Stephanie Horion

Using earth observation-based dry season NDVI trends for assessment of changes in tree cover in the Sahel

The co-existence of trees and grasses is a defining feature of savannah ecosystems and landscapes. During recent decades, the combined effect of climate change and increased demographic pressure has led to complex vegetation changes in these ecosystems. A number of recent Earth observation (EO)-based studies reported positive changes in biological productivity in the Sahelian region in relation to increased precipitation, triggering an increased amount of herbaceous vegetation during the rainy season. However, this ‘greening of the Sahel’ may mask changes in the tree–grass composition, with a potential reduction in tree cover having important implications for the Sahelian population. Large-scale EO-based evaluation of changes in Sahelian tree cover is assessed by analysing long-term trends in dry season minimum normalized difference vegetation index (NDVImin) derived from three different satellite sensors: Système Pour l’Observation de la Terre (SPOT)-VEGETATION (VGT), Terra Moderate Resolution Imaging Spectroradiometer (MODIS), and the Advanced Very High Resolution Radiometer (AVHRR) Global Inventory Modeling and Mapping Studies (GIMMS) dataset. To evaluate the reliability of using NDVImin as a proxy for tree cover in the Sahel, two factors that could potentially influence dry season NDVImin estimates were analysed: the total biomass accumulated during the preceding growing season and the percentage of burned area observed during the dry season. Time series of dry season NDVImin derived from low-resolution satellite time series were found to be uncorrelated to dry grass residues from the preceding growing season and to seasonal fire frequency and timing over most of the Sahel (88%), suggesting that NDVImin can serve as a proxy for assessing changes in tree cover. Good agreement (R2 = 0.79) between significant NDVImin trends (p < 0.05) derived from VGT and MODIS was found. Significant positive trends in NDVImin were registered by both MODIS and VGT dry season NDVImin time series over the Western Sahel, whereas trends based on GIMMS data were negative for the greater part of the Sahel. EO-based trends were generally not confirmed at the local scale based on selected study cases, partly caused by a temporal mismatch between data sets (i.e. different periods of analysis). Analysis of desert area NDVImin trends indicates less stable values for VGT and GIMMS data as compared with MODIS. This suggests that trends in dry season NDVImin derived from VGT and GIMMS should be used with caution as an indicator for changes in tree cover, whereas the MODIS data stream shows a better potential for tree-cover change analysis in the Sahel.

Global Biogeographical Pattern of Ecosystem Functional Types Derived From Earth Observation Data

The present study classified global Ecosystem Functional Types (EFTs) derived from seasonal vegetation dynamics of the GIMMS3g NDVI time-series. Rotated Principal Component Analysis (PCA) was run on the derived phenological and productivity variables, which selected the Standing Biomass (approximation of Net Primary Productivity), the Cyclic Fraction (seasonal vegetation productivity), the Permanent Fraction (permanent surface vegetation), the Maximum Day (day of maximum vegetation development) and the Season Length (length of vegetation growing season) variables, describing 98% of the variation in global ecosystems. EFTs were created based on Isodata classification of the spatial patterns of the Principal Components and were interpreted via gradient analysis using the selected remote sensing variables and climatic constraints (radiation, temperature, and water) of vegetation growth. The association of the EFTs with existing climate and land cover classifications was demonstrated via Detrended Correspondence Analysis (DCA). The ