrubs and trees (Stow et al 2004, Forbes the temporal and spatial scales of currently available et al 2010, Frost and Epstein 2014). Spatially extensive remote-sensing products may pose a challenge during © 2017 IOP Publishing Ltd Environ. Res. Lett. 12 (2017) 095002 20°0′0″E 50°0′0″E 90°0′0″E 130°0′0″E 160°0′0″E Bare Water Bog Flux tower Graminoid tundra Survey plots Dry fen Wet fen / half water Lichen tundra Shrub tundra Figure 1. Location of Tiksi (left) and the land-cover map of the study area with survey plot locations (right). The graminoid tundra and flood meadow were grouped in the land-cover classification. mapping of spatially heterogeneous landscapes, such small time difference between the image and the as the northern tundra (Virtanen and Ek 2014). ground-truth data can result in wide variations in The pixel size of the commonly used satellite terms of growth stage. imageries, e.g. those obtained from the Landsat and In this study, therefore, our objective was to Moderate Resolution Imaging Spectroradiometer quantify the spatial and seasonal variation in LAI (MODIS) satellites, ranges from tens to hundreds of among the dominant plant communities in the arctic meters and cannot reveal fine-scale heterogeneity in tundra and to evaluate how the seasonality of the vegetation and ecosystem properties (Laidler et al vegetation affects the interpretation of vegetation 2008, Virtanen and Ek 2014, Mora et al 2015, Bratsch structure from satellite images. We measured the et al 2016). This complicates the examination of plant seasonal development and spatial pattern of the LAI in growth responses to warming, which may vary among the coastal arctic tundra near Tiksi, NE Russia, in the neighboring communities (e.g. McManus et al 2012, summer of 2014 and examined how the normalized Bratsch et al 2016). To resolve this problem, images of difference vegetation index (NDVI), derived from the very high spatial resolution (VHSR, 0.32 m pixel reflectance data, varied among the plant communities size) have become available in recent years. Their and between the VHSR multispectral satellite images usage, however, is hampered by the high price, limited of two different growing seasons. To quantify the temporal availability, which is caused by the relatively dependence of the seasonal LAI development on the long revisit periods of high-resolution image sensor weather and for reconstructing the LAI values, we satellites in the same location, and by the frequent developed regression models between the LAI, degree- cloud cover and low solar angle in the Arctic (Hope days (DD) accumulation, and satellite-based NDVI. and Stow 1996, Rees et al 2002, Stow et al 2004, Based on these models, we then evaluated the impacts Westergaard-Nielsen et al 2013). Therefore, the best- that the temporal mismatch between the satellite quality image often does not temporally match the imagery and field data may have on the interpretation ground-truth data. of the NDVI and LAI distributions in the landscape. This temporal mismatch is a source of uncertainty in the end product, because the reflectance is Methods dependent on the amount of biomass, plant species composition, and the water, nutrient, and pigment Study site contents of plant tissues, all of which are affected by the growth stage of the vegetation (e.g. Ustin and The study site is located about 500 m from the coast of the Arctic Ocean near the Hydrometeorological Gamon 2010). Therefore, the spectral responses to the seasonal changes in vegetation properties should be Observatory of Tiksi in NE Russia (71.5936°N, 128.8850°E, figure 1). The climate at Tiksi is arctic, better understood (Garrigues et al 2008, Ustin and Gamon 2010, Rautiainen et al 2011, Westergaard- with very cold and windy winters, short but relatively Nielsen et al 2013). Since the growing season is short warm summers, and short shoulder seasons between and the changes in the LAI and biomass of the these. The mean annual temperature was 12.7 °C vegetation are rapid in the Arctic, it is likely that even a and the mean annual precipitation 323 mm in 60°0′0″N Environ. Res. Lett. 12 (2017) 095002 Sphagnum mosses (Sphagnum L.) and feathermosses a Dwarf shrub (e.g. Pleurozium schreberi (Willd. ex Brid.) Mitt) with Graminoid 1.0 shrubs were abundant in the dry fens, while the moss Salix spp. B. nana cover was sparse, due to aboveground water in the wet 0.8 Herb fens. The bogs showed typical microtopographic 0.6 variation, and their vegetation was characterized by 0.4 the presence of dwarf shrubs, dwarf birch (Betula nana 0.2 L.), Sphagnum, and feathermosses. The vegetation of the flood meadows along the stream and drier 0.0 graminoid tundra was dominated by graminoids Lichens 1.0 (sedges, grasses) and willows (Salix L. spp.). Abundant Other mosses Sphagnum spp. feathermoss coverage on the ground layer and dwarf 0.8 shrubs in the field layer characterized the shrub-moss 0.6 tundra. Lichen tundra patches alternated with stony 0.4 bare-ground surfaces. 0.2 Flow of the study 0.0 Several steps were needed to obtain time series of LAI, to model LAI for each land cover class (LC) for the years of satellite data, and to produce the LAI maps over the study area. The steps were, in brief, as follows, while the details are given in following paragraphs. Land cover class 1. Phenological dynamics of vegetation: Vegetation Figure 2. Mean leaf-area index (LAI) of (a) all vascular plants and (b) all non-vascular plants in the various land cover surveys (% cover and mean height of each plant classes (LC). In both cases, the LAI is divided into plant functional types) in sample of plots seven times functional types, while the LCs are shown in order of decreasing wetness from the left (the number of harvested over the study period in 2014. field plots in each type is given in parentheses). 2. Estimation of LAI on basis of % cover and height: Sample of plots were harvested after vegetation survey during the peak biomass and 1981–2010. Within this reference period, the average LAI of harvested material was measured at PFT growing season (0 °C threshold) lasted from 7 June to level. Data were used to develop regression 26 September, with DD of 668 (Arctic and Antarctic models to predict LAI. Research Institute AARI 2016). Meteorological data from the Hydrometeorological Observatory was used 3. Producing time series of total and vascular plant to calculate the DD for the examined periods in this LAI: A model using degree-day accumulation study. and chilling temperature accumulation as drivers The site represents a typical coastal tundra of was fitted for each land cover class and LAI was Eastern Siberia with alkaline bedrock and high plant modelled for years 2005, 2012, and 2014. species diversity. We focused on an area of approxi- 4. Mapping spatial distribution of LAI using satellite mately 1 km around the micrometeorological station, data: LAI was modeled over the study area using established in 2010 for eddy covariance (EC) relationship between satellite derived NDVI and measurements of the land-atmosphere exchange of LAI in the study plots. water, heat, carbon dioxide (CO ), and methane (CH ) (Uttal et al 2016, figure 1). The terrain around the EC mast was relatively flat; in addition to microtopographic variation there was a gentle slope Field data rising towards the north and a small stream running The field data on the vegetation were collected in the through the area. The vegetation and land-cover types summer of 2014, which had a warmer (871 DD) within the area were classified, using ground-based growing season than the long-term average. The 92 visual judgement during an expedition in 2012. These inventory plots with radius of 2.5 m were placed along land cover classes (LC) were characterized as (1) dry 16 compass points at regular distances of 25, 50, 75, fen, (2) wet fen, (3) bog, (4) lichen tundra, (5) shrub- 100, 150, and 250 m from the EC mast (figure 1). moss tundra, (6) graminoid tundra, (7) flood meadow, Several additional plots were monitored at distances of (8) bare ground, and (9) water. 300, 350, and 400 m to balance the number of plots in The fen and bog were peat-forming environments, the various LCs. In each plot, the vegetation was while the other land cover classes showed no clearly inventoried in four subplots (45 cm  45 cm in area), discernible peat. Sedges (Carex L. spp.) characterized located 2 m from the plot midpoint in four main the vascular plant vegetation in the fens (figure 2). compass directions. Each plot was classified according Wet fen (6) Flood meadow (11) Graminoid tundra (10) Dry fen (16) Bog (21) Shrub tundra (11) Lichen tundra (19) Bare ground (3) Vascular LAI Non-vascular LAI Environ. Res. Lett. 12 (2017) 095002 Table 1. Regression models describing the dependence of the leaf-area index (LAI) of the various plant functional types on their areal cover (C, %) and height (H, cm) in the Tiksi field plots surveyed in 2014. All regressions and parameters were significant at p < 0.05. a 2 b Plant functional type df reg., res. R adj Salix spp. LAI = 0.0126  C 1, 63 0.88 Dwarf shrub LAI = 0.0192  (C þ 0.001) þ 0.0397  H 2, 63 0.93 Betula nana LAI = 0.0132  C 1, 36 0.91 Graminoids LAI = 0.0150  C 1, 85 0.88 Herbs LAI = 0.0098  C þ 0.0046  H 2, 84 0.93 Degrees of freedom for regression and residuals. Adjusted coefficient of determination. to the apriori LC scheme described above (dry fen, substitution, to estimate the LAI for the subplots that wet fen, etc.), and the plot midpoint was georefer- were monitored for seasonal dynamics. enced, using a Global Positioning System (GPS) device and a measuring tape to achieve a location Satellite image acquisition and processing accuracy of 1–3m. To examine the spatial and temporal variation in the The vegetation was surveyed to characterize the vegetation patterns within the study area, we acquired species composition and to quantify the LAI and its two VHSR multispectral satellite images from the seasonal development in the various LCs of the archive of DigitalGlobe (Westminster, CO, USA). To tundra. The vegetation was inventoried as plant enable the comparison of images taken under different functional types (PFTs), following the typification by atmospheric conditions, the images were corrected for Hugelius et al (2011), which is a modification of that atmospheric scattering and transformed into surface by Chapin et al (1996). The PFTs included: (1) reflectance values, using the dark-object subtraction Sphagnum mosses, (2) other mosses, (3) lichen (4) method (Chavez 1988). dwarf shrubs, (5) deciduous shrub Betula nana,(6) The images, QuickBird (QB, DigitalGlobe, 15 July deciduous shrub Salix spp., (7) herbaceous species, 2005) and WorldView-2 (WV2, DigitalGlobe, 12 and (8) graminoids. August 2012), were chosen because they were of good The projection cover percentage of each PFT was quality and showed the best temporal matching with visually estimated, and the mean height was measured the collection of the peak season field verification data in each of the subplots during the main survey, 23–24 (23–24 July 2014) in terms of calendar days. The July 2014. These values were then averaged for the growing season of the QB image was shorter and main 2.5 m radius plot to relate the vegetation and the cooler (10 June–30 September 2005, 646 DD) than the satellite image-based spectral reflectance. The seasonal growing seasons of the WV2 image (22 May–30 development of the vegetation in one of the subplots September 2012, 1071 DD) and the verification data was monitored by seven successive surveys performed (6 June–24 September 2014, 863 DD). Thus, the QB between 2 July and 15 August 2014. Another subplot image captured a period during which the vegetation was harvested immediately after the main survey to was in the fast-growing phase, while the WV2 image quantify the one-sided LAI. The vascular plant captured the peak LAI period (figure 3 (a)(c)). material harvested was scanned, using a Canon MP The mean reflectance values were extracted for Navigator EX scanner (Canon Inc., Tokyo, Japan) and circular plots with a 2.5 m radius. The reflectance data by calculating the green surface area (= LAI) of the were used to calculate the normalized difference scanned images, using GNU Image Manipulation vegetation index, NDVI = (NIR  VIS)/(NIR þ VIS), Program 2 (GIMP 2) software. The LAI of the mosses which describes the absorbance of the red portion of was estimated as a projection coverage, i.e. we visible (VIS) light and the reflectance of near-infrared determined that a 100% cover would represent an (NIR) radiation by green vegetation. Thus, the NDVI LAI of 1 (Riutta et al 2007). This approach under- is an indicator of the quantity and photosynthetic estimates the true multilayered leaf area of mosses, but capacity of green vegetation and has commonly been probably estimates reasonably well their light-captur- used in spatial extrapolations of LAI (Tucker 1979, ing and reflectance properties, due to the lower Laidler and Treitz 2003, Shaver et al 2013). Supervised pigment and nutrient contents in moss tissues land-cover classification was carried out, based on a (Tieszen and Johnson 1968, Moore et al 2006, Street WV2 image (12 August 2012) to visualize the spatial et al 2012). Using the data of the harvested subplots, distribution of vegetation in the area (figure 1). we then calculated for each PFT the relationship between the areal cover, plant height, and LAI and, Temporal modeling of the LAI using these relationships, estimated the LAI for each We examined the factors determining the seasonal subplot at the time of the main survey (table 1). These development of the LAI to estimate it for the specific relationships were also used, assuming space-for-time dates of the two satellite images. Since our field data on 4 Environ. Res. Lett. 12 (2017) 095002 800 2012 2014 b 2005 d f Dry fen Wet fen Bog 1.0 Lichen tundra Shrub tundra Graminoid tundra Flood meadow 0.5 0.0 1.0 0.5 0.0 Aug Jun Jul Aug Jun Jul Aug Jun Jul Month Figure 3. Accumulation of (a)(c) the temperature sum above 0 °C (degree-days, DD), (d)(f) vascular plant leaf-area index (LAI), and (g)(i) total LAI in the dominant plant community types for the years of the field data (2014), QuickBird (QB) image (2005), and WorldView-2 (WV2) image (2012). The vascular plant LAI was modeled based on seasonal accumulated temperature (DD) (table 2), while the non-vascular LAI was assumed to be constant and directly proportional to the areal cover. The vertical dotted lines indicate the timing of the satellite images and of the field data. LAI development originated from a single growing In this equation, t is the time, S is the DD season only, we used long-term (years 2010–2014) accumulated from all daily mean air temperatures over data on the daily maximum ecosystem photosynthesis 0 °C, and S is the DD accumulated during the latter (GP )to define a functional form that describes the part of the growing season (after 15 July) from the max seasonal growth of vegetation from soil thawing to daily mean air temperatures above 0 °C, but below maximum activity and further to senescence. This 10 °C and a, b , b , c , and c are the parameters to be 1 2 1 2 approach is justified by the close relationship between estimated. The equation was fitted to the total vascular the LAI and GP (e.g. Laurila et al 2001, Street et al plant LAI data for each LC (table 2). The fits obtained max 2007). The daily GP was derived from the were favorable for the period covered by field max continuous CO flux data measured with the EC observations in 2014, and the values outside the method at the site. The GP was determined from measurement period (2 July–15 August) were not used max the eddy covariance flux as the night-day difference in in any further analysis. net ecosystem CO exchange (NEE). The daily GP 2 max Data analysis was obtained as the difference of the 7 day running A paired sample Wilcoxon signed-rank test was used mean of the nighttime NEE (photosynthetic photon 2 1 to determine whether the vegetation classes showed flux density < 20 mmol m s ) and 3 day running significant seasonal differences in their NDVI signals; mean of the daytime NEE (PPFD > 600 mmol m i.e. the NDVI (15 July 2005) and NDVI s ). A function including two temperature-depen- QB WV2 (12 August 2012) were paired for each PCT. To dent sigmoid terms operating during different phases further illustrate the seasonal changes in the various of the growing season proved suitable for modeling the plant communities, we calculated the difference in GP cycle (supplement 1 available at stacks.iop.org/ max NDVI between the late-season and early-season image ERL/12/095002/mmedia). It was fitted to the LAI data (NDVI NDVI ). The relationships between the WV2 QB 2 3 LAI (either measured in the 2014 field survey or 1 1 estimated for the actual dates of the satellite images) 4 5 LAIðÞ t ¼a and the NDVI derived from the QB and WV2 images S ðtÞb S ðtÞb 1 1 2 2 1þexp  1þexp c c 1 2 were examined, using regression analysis. This ð1Þ relationship is commonly used to extrapolate the LAI LAI DD (°C) tot vascular Environ. Res. Lett. 12 (2017) 095002 Table 2. Fit statistics and parameter values of the vascular plant leaf-area index (LAI) model (equation (1)) for each land cover class. The model was fitted to the community mean LAI of 7 measurement days. 2 a b Land cover class R RMSE ab c b c adj 1 1 2 2 Wet fen 0.98 0.055 1.09 227 74 153 9 Flood meadow 0.96 0.077 0.90 212 70 200 20 Dry fen 0.97 0.032 0.52 220 74 150 9 Graminoid tundra 0.96 0.063 0.83 241 126 200 9 Bog 0.95 0.032 0.53 138 113 200 9 Shrub tundra 0.76 0.045 0.69 83 68 200 9 Lichen tundra 0.63 0.032 0.26 58 43 130 9 Adjusted coefficient of determination. Root-mean-squared error. LAI over a landscape, using the NDVI as input (e.g. Shaver et al 2007, Williams et al 2008, Shaver et al 0.8 a 2013, Marushchak et al 2013). The regression relationships for both the total and vascular LAI were 0.6 applied to calculate the LAI maps, using the QB and WV2 images. We used IBM SPSS Statistics Version 22 0.4 (IBM Corp., Armonk, NY, USA) and JMP Pro 10.02 software (SAS Institute Inc., Cary, NC, USA) for statistical modeling and testing. 0.2 0.8 Results 0.6 At harvest time, the average vascular plant LAI was 0.55 across all the harvested plots. The highest values 0.4 were found in the wet fen and flood meadow, both dominated by graminoids, while the lichen and bare- 0.2 ground tundra showed the lowest values (figure 2(a)). 0.4 The moss LAI was highest in the dry fen and bog (figure 2(b)). The total LAI did not differ greatly among the LCs, excluding the lichen and bare-ground 0.2 tundra, due to the contrasting distribution of mosses and vascular plants in the communities. 0.0 The seasonal amplitudes of the vascular and total LAIs was largest in those communities with abundant graminoid vegetation, i.e. in the wet fen, graminoid tundra and flood meadow (figures 3(d)(i). Our modeled LAI showed that the development of the vegetation was delayed in the cool growing season of 2005 (figures 2(e) and (h)), whereas in 2012 the LAI Land cover class developed rapidly, due to the early start of the growing Figure 4. Boxplots of the normalized difference vegetation season and higher DD accumulation (figures 3(c), (f) index (NDVI) derived from the (a) QuickBird (QB) and (b) and (i)). The yearly differences were pronounced in WorldView-2 (WV2) reflectance, and (c) the difference the graminoid-dominated communities and small in between these for the various land cover classes (shown in order of decreasing wetness from the left). The horizontal the moss- and evergreen-dominated communities line, box, and whisker ends indicate the median, 25th and (3(d)(f)). The varying meteorological conditions 75th percentiles, and the 10th and 90th percentiles, respectively, and the data points outside this range are shown during the years of the satellite images led to by dots. The Z- and p-statistics of the Wilcoxon signed-rank differences in the LAI that were larger than would test of the difference between the early (QB image) and late- be expected, based solely on the difference between the season (WV2 image) NDVI in the various land cover classes are shown in panel (c) p < 0.05, p < 0.01, p < 0.001. dates of the images. The differences in the LC-specific NDVI values between the two satellite images were in line with the differences in vegetation development illustrated by the WV2 image, which represents the maximum LAI, the LAI. In the QB image, which represents the LAI in the NDVI maxima were found in the wet fen, flood the growth phase (figure 4(a)), the NDVI values were meadow, and graminoid tundra (figure 4(b)). The median difference in the NDVI and NDVI largest in the shrub-moss tundra and bog, whereas in QB WV2 Wet fen Flood meadow Graminoid tundra Dry fen Bog Shrub tundra Lichen tundra Bare ground NDVI - NDVI NDVI NDVI WV2 QB WV2 QB z=36* z=55** z=153*** z=78** z=210*** z=66** z=8 z=42 Environ. Res. Lett. 12 (2017) 095002 a QB WV2 this study Dry fen 2 Shaver et al 2007 Wet fen Bog Lichen tundra Shrub tundra Graminoid tundra Flood meadow 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 NDVI NDVI Figure 5. Relationships between normalized difference vegetation index (NDVI) and the field data of (a) the total leaf-area index (LAI) and (b) vascular plant LAI, with the NDVI derived from the QuickBird (QB) and WorldView-2 (WV2) satellite images, i.e. comparing LAI observations from 2014 with NDVI from 2005 and 2012. The open and closed circular data points denote individual study plots, while the land cover class means are indicated by the gray triangle and square symbols. For the regressions, the community 2 2 means were used as dependent variables: LAItot = 0.06 e(5.34NDVI ), R = 0.81, LAItot = 0.12 e(3.24NDVI , R = 0.86, QB adj. wv2 adj. 2 2 LAIvasc = 0.1 e(3.48NDVI ), R = 0.42, and LAIvasc = 0.05 e(3.36NDVI ), R = 0.92. The relationships between the QB adj. wv2 adj. NDVI and (c) total LAI and (d) vascular plant LAI, modeled to temporally match the QB and WV2 images. The data points represent the land cover class means and the regressions show are: LAItot = 0.1 e(2.83NDVI), R = 0.65 and LAIvasc. = 0.02 e adj. 2 8.0783NDVI (5.18NDVI), R = 0.84. LAI = modeled LAI, LAI = observed LAI. Equation (LAI = 0.0026e ) presented in adj. mod obs Shaver et al (2007 and 2013) also shown in the paned (d). values was significantly different from zero for all the Discussion plant communities, except for the lichen tundra and bare ground, which showed equally low NDVI values In this study, we documented the spatial and temporal in both images (figure 4(b) and (c)). The NDVI patterns of the LAI in the low-growth vegetation of the differences were larger among the LCs in the WV2 coastal arctic tundra and generalized these observa- image than in the QB image. tions for the entire study area, using VHSR satellite The relationship between the NDVI and LAI images. Our study illustrates the impact of the spatial showed a typical exponential shape with reasonably and temporal variation in the seasonality of vegetation, favorable adjusted coefficients of determination (R adj measured as the LAI, on the inference of remote- = 0.42–0.92, p > 0.05). The regression parameters, sensing products in the tundra. Interestingly, both the however, varied depending on the satellite image used plant communities and the LAI showed substantial due to the different timing of the images in relation to small-scale spatial variation in our study area, which the field data and growth stage (figure 5(a) and (b)). appears to be typical of tundra landscapes (Marush- The variation in total and vascular plant LAI was better chak et al 2013, Virtanen and Ek 2014). The temporal explained by the NDVI than the NDVI , due to dynamics of the LAI also differed among the plant WV2 QB the growth stage in the WV2 image better resembling communities. The graminoid-dominated vegetation the growth stage in the field data (figure 3). To showed intensive growth within the short arctic illustrate the spatial variation of the LAI in the study summer, leading to a large seasonal amplitude in area at the time of the satellite images, we calculated the LAI. In the Carex-dominated wet fens and flood the NDVI-LAI regressions, using values adjusted with meadows, the sparse moss cover further amplified the phenological model (figure 5(c) and (d)). For these seasonal differences in total greenness. The vascular LAI the relationship was roughly agreeable mixed vegetation of the evergreen and deciduous with a spatially more representative transfer function dwarf shrubs showed a smaller vascular plant LAI and in Shaver et al (2007, 2013) based on multiple arctic a more abundant moss cover and thus less seasonal sites (figure 5(d)). In the scale of whole focus area variation in its NDVI (figures 2 and 3). These particularly the vascular LAI differed between the differences in the LAI and NDVI dynamics were driven images. In the WV2 image, the average vascular LAI by the variation in the overwintering green biomass was 30% higher and the graminoid LAI in the and leaf production dynamics among the various graminoid communities was 2.3–2.9 f old compared to PFTs. Graminoids have little overwintering green their respective values in the QB image (figure 6). biomass, while the evergreen dwarf shrubs show rapid LAI LAI obs. mod. Environ. Res. Lett. 12 (2017) 095002 Vascular LAI, Vascular LAI, early season later season 0 - 0.074 0 - 0.139 0.075 - 0.198 0.14 - 0.376 0.199 - 0.309 0.377 - 0.585 0.31 - 0.396 0.586 - 0.739 0.397 - 0.47 0.74 - 0.892 0.471 - 0.544 0.893 - 1.073 0.545 - 0.631 1.074 - 1.296 0.632 - 0.779 1.297 - 1.575 0.78 - 1.113 1.576 - 2.411 1.114 - 3.153 2.412 - 3.554 Figure 6. Spatial extrapolation of the vascular plant leaf-area index (LAI), based on the normalized difference vegetation index (NDVI) from the QuickBird (QB) and WorldView-2 (WV2) images representing stages of a) increasing and b) peak LAI. The images were taken on 15 July 2005 and 12 August 2012, respectively. See figure 5(c) for the equation used to calculate the LAI and figure 1 for the distribution of the vegetation types. green-up in the spring and gradual leaf turnover the effects of chilling temperatures in the latter part of during the late summer (Johnson and Tieszen 1976, the growing season. The phenological dynamics are Saarinen 1998, Street et al 2007, Maanavilja et al 2010). also affected by the timing of thawing, thaw depth, Mosses, in turn, show little seasonal variation in previous year’s conditions, soil moisture variations, nutrient and pigment contents (Moore et al 2006, herbivory, and photoperiod (Arft et al 1999, March- Street et al 2012), but their water content affects their and et al 2004, Körner and Basler 2010, Oberbauer reflective properties (Vogelmann and Moss 1993). et al 2013). These factors may be only partly accounted We applied a simple regression model to estimate by the two-part DD model. the interannual variation in the seasonal LAI Our results show how the phenological stage of the development and to match the phenological phases vegetation can rapidly change in the short growing of the LAI illustrated in the two satellite images. The season of the Arctic and, therefore, that the timing of images were taken in different years and were 9 and satellite image acquisition really matters. We illustrate þ19 ordinal days apart from the date of the field data how the spatial pattern of the LAI in the two images of collection, which in turn represented the peak LAI the same area differ and how linking the vegetation season (figure 3). Both the NDVI values extracted patterns to the NDVI values of our QB image would for the survey plots and the modeled LAI values result in a biased relationship and affect the indicated that the earlier, i.e. 9 days, image interpretation of the remote-sensing products (Wil- represented a phase of vegetation development in liams et al 2008, Ustin and Gamon 2010). VHSR which the graminoid LAI had not yet reached its satellite imagery is a significant improvement over growing season maximum (figures 3 and 4). traditional imagery, such as Landsat, for mapping the Apparently, despite a larger deviation in the number vegetation and associated processes of heterogeneous of ordinal days, the later image was similar to the field landscapes, such as tundra (Laidler and Treitz 2003, data in terms of the growth phase. Virtanen and Ek 2014, Siewert et al 2015, Shrestha et al We acknowledge modeling the LAI values for years 2016). However, it is not unusual for these images to with no field measurements, but the continuous data be acquired in different years or dates than the field of daily GP that we have available for 2010–2014 data, because the availability of VHSR images of the max (S1) and the robust correlation between the LAI and Arctic is limited, due to infrequent satellite visits and GP (e.g. Laurila et al 2001, Street et al 2007), unfavorable cloud conditions and zenith angles in max however, provide strong support for the function that these areas (Hope and Stow 1996, Rees et al 2002, Stow we chose for modeling the seasonal LAI patterns. Our et al 2004, Westergaard-Nielsen et al 2013). Our data simple model with DD as an environmental driver showt that one must be cautious in interpreting seemed to be suitable for our site and for this time images without knowing the phenological stage of the period, but without cross-site validation it cannot be most abundant plant species and communities at the considered general, but rather a technical tool for local time of imaging. LAI estimation. Degree-day accumulation has found Moreover, these data add knowledge on arctic to be a reasonable driver of LAI in Arctic communities vegetation patterns and satellite derived NDVI in a less (e.g. Hollister et al 2005) and our model also included studied region in eastern Siberia representing a coastal 8 Environ. Res. Lett. 12 (2017) 095002 lowland middle Arctic tundra landscape. In order to our stay at the Tiksi Observatory and Yakutian Service spatially extrapolate LAI over the study area we applied for Hydrometeorology and Environmental Monitor- an empirical relationship between plot scale LAI ing for providing accommodation and access to the and VHRS NDVI with a typical exponential form observatory. (figure 5) observed across the Arctic (Van Wijk and Williams 2005, Stelzer and Welker 2006, Street et al 2007, Shaver et al 2007 and 2013, Williams et al 2008, References Stoy et al 2009, Stoy and Quaife 2015). We found the AARI 2016 Electronic archive AARI term meteorological and relationship roughly similar between our study with upper-air observations Hydrometeorological Observatory VHSR derived NDVI and, for example, study by AARI 2016 (station) Tiksi for 19322014 (www.aari.ru/ Shaver et al (2013) with multi-site data of field- main.php?sub=2&id=3) Arft A M et al 1999 Responses of tundra plants to experimental spectrometer derived NDVI (figure 5(d)). However, warming: meta-analysis of the international tundra similarities or differences in the NDVI-LAI relation- experiment Ecol. Monogr. 69 491–511 ship among sites and sensors deserve further attention, Bratsch S N, Epstein H E, Buchhorn M and Walker D A 2016 Differentiating among four arctic tundra plant because it is commonly used for large scale communities at Ivotuk, Alaska using field spectroscopy extrapolations, but are known to be affected by Rem. Sens. 8 51 changes in spatial scales (Spadavecchia et al 2008, Stoy Chapin F S III, Bret-Harte M S, Hobbie S E and Zhong H 1996 et al 2009, Stoy and Quaife 2015, Williams et al 2008), Plant functional types as predictors of transient responses of arctic vegetation to global change J. Veg. Sci. 7 347–58 between vegetation type or content (Stelzer and Chavez P S Jr 1988 An improved dark-object subtraction Welker 2006, Street et al 2007) and growing season technique for atmospheric scattering correction of (Street et al 2007). To our knowledge none of earlier multispectral data Rem. Sens. Environ. 24 459–79 studies has focused on similar Siberian calcareous Cramer W et al 2001 Global response of terrestrial ecosystem structure and function to CO and climate change: results tundra landscapes nor employed VHSR satellite 2 from six dynamic global vegetation models Glob. Change imagery to establish LAI-NDVI relationship there. Biol. 7 357–73 Data from multiple sites and including seasonal and Forbes B C Macias Fauria M and Zetterberg P I 2010 Russian multi-year variation would be valuable as verification Arctic warming and ‘greening’ are closely tracked by tundra shrub willows Glob. Change Biol. 16 1542–54 material for developing models to generate LAI maps Frost G V and Epstein H E 2014 Tall shrub and tree expansion of the Arctic based on satellite images in Siberian tundra ecotones since the 1960s Glob. Change Biol. 20 1264–77 Garrigues S et al 2008 Validation and intercomparison of global Conclusions leaf area index products derived from remote sensing data J. Geophys. Res. 113 G02028 Hollister R D, Webber P J and Bay C 2005 Plant response to Our objective was to determine the composition, temperature in northern Alaska: implications for predicting distribution, and seasonal dynamics of the LAI in the vegetation change Ecology 86 562–1570 plant communities of the arctic tundra and to assess Hope A S and Stow D A 1996 Shortwave reflectance properties of arctic tundra landscapes Landscape Function and how the seasonality of plant growth affects the Disturbance in Arctic Tundra ed J Reynolds and J interpretation of the vegetation signal derived from Tenhunen (Berlin Heidelberg: Springer) pp 155–64 satellite images. Our observations have significant Hugelius G, Virtanen T, Kaverin D, Pastukhov A, Rivkin F, implications for the evaluation and planning of optical Marchenko S, Romanovsky V and Kuhry P 2011 High- resolution mapping of ecosystem carbon storage and Earth observations and for compromising between potential effects of permafrost thaw in periglacial terrain, spatial and temporal resolution. Understanding the European Russian Arctic J. Geophys. Res. 116 G03024 small-scale spatial variation in plant communities, Johnson D A and Tieszen L L 1976 Aboveground biomass plant growth dynamics, and their constraints are allocation, leaf growth, and photosynthesis patterns in tundra plant forms in Arctic Alaska Oecologia 24 159–73 highly important in heterogeneous landscapes when Körner C and Basler D 2010 Phenology under global warming biological variables are interpreted, using VHSR Earth Science 327 1461–2 observation data. We conclude that the short growing Laidler G J and Treitz P 2003 Biophysical remote sensing of season of high latitudes, in association with climatic arctic environments Progr. Phys. Geogr. 27 44–68 Laidler G, Treitz P and Atkinson D 2008 Remote sensing of variation, sets special requirements for linking non- arctic vegetation: relations between the NDVI, spatial matching field data and satellite images in these areas. resolution and vegetation cover on Boothia Peninsula, Nunavut Arctic 61 1–13 Laurila T, Soegaard H, Lloyd C R, Aurela M, Tuovinen J-P and Nordstroem C 2001 Seasonal variations of net CO Acknowledgments exchange in European Arctic ecosystems Theor. Appl. Climatol. 70 183–201 We thank Emmi Vähä and Lauri Rosenius for field and Maanavilja L, Riutta T, Aurela M, Laurila T and Tuittila E-S 2010 laboratory assistance, and James Thompson for Spatial variation in CO exchange at a northern aapa mire Biogeochemistry 104 325–45 English revision. This research was supported by the Marchand F L, Nijs I, Heuer M, Mertens S, Kockelberh I, Academy of Finland (projects 269095 and 291736 for Pontailler J-Y, Impens I and Beyens L 2004 Climatic MA and TV). We thank G Chumachenko and warming responses of high arctic tundra Arct. Antarct. Alp. Res. 36 390–4 O Dmitrieva for kindly making arrangements for 9 Environ. Res. Lett. 12 (2017) 095002 Marushchak M E, Kiepe I, Biasi C, Elsakov V, Friborg T, Spadavecchia L, Williams M, Bell R, Stoy P C, Huntley B and