No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Spatio-Temporal Statistical Methods for Monitoring of Land Surface Phenology
Abstract
The onset of greening, the start of senescence, the timing of the maximum of the growing season and the growing season length are frequently calculated phenology metrics based on satellite imagery. However, since it is complicated to validate the much coarser spatial resolution observations of land surface phenology even with networks of ground observations of phenology, it is often unclear as to what the land surface phenology metrics actually quantify. For example, in northern biomes, the greatest increase in a satellite derived vegetation index indicated as "start of the season" (SOS) in some methods, can often be due to snow melt. The end of the greenness (EOS) metric, on the other hand, could measure an extended period of cloudiness instead of actual vegetation senescence. Since the relationship between satellite and ground observations of phenological events is ambiguous, many techniques have been developed. Here I discuss and compare some of the more commonly applied methods to derive land surface phenology metrics: delayed moving average method, percent threshold method, quadratic models based on accumulated growing degree-days and the MOD12 phenology product, which relies on piecewise sigmoidal models. I demonstrate the methods using both NDVI and EVI time series from 2001 and 2007 derived from MODIS/Terra+Aqua Nadir BRDF-Adjusted Reflectance 16-Day L3 Global 0.05Deg CMG V005 (MCD43C4) data for North America north of 30°N. To remove snow-covered pixels, I use the snow and ice QA flags. To compare the methods I evaluate the spatio-temporal differences in phenological metrics and discuss the bias- variance dilemma that involves the trade-off between an over-fitted model that is too complicated and an over- smoothed model that is too simple. Finally, I discuss the related issue of the possibility of statistically comparing the land surface phenology metrics for two years of data which represent extremes in the polarity of the North Atlantic Oscillation (2000 and 2007).
... In the case where the typical dynamic is periodic, such as the seasonal and growth patterns of vegetation, Fourier (harmonic) curves can be fit to any pixel, provided that the time period yields a sufficient quantity of quality data points to obtain a good representation of a curve (Brooks et al., 2012). Numerous authors have used this approach to characterize land surface phenology (e.g., Moody and Johnson, 2001;De Beurs and Henebry, 2010) and to extract the harmonic function coefficients as predictor variables (Immerzeel et al., 2005;Brooks et al., 2016;Wilson et al., 2018;Derwin et al., 2020). ...
... Deviation from the harmonic curve is evidence of a response to large disturbances, such as thins or clear cuts. One of the advantages of harmonic regression approaches is that they are sensitive to these larger disturbances, but are otherwise insensitive to noise and missing data (De Beurs and Henebry, 2010;Brooks et al., 2012;Wilson et al., 2018). As such, harmonic analysis has been used to characterize land surface phenology (Moody and Johnson, 2001;De Beurs and Henebry, 2010), forest growth and disturbance (Brooks et al., 2016), tree canopy cover (Derwin et al., 2020), and other forest attributes (Wilson et al., 2018). ...
... One of the advantages of harmonic regression approaches is that they are sensitive to these larger disturbances, but are otherwise insensitive to noise and missing data (De Beurs and Henebry, 2010;Brooks et al., 2012;Wilson et al., 2018). As such, harmonic analysis has been used to characterize land surface phenology (Moody and Johnson, 2001;De Beurs and Henebry, 2010), forest growth and disturbance (Brooks et al., 2016), tree canopy cover (Derwin et al., 2020), and other forest attributes (Wilson et al., 2018). The use of the constant, sine and cosine coefficients as predictors for a variety of vegetation characteristics, including forest cover and disturbance, is well established in the literature (Immerzeel et al., 2005;Brooks et al., 2016;Wilson et al., 2018;Derwin et al., 2020). ...
The southeastern United States is unique in terms of both the intensity and scale of forest management, which includes substantial thinning and other forms of harvesting. Because thinning is not a land use transition, and the disturbance signal is relatively subtle compared to a clear cut, there is a dearth of studies that attempt to detect thinning over large areas. Our goal was to detect pine thins as an indicator of active forest management using Landsat data. Areas which undergo thinning are indicative of active forest management in the region.
Our approach uses a machine learning method which combines first-order harmonics and metrics from 3-year Fourier regression of Landsat time series stacks, layers from the Global Forest Change product, and other vetted national products into a random forests model to classify forest thins in the southeastern US. Forest Harvest Inspection Records for Virginia were used for training and validation. Models were successful separating thins from clear cuts and non-harvested pines (overall accuracy 86%, clear cut accuracy 90%, thin accuracy 83% for a simplified 10-predictor variable model).
Examination of variable importance illustrates the physical meaning behind the models. The curve fit statistics (R² or RMSE) of the NDVI, Pan, and SWIR1 harmonic curve fits, which are an indication of a departure from typical vegetation phenology caused by thinning or other disturbances, were consistently among the top predictors. The harmonic regression constant, sine and cosine from the Landsat 8 panchromatic band were also important. These describe the visible reflectance (500–680 nm) phenology over the time period at a high spatial resolution (15 m). The Loss Year from the Global Forest Change product, which is an indication of stand replacing disturbance, was also consistently among the most important variables in the classifiers.
High performance computing, such as Google Earth Engine, and analysis-ready data are important for this approach. This work has importance for quantification of actively managed forests in a region of the world where production forestry is the dominant land disturbance signal and a significant economic engine.
... Land surface phenology (LSP) deals with the timing of vegetated land surface dynamics as observed by satellite remote sensors, particularly at spatial resolutions and extents relevant to meteorological processes in the atmospheric boundary layer [20,21]. LSP plays an important role in monitoring cropland dynamics. ...
... For the AMSR-E data, 8-day moving average filter was applied to the daily data to minimize data gaps due to orbit and swath width. Growing degree-day (GDD) is the daily thermal-time increment above a certain threshold (base temperature) for plant growth [20,41,45], and accumulated growing degree-days (AGDD) are the simple summation of heat units throughout the annual observation period [21,65,66]; that is, "the passage of days is weighted by the quantity of growing degrees occurring that day, with zero (but not negative) degrees being a permissible weight" [67]. McMaster [68] concluded that the base temperature of 0 °C is a robust and sufficient threshold for wheat phenology, which is the main crop in our study area. ...
... To characterize the seasonal progression of thermal time, we fitted the GDDs as a convex quadratic (CxQ) function of AGDD. The CxQ model have been successfully applied in temperate herbaceous vegetation and boreal ecosystems [21,47,65]. ...
We demonstrate the synergistic use of surface air temperature retrieved from AMSR-E (Advanced Microwave Scanning Radiometer on Earth observing satellite) and two vegetation indices (VIs) from the shorter wavelengths of MODIS (MODerate resolution Imaging Spectroradiometer) to characterize cropland phenology in the major grain production areas of Northern Eurasia from 2003–2010. We selected 49 AMSR-E pixels across Ukraine, Russia, and Kazakhstan, based on MODIS land cover percentage data. AMSR-E air temperature growing degree-days (GDD) captures the weekly, monthly, and seasonal oscillations, and well correlated with station GDD. A convex quadratic (CxQ) model that linked thermal time measured as growing degree-days to accumulated growing degree-days (AGDD) was fitted to each pixel’s time series yielding high coefficients of determination (0.88 ≤ r2 ≤ 0.98). Deviations of observed GDD from the CxQ model predicted GDD by site corresponded to peak VI for negative residuals (period of higher latent heat flux) and low VI at beginning and end of growing season for positive residuals (periods of higher sensible heat flux). Modeled thermal time to peak, i.e., AGDD at peak GDD, showed a strong inverse linear trend with respect to latitude with r2 of 0.92 for Russia and Kazakhstan and 0.81 for Ukraine. MODIS VIs tracked similar seasonal responses in time and space and were highly correlated across the growing season with r2 > 0.95. Sites at lower latitude (≤49°N) that grow winter and spring grains showed either a bimodal growing season or a shorter unimodal winter growing season with substantial inter-annual variability, whereas sites at higher latitude (≥56°N) where spring grains are cultivated exhibited shorter, unimodal growing seasons. Sites between these extremes exhibited longer unimodal growing seasons. At some sites there were shifts between unimodal and bimodal patterns over the study period. Regional heat waves that devastated grain production in 2007 in Ukraine and in 2010 in Russia and Kazakhstan appear clearly anomalous. Microwave based surface air temperature data holds great promise to extend to parts of the planet where the land surface is frequently obscured by clouds, smoke, or aerosols, and where routine meteorological observations are sparse or absent.
... In this study, comparisons with in-situ trailcam observations have helped us quantify uncertainties in growing-season phenology resulting from different curve-fitting methods that range from − 8.9 to + 14.6 days. Though this is a smaller range than found in previous studies (de Beurs and Henebry, 2010;Schwartz and Hanes, 2010;White et al., 2014), large uncertainties in satellite-derived phenology may have significant ramifications in the modelling of vegetation processes, quantifying nutrient cycles, and linking wildlife patterns to vegetation dynamics. Quantification of these uncertainties is important given that large areas of Earth lack in-situ data necessary to facilitate such comparisons. ...
... At smaller scales, as in wildlife ecology, such uncertainty can affect predictions of the availability of food resources (Bater et al., 2011), leading to large variability in population modeling outcomes. However, there remains a general lack of consensus on a preferred approach to satellite-based phenological analysis that could effectively reduce such uncertainties at all scales of application (Atkinson et al., 2012;de Beurs and Henebry, 2010;White et al., 2014White et al., , 2009. ...
Repeat digital photography at or near ground-level is a proven and efficient approach for tracking plant phenology. Here, we explored the potential to monitor phenology using the Snapshot Wisconsin (SW) trail camera network, a citizen science program. Using three curve-fitting methods for characterizing phenological transition dates, we assessed the phenological offset between understory vegetation and the overstory canopy in the trailcam observations and compared variations in derived phenology over the different spatial scales represented by trailcams (~20-50 m), Harmonized Landsat and Sentinel-2 (HLS, 30 m), and Moderate Resolution Imaging Spectroradiometer (MODIS, 500 m). Our results showed that the apparent phenological offset between understory and overstory vegetation differed among forest types: in broadleaf deciduous forests, understory vegetation had an earlier start-of-spring (SOS) and later end-of-autumn (EOA) than the overstory canopy; in mixed forests, the understory showed an earlier SOS than the overstory, but no significant difference in EOA; in evergreen conifer forests, neither SOS nor EOA differed significantly between the understory and overstory. We found moderate correlations (0.25 ≤ r ≤ 0.57) between trailcam-and satellite-derived phenological dates. Moreover, those derived dates varied significantly among the applied curve-fitting methods: total growing season length (from SOS to EOA) could be 19 days longer for a threshold-based method than for a logistic curve-fitting method (our reference model), but 17 days shorter than the logistic method when using a piecewise-continuous method based on fitted sine curves. Despite the spatial limitations of trailcams for characterizing phenology on landscape and regional scales, trailcam networks have considerable potential for informing local phenological studies and disentangling the many drivers of phenology that can remain undetected from the satellite perspective.
... These conditions encompass a range of factors, including species presence, plant ages, and the diversity of canopy layers present (D'Odorico et al., 2015). Land surface phenology (LSP) is the study of vegetation phenology from satellite-derived vegetation indices or satellite-derived biophysical variables (de Beurs and Henebry, 2004;Helman, 2018), which can be used as proxy to obtain specific phenological metrics (phenometrics) such as the start, peak and end of the growing season (SOS, POS and EOS, respectively) as well as statistical metrics of time series, such as the maximum, minimum or amplitude of the growing season (Xu et al., 2016). Phenometrics are widely used for land cover mapping (Wessels et al., 2011;Yan et al., 2015), tree species discrimination (Chuine and Beaubien, 2001;Fassnacht et al., 2016), analysis of land cover dynamics and climate change (Friedl et al., 2014;Jin et al., 2019;Piao et al., 2019;Prasad et al., 2007;Silveira et al., 2021), habitat quality characterization (Ganzhorn et al., 2011;Weber et al., 2018Weber et al., , 2018, and habitat suitability mapping (Hoagland et al., 2018), among other applications. ...
... Some bias can also be introduced due to vegetation changes caused by diseases or plant stress (Vina et al., 2004) or by specific grazing and agricultural practices (Hall-Beyer, 2003;Wardlow et al., 2006). Lastly, the statistical methods to determine the start and end of the growing season also have limitations, and linking remotely sensed observations with field collected data is one of the major challenges inherent in remotely-sensed phenology studies (de Beurs and Henebry, 2004). ...
Over the course of a year, vegetation and temperature have strong phenological and seasonal patterns, respectively , and many species have adapted to these patterns. High inter-annual variability in the phenology of vegetation and in the seasonality of temperature pose a threat for biodiversity. However, areas with high spatial variability likely have higher ecological resilience where inter-annual variability is high, because spatial variability indicates presence of a range of resources, microclimatic refugia, and habitat conditions. The integration of inter-annual and spatial variability is thus important for biodiversity conservation. Areas where spatial variability is low and inter-annual variability is high are likely to limit resilience to disturbance. In contrast, areas of high spatial variability may be high priority candidates for protection. Our goal was to develop spatio-temporal remotely sensed indices to identify hotspots of biodiversity conservation concern. We generated indices that capture the inter-annual and spatial variability of vegetation greenness and land surface temperature and integrated them to identify areas of high, medium, and low biodiversity conservation concern. We applied our method in Argentina (2.8 million km 2), a country with a wide range of climates and biomes. To generate the inter-annual variability indices, we analyzed MODIS Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) time series from 2001 to 2018, fitted curves to obtain annual phenological and seasonal metrics, and calculated their inter-annual variability. To generate the spatial variability indices, we calculated standard deviation image texture of Landsat 8 EVI and LST. When we integrated our inter-annual and spatial variability indices, areas in the northeast and parts of southern Argentina were the hotspots of highest conservation concern. High inter-annual variability poses a threat in these areas, because spatial variability is low. These are areas where management efforts could be valuable. In contrast, areas in the northwest and central-west are where protection should be strongly considered because the high spatial variability may confer resilience to disturbance , due to the variety of conditions and resources within close proximity. We developed remotely sensed indices to identify hotspots of high and low conservation concern at scales relevant to biodiversity conservation, = which can be used to target management actions in order to minimize biodiversity loss.
... Remote sensing presents a wider view of the land surface, and in any given satellite-observed "forest" pixel there may be several species of trees, undergrowth, bare soil, roads and other development, and any number of additional land components, including open water. The satellite perspective thus provides a view not on the phenology of individual trees, even at Landsat resolution (30-m pixels), but on "land surface phenology" [de Beurs and Henebry, 2010] as it is viewed in total over an area. This is especially the case for coarser multispectral remote sensing products: ~8-km pixels from AVHRR [Buitenwerf et al., 2015;Zhao et al., 2015]; 250-, 500-, and 1000-m pixels from MODIS [Huete et al., 2002;Friedl et al., 2010;Ganguly et al., 2010]; 1-km pixels from SPOT VEGETATION [Ivits et al., 2013]. ...
Landsat has a history of use in the diagnosis of land surface phenology, vegetation disturbance, and their impacts on numerous forest biological processes. Studies have connected remote sensing-based phenology to surface climatological patterns, often using average temperatures and derived growing degree day accumulations. I present a detailed examination of remotely sensed forest phenology in the region of western Lake Superior, USA, based on a comprehensive climatological assessment and 1984-2013 Landsat imagery. I use this climatology to explain both the mean annual land surface phenological cycle and its interannual variability in temperate mixed forests. I assess long-term climatological means, trends, and interannual variability for the study period using available weather station data, focusing on numerous basic and derived climate indicators: seasonal and annual temperature and precipitation, the traditionally defined frost-free growing season, and a newly defined metric of the climatological growing season: the warm-season plateau in accumulated chilling days. Results indicate +0.56°C regional warming during the 30-year study period, with cooler springs (–1.26°C) and significant autumn warming (+1.54°C). The duration of the climatological growing season has increased +0.27 days/y, extending primarily into autumn. Summer precipitation in my study area declined by an average –0.34 cm/y, potentially leading to moisture stress that can impair vegetation carbon uptake rates and can render the forest more vulnerable to disturbance. Many changes in temperature, precipitation, and climatological growing season are most prominent in locations where Lake Superior exerts a strong hydroclimatological influence, especially the Minnesota shoreline and in forest areas downwind (southeast) of the lake. I then develop and demonstrate a novel phenoclimatological modeling method, applied over five Landsat footprints across my study area, that combines (1) diagnosis of the mean phenological cycle from remote sensing observations with (2) a partial-least-squares regression (PLSR) approach to modeling vegetation index residuals using these numerous meteorological and climatological observations. While the mean phenology often used to inform land surface models in meteorological and climate modeling systems may explain 50-70% of the observed phenological variability, this mixed modeling approach can explain more than 90% of the variability in phenological observations based on long-term Landsat records for this region.
... Fixed VI thresholds were tested and compared to other extraction methods, sometimes with results more related to ground phenology respective to the others (Studer et al., 2007), but a universally applicable VI threshold has not yet been recognized. A review is available in de Beurs and Henebry (2010b). In addition to the fact that there is no clear consensus on the most efficient extraction algorithms, there are no conclusions that emerge clearly from previous studies regarding the best performing VIs and the uncertainty of satellite-based estimates of phenological dates related to temporal resolution used in time-series composite data. ...
Monitoring forest phenology allows us to study the effects of climate change on vegetated land surfaces. Daily and composite time series (TS) of several vegetation indices (VIs) from MODerate resolution Imaging Spectroradiometer (MODIS) data have been widely used in scientific works for phenological studies since the beginning of the MODIS mission. The objective of this work was to use MODIS data to find the best VI/TS combination to estimate start-of-season (SOS) and end-of-season (EOS) dates across 50 temperate deciduous forests. Our research used as inputs 2001–2012 daily reflectance from MOD09GQ/MOD09GA products and 16-day composite VIs from the MOD13Q1 dataset. The 50 pixels centered on the 50 forest plots were extracted from the above-mentioned MODIS imagery; we then generated 5 different types of TS (1 daily from MOD09 and 4 composite from MOD13Q1) and used all of them to implement 6 VIs, obtaining 30 VI/TS combinations. SOS and EOS estimates were determined for each pixel/year and each VI/TS combination. SOS/EOS estimations were then validated against ground phenological observations. Results showed that, in our test areas, composite TS, if actual acquisition date is considered, performed mostly better than daily TS. EVI, WDRVI0.20 and NDVI were more suitable to SOS estimation, while WDRVI0.05 and EVI were more convenient in estimating early and advanced EOS, respectively.
... To derive spatially exhaustive information, satellite remote sensing data that trace seasonal changes in the spectral signature of vegetation photosynthetic activity have increasingly been used during the last decade [9,10]. Such remote sensing-based analyses are referred to as land surface phenology (LSP) [11][12][13][14][15][16][17][18]. LSP observations provide a spatially integrative view of continuous biophysical canopy properties at coarse scales instead of plant-specific phenological stages. ...
... Because cloud and snow cover can significantly affect NDVI or Enhanced Vegetation Index (EVI) values, data smoothing is important to reduce signal noise caused by clouds and snow in time-series remote sensing data[2,16,23]. Phenological parameters are then estimated by using derivatives of the time-series data or applying user-defined thresholds (e.g., 50% of seasonal amplitude)[20,[24][25][26]. Among the various methods or tools, TIMESAT is one of the most commonly used packages because it provides an integrated framework for data smoothing and phenological parameter estimation[27]. ...
Understanding crop phenology is fundamental to agricultural production, management, planning, and decision-making. This study used 250 m 16-day Moderate Resolution Imaging Spectroradiometer (MODIS) Enhanced Vegetation Index (EVI) time-series data to detect crop phenology across the Midwestern United States, 2007–2015. Key crop phenology metrics, start of season (SOS) and end of season (EOS), were estimated for corn and soybean. For such a large study region, we found that MODIS-estimated SOS and EOS values were highly dependent on the nature of input time-series data, analytical methods, and threshold values chosen for crop phenology detection. With the entire sequence of MODIS EVI time-series data as input, SOS values were inconsistent compared to crop emergent dates from the United States Department of Agriculture (USDA) Crop Progress Reports (CPR). However, when we removed winter EVI images from the time-series data to reduce impacts of snow cover, we obtained much more consistent SOS estimation. Various threshold values (10 to 50% of seasonal EVI amplitude) were applied to derive SOS values. For both corn’s and soybean’s SOS estimation, a threshold value of 25% generated the best overall agreement with the CPR crop emergent dates. Root-mean-square error (RMSE) values were 4.81 and 5.30 days for corn and soybean, respectively. For corn’s EOS estimation, a threshold value of 40% led to a high R2 value of 0.82 and RMSE value of 5.16 days. We further examined spatial patterns of SOS and EOS for both crops—SOS for corn displayed a clear south-north gradient; the southern portion of the Midwest US has earlier SOS and EOS dates.
... Based on our extensive prior research using vegetation index time series to measure land surface phenology (LSP) ( de Beurs and Henebry 2010;Henebry and Beurs 2013), we summarized the MODIS image time series into a set of annual metrics. The LSP metrics that we calculated included (1) the start of the growing season (SOS), (2) the length of the growing season (LOS), (3) the AGDD at SOS, (4) the NDVI peak timing measured in AGDD, (5) the NDVI peak height, and (6) the coefficient of determination (R 2 ) for convex quadratic LSP model. ...
... Considering that DM and DA indicate different ecological implications and they are also somewhat correlated with each other, the classification experiments were performed respectively with both of the indicators or either of them as input features. Support Vector Machine (SVM) (Cortes and Vapnik, 1995) was used as the classifier in the current study. Considering that the objective of this study is to identify C. microphylla from grass species in the region, the classification was not implemented on other land cover types (e.g., barren soil, urban, water, etc.). ...
... over time is used as a leading predictor for phenological events of plants (see for example Chuine, 2000; Beurs and Henebry, 2010). ...
Phenology, the study of the association between biological development stages and variation in climate, has greatly increased in importance due to concerns arising from climate change. This paper presents a general stochastic approach to the modeling of the relationship between phenological events and climate variables, and gives a prediction method on this approach to provide full predictive distributions for future events. The proposed methods are then applied to the modeling and prediction of the bloom dates of six highly-valued fruit crops. In particular, we use our approach to explore how the bloom dates are related to the accumulation of growing degree days, to provide a sensible estimate of an important parameter Tbase in phenological study, and to assess the prediction of bloom dates with a leave-one-out procedure. Most importantly, the impact of future climate change on bloom dates is studies with temperature outputs from well-established coupled global climate models under a high greenhouse gases scenario.
... This lack of pronounced seasonality trends confounded efforts to retrieve variables such as growing season length using standard assessment techniques. We therefore limited our analysis to a single phenology metric: the timing of maximum greenness, i.e., the day of year (DOY) of the maxi- mum annual NDVI or EVI value ( de Beurs & Henebry, 2010). The determi- nation of the timing of the peak vegetation greenness in each 30 m STARFM is a straightforward extraction of the date on which the highest index value was recorded. ...
The patchy and heterogeneous arrangement of vegetation in dryland areas complicates the extraction of phenological signals using existing remote sensing data. This study examined whether the phenological analysis of a range of dryland land cover classes would benefit from the availability of synthetic images at Landsat spatial resolution and MODIS time intervals. We assembled a series of 500 m MODIS and Landsat-5 TM datasets from April to November, 2005–2009, over a study site in central Arizona that encompasses diverse dryland vegetation classes along an elevation gradient of 2000 m. We applied the spatial and temporal adaptive reflectance fusion model (STARFM) to each MODIS image to create a time series of synthetic images at 30 m resolution. We subjected a subset of the synthetic imagery to a pixel-based regression analysis with temporally coincident Landsat images to analyze the effect of the underlying vegetation class on the accuracy of the STARFM results. To evaluate the usefulness of the increased spatial resolution compared to a MODIS product, we analyzed the variability of the date of peak greenness values of all 30 m pixels within unmixed MODIS pixels. Finally, we examined differences in the temporal distributions of peak greenness extracted from both the Enhanced Vegetation Index (EVI) and Normalized Difference Vegetation Index (NDVI) synthetic imagery time series. Our results indicate that characteristics of the vegetation classes strongly influence STARFM algorithm performance, with Pearson correlation coefficient values ranging from 0.72 to 0.96 depending on the Landsat band and the land cover class. Responses in the near-infrared (NIR) spectrum yielded the lowest correlations, particularly for the Ponderosa Pine class. The phenological variability exhibited by each land cover class was dependent on the precipitation patterns of each growing season, but was sufficiently high to make the application of STARFM imagery at this scale uniformly beneficial. The peak greenness dates extracted from EVI and NDVI time series were temporally synchronized for the Grassland class but diverged for the classes of mixed woody and herbaceous vegetation types.
Investigating vegetation changes, especially plant phenology, can yield valuable information about global warming and climate change. Time series satellite observations and remote sensing methods offer a great source of information on distinctions and changing aspects of vegetation. The current study aimed to determine the trend and rate of changes in some phenological components of Iran’s vegetation. In this regard, the current study employed the daily NDVI (Normalized Difference Vegetation Index) product of the AVHRR sensor with a spatial resolution of 0.05° × 0.05°, named AVH13C1. Then, using the HANTS algorithm, images of amplitude zero, annual amplitude, and annual phase were prepared annually from 1982 to 2019. Using TIMESAT software, the starting, end, and length of time of growing season were calculated for each pixel time series to prepare annual maps. The Mann–Kendall statistical test was used to investigate the significance of changes during the study period. On average in the entire area of Iran, the annual phase was declining with a trend of −0.6° per year, and the time for the start and end of the season was declining by −0.3 and −0.65 days per year, respectively. Major changes were noticed in the northeast, west, and northwest regions of Iran, where the annual phase declined with a trend of −0.9° per year. Since the annual growth cycle of the plant (equivalent to 356 days) was in the form of a sinusoidal signal, and the angular changes in the sine wave were between zero and 360°, each degree of change was equivalent to 1.01 days per year. Therefore, the reduction in the annual phase by −0.9 degrees almost means a change in the time (due to the earlier negative start phase) of the start of the annual growth signal by −0.9 days per year. The time of the start and end of the growing season declined by −0.6 and −1.33 days per year, respectively. The reduction in annual phase and differences in time of the starting season from 1982 to 2019 indicate the acceleration and earlier initiation of various phenological processes in the area.
Accurate and continuous crop phenology information at a regional scale is important for agronomic management and yield estimation. However, detecting continuous crop phenology remains challenging due to the low sensitivity of remote sensing signals to certain phenological stages and the limited of availability remote sensing images. Therefore, this study developed a layered strategy to detect continuous crop phenology. First, a novel Phenology Separability Index (PSI) is established to select features from the Gaussian probability density distribution. PSI quantifies the capability of optical vegetation indexes (VIs), Synthetic Aperture Radar (SAR) signals, and meteorological factors to distinguish between various phenological stages. Then, the multi-temporal sample is established to enhance training sample representation and quantity. Finally, a random forest model is trained using features extracted from multi-temporal samples to improve detection accuracy. This model effectively reduces phenological stage confusion due to redundant features and limited samples. In addition, this study validated its extensibility by mapping crop phenology in the cities of Acheng, Zhaozhou, Lishu, and Buxin and assessed its uncertainty using the Sobol approach. Results indicated that growing degree day has the highest separability among meteorological factors, surpassing both SAR singles and optical VIs. Moreover, the proposed layered strategy was robust, explaining 96% of spatial variation in crop phenology at the regional scale. The accuracy of the layered strategy method (total RMSE = 8.74 days) surpassed that of the multi-temporal sample method (total RMSE = 15.76 days) and the traditional method with a single-temporal sample (total RMSE = 17.21 days). In addition, this study indicated that optical VIs are prone to confuse with the early or late phenological stage of corn, whereas SAR singles are highly sensitive to jointing date.
In today’s world of increasing food insecurity due to more frequent and extreme events (droughts, floods), a comprehensive understanding of global cropland dynamics is critically needed. Land surface parameters derived from the passive microwave Advanced Microwave Scanning Radiometer on EOS (AMSR-E) and AMSR2 data enable monitoring of cropland dynamics and they can complement visible to near infrared (VNIR) and thermal infrared (TIR) data. Passive microwave data are less sensitive to atmospheric effects, cloud contamination, and solar illumination constraints resulting in finer temporal resolution suitable to track the temporal progression of cropland cover development compared to the VNIR data that has coarser temporal resolution due to compositing to lessen the atmospheric effects. Both VNIR and TIR data have moderate to fine spatial resolution compared to passive microwaves, due to the faint microwave flux from the planetary surface. I used AMSR, MODIS, TRMM, and simplified surface energy balance (SSEB) data to study cropland dynamics from 2003-2015 in North Dakota, USA, the Canadian Prairie Provinces, Northern Eurasia, and East Africa: a contrast between crop exporting regions and a food insecure region. Croplands in the temperate region are better studied compared to that of the tropics. The objective of this research was to characterize cropland dynamics in the tropics based on the knowledge gained about the microwave products in the temperate croplands. This study also aimed at assessing the utility of passive microwave data for cropland dynamics study, especially for tropical cropland regions that are often cloud-obscured during the growing season and have sparse in situ data networks. Using MODIS land cover data, I identified 162 AMSR grid cells (25km*25km=625km2) dominated by croplands within the study regions. To fit the passive microwave time series data to environmental forcings, I used the convex quadratic (CxQ) model fit that has been successfully applied with the VNIR and TIR data to herbaceous vegetation in temperate and boreal ecoregions. Land surface dynamics in the thermally-limited temperate croplands were characterized as a function of temperature; whereas, a function of moisture to model land surface dynamics in the tropical croplands. In the temperate croplands, growing degree-day (GDD), NDVI, and vegetation optical depth (VOD) were modeled as a convex quadratic function of accumulated GDD (AGDD) derived from AMSR air temperature data, yielding high coefficients of determination (0.88� r2 �0.98). Deviations of GDD from the long term average CxQ model by site corresponded to peak VI producing negative residuals (arising from higher latent heat flux) and low VI at beginning and end of growing season producing positive residuals (arising from higher sensible heat flux). In Northern Eurasia, sites at lower latitude (44o - 48o N) that grow winter grains showed either a longer unimodal growing season or a bimodal growing season; whereas, sites at higher latitude (48o - 56o N) where spring grains are cultivated showed shorter, unimodal growing seasons. Peak VOD showed strong linear correspondence with peak greenness (NDVI) with r2>0.8, but with a one-week lag. The AMSR data were able to capture the effects of the 2010 and 2007 heat waves that devastated grain production in southwestern Russia and Northern Kazakhstan, and Ukraine, respectively, better than the MODIS data. In East African croplands, the AMSR, TRMM, and SSEB datasets modeled as a convex quadratic function of cumulative water-vapor-days displayed distinct cropland dynamics in space and time, including unimodal and bimodal growing seasons. Interannual moisture variability is at its highest at the beginning of the growing season affecting planting times of crops. Moisture time to peak from AMSR and TRMM land surface parameters displayed strong correspondence (r2 > 0.80) and logical lags among variables. Characterizing cropland dynamics based on the synergistic use of complementary remote sensing data should help to advance and improve agricultural monitoring in tropical croplands that are often associated with food insecurity.
The annual cyclic phenomena of soil surface wetness influences for instance vegetation growth, drought, flooding, and soil properties. This study presents an attempt to define metrics relevant for capturing the soil moisture dynamics from an annual series of wetness estimates derived from global Moderate-resolution imaging spectroradiometer (MODIS) images. Different algorithms for both smoothing and gap-filling the time series are tested with the results compared to in-situ data. Neither the smoothing nor the gap-filling improve the capturing of the surface wetness phenology compared to using unsmoothed time series data. The smoothing, however removes the effects of erratic rainfall events and noise, and the smoothed time series was considered more robust for identification of wetness phenology. Metrics capturing the global surface wetness phenology for 2011, extracted after smoothing using a simplified locally weighted scatterplot smoothing (LOWESS) model, are presented at a spatial resolution of 500 m for the calendar year 2011.
Satellite observations of the terrestrial biosphere cover a period of time sufficiently extended to allow the calculation of a reliable climatology. The latter is particularly relevant for studies of vegetation response to climate variability. Observations from space of the land surface are hampered by clouds at shorter wavelength and affected by water in the atmosphere in the microwave range. Both polar orbiting and geostationary satellites have a revisit frequency high enough to allow for some redundancy relative to the processes being observed, so that time series where a fraction of observations are removed and the resulting gaps filled are still very useful to monitor land surface processes. Two examples illustrate this concept in two different spectral regions: Thermal Infrared (TIR) and observations of land surface temperature to study the thermal behavior of the land surface in response to weather and climate and 37 GHz observations of the polarization difference in brightness temperature to retrieve the fractional abundance of water-saturated soil. Three applications of time series of land surface temperature are presented: (a) monitoring of spectral thermal admittance of the land surface; (b) estimation and mapping of air temperature and (c) monitoring of thermal load to assess the risk of forest fires.
Plateau vegetation is considered to be highly sensitive to climate change, especially at higher altitudes. Although the Tibetan Plateau has experienced intensive warming over the past few decades, there is much contradictory evidence regarding its phenological variations and the impact of climatic change. In this study, we explored vegetation phenology through the inflexion point-based method with the weekly 0.05° EVI2 datasets from 1982 to 2010. We observed complex spatiotemporal variations in vegetation phenology on the higher Tibetan Plateau from three aspects. From a spatial aspect, the altitudinal gradients of phenological dates, as well as their directions, varied among different altitudes over the past three decades. Compared with delaying with elevation at altitudes below 5000 m, the phenological parameters at altitudes above 5000 m significantly advanced with increasing altitudes. At higher altitudes, much stronger altitudinal gradients (slope) of phenological dates were observed in the 2000s than in the 1980s and 1990s, i.e., 2.19, 3.47, and 3.68 days' advance for start, maximum, and end dates, respectively, compared to less than 1 day's change per 100 m increase in altitude. From a temporal dynamic aspect, when analyzed at different altitudinal bands, the dynamic trends in phenological dates were generally not significant except the advancing trends in the maximum dates at altitudes above 5000 m and the delaying trend in the end dates at altitudes of 4500-5000 m in the twenty-first century. Remarkable elevation dependency was also observed at the pixel level: increasing amplitudes of phenological dynamic trends were observed at higher altitudes when obtaining their minimum around 5000 m. These spatiotemporal variations of vegetation phenology were due to combined effects from both temperature and precipitation: more abundant rainfall and greater magnitudes of dynamic trends were observed in the average daily minimum temperature (slope = 0.08 °C/year) and annual precipitation (slope = 2.17 mm/year) at higher altitudes.
Advanced Very High Resolution Radiometer (AVHRR) (8 km) Normalized Differential Vegetation Index (NDVI) data and Moderate Resolution Imaging Spectroradiometer (MODIS) (16-day, 250m) NDVI data products were considered to analyze vulnerability of Indian agriculture to rainfall variability under climate change impact studies. Predicted higher temperature and altered rainfall patterns accompanied by extreme weather events would impact vegetation growth in natural forest, open scrub, agricultural land and plantations. NDVI derived from 2-band information (Red and Near-infra Red) of multi-spectral imagery of AVHRR (1982 to 2006) and from MODIS (20002010) were analysed to understand spatial and temporal variability. Coefficient of Variation (CV) of maximum NDVI from 15-day composites for the total length of the study period was used to assess vulnerability of rain-fed agriculture and results were corroborated with the Standard Precipitation Index (SPI) rather than actual rainfall received during the study period. AVHRR time-series data helped to identify vulnerable areas at regional-scale, i.e., agro-ecological subregions (AESR) due to coarser ground resolution while MODIS data products with 250m pixel resolution helped identify vulnerability at the district level. It was estimated that over 241 Mha areas in the country may not be vulnerable to rainfall variability-induced climate change, whereas over 81.3 Mha in arid, semi-arid and dry sub-humid regions in the country may be vulnerable to extreme weather events. Study indicated that over 12.1 and 1.81 Mha of Kharif cropland would be mildly and severely vulnerable, whereas 6.86 and 0.5 Mha of Rabi cropland may be adversely affected in a similar manner. Of the remaining agricultural lands, 29.93 and 5.24 Mha would also be vulnerable to climate change in a similar manner. Studies also indicated a decrease in length of Kharif and Rabi seasons and a delay in the start of Kharif season based on preliminary findings.
This research examines the spatio-temporal trends in Advanced Very High Resolution Radiometer (AVHRR) Global Inventory Modelling and Mapping Studies (GIMMS) normalized difference vegetation index (NDVI) time series to ascribe land use change and precipitation to observed changes in land cover from 1982 to 2007 in the Mexican Yucatán Peninsula, using seasonal trend analysis (STA). In addition to discrete land cover transitions across the study region, patterns of agricultural intensification, urban expansion and afforestation in protected areas have enacted changes to the seasonal patterns of apparent greenness observed through STA greenness parameters. The results indicate that the seasonal variation in NDVI can be used to distinguish among different land cover transitions, and the primary differences among these transitions were in changes in overall greenness, peak annual greenness and the timing of the growing season. Associations between greenness trends and precipitation were weak, indicating a human-dominated system for the 26 years examined. Changes in the states of Campeche, Quintana Roo and Yucatán appear to be associated with pasture cultivation, urban expansion-extensive cultivation and urban expansion-intensive cultivation, respectively.
ResearchGate has not been able to resolve any references for this publication.