Johansson T, Soegaard H, Virtanen T and Martikainen P Van Wijk M T 2008 Topographic controls on the leaf area 2013 Carbon dioxide balance of subarctic tundra from plot index and plant functional type of a tundra ecosystem J. to regional scales Biogeosciences 10 437–52 Ecol. 96 1238–51 McManusKM, Morton DC,Masek J G,WangD,SextonJO, Nagol Stelzer H and Welker J M 2006 Modeling the effect of J R, Ropars P and Boudreau S 2012 Satellite-based evidence for photosynthetic vegetation properties on the NDVI-LAI shrub and graminoid tundra expansion in northern Quebec relationship Ecology 87 2765–72 from 1986 to 2010 Glob. Change Biol. 18 2313–23 Stow D A et al 2004 Remote sensing of vegetation and land- Melton J R et al 2013 Present state of global wetland extent and cover change in arctic tundra ecosystems Rem. Sens. wetland methane modelling: conclusions from a model Environ. 89 281–308 inter-comparison project (WETCHIMP) Biogeosciences 10 Stoy P C and Quaife T 2015 Probabilistic downscaling of remote 753–88 sensing data with applications for multi-scale Moore T R, Lafleur P M, Poon D M, Heuman B W, Seaquist J biogeochemical flux modeling PLoS ONE 10 e0128935 W and Roulet N T 2006 Spring photosynthesis in a cool Stoy P C, Williams M, Spadavecchia L, Bell R A, Prieto-Blanco temperate bog Glob. Change Biol. 12 2323–35 A, Evans J G and Van Wijk M T 2009 Using information Mora C, Vieira G, Pina P, Lousada M and Christiansen H H theory to determine optimum pixel size and shape for 2015 Land cover classification csing high-resolution aerial ecological studies: aggregating land surface characteristics photography in Adventdalen, Svalbard Geogr. Ann. Ser. A: in arctic ecosystems Ecosystems 12 574–89 Phys. Geog. 97 473–88 Street L E, Shaver G R, Williams M and Van Wijk M T 2007 Oberbauer S F et al 2013 Phenological response of tundra plants What is the relationship between changes in canopy leaf to background climate variation tested using the area and changes in photosynthetic CO flux in arctic international tundra experiment Phil. Trans. R. Soc. B 368 ecosystems? J. Ecol. 95 139–50 20120481 Street L E, Stoy P C, Sommerkorn M, Fletcher B J, Sloan V L, Rautiainen M, Mõttus M, Heiskanen J, Akujärvi A, Majasalmi T Hill T C and Williams M 2012 Seasonal bryophyte and Stenberg P 2011 Seasonal reflectance dynamics of productivity in the sub-arctic: a comparison with vascular common understory types in a northern european boreal plants Funct. Ecol. 26 365–78 forest Rem. Sens. Environ. 115 3020–8 Tieszen L L and Johnson P L 1968 Pigment structure of some Rees G, Brown I, Mikkola K, Virtanen T and Werkman B 2002 arctic tundra communities Ecology 49 370–373 How can the dynamics of the tundra-taiga boundary be Tucker C J 1979 Red and photographic infrared linear remotely monitored? Ambio Spec. Rep. 12 56–62 combinations for monitoring vegetation Rem. Sens. Riutta T, Laine J and Tuittila E-S 2007 Sensitivity of CO Environ. 8 127–50 exchange of fen ecosystem components to water level Ustin S L and Gamon J A 2010 Remote sensing of plant variation Ecosystems 10 718–33 functional types New Phytol. 186 795–816 Saarinen T 1998 Demography of carex rostrata in a boreal Uttal T et al 2016 International arctic systems for observing the mesotrophic fen: shoot dynamics and biomass atmosphere (IASOA): an international polar year legacy development Ann. Bot. Fenn. 35 203–9 consortium Bull. Am. Meteorol. Soc. 97 1033–56 Shaver G R, Street L E, Rastetter E B, Van Wijk M T and Van Wijk M T and Williams M 2005 Optical instruments for Williams M 2007 Functional convergence in regulation of measuring leaf area index in low vegetation: application in net CO flux in heterogeneous tundra landscapes in Alaska Arctic ecosystems Ecol. Appl. 15 1462–70 and Sweden J. Ecol. 95 802–17 Virtanen T and Ek M 2014 The fragmented nature of tundra Shaver G R, Rastetter E B, Salmon V, Street L E, van de Weg M landscape Int. J. Appl. Earth Obs. Geoinf. 27 4–12 J, Rocha A, van Wijk M T and Williams M 2013 Pan-arctic Vogelmann J E and Moss D E 1993 Spectral reflectance modelling of net ecosystem exchange of CO Phil. Trans. measurements in the genus Sphagnum Rem. Sens. Environ. R. Soc. B 368 20120485 45 273–9 Shrestha R K, Arora V K and Melton J R 2016 The sensitivity of Westergaard-Nielsen A, Lund M, Hansen B U and Tamstorf M P simulated competition between different plant functional 2013 Camera derived vegetation greenness index as proxy types to subgrid-scale representation of vegetation in a for gross primary production in a low Arctic wetland area land surface model J. Geophys. Res.-Biogeosci. 121 809–28 ISPRS J. Photogramm. Rem. Sens. 86 89–99 Siewert M B, Hanisch J, Weiss N, Kuhry P, Maximov T C and Williams M, Bell R, Spadavecchia L, Street L E and van Wijk M Hugelius G 2015 Comparing carbon storage of siberian T 2008 Upscaling leaf area index in an Arctic landscape tundra and taiga permafrost ecosystems at very high spatial through multiscale observations Glob. Change Biol. 14 resolution J. Geophys. Res.-Biogeosci. 120 1973–94 1517–30

Journal

Environmental Research LettersIOP Publishing

Published: Sep 1, 2017

There are no references for this article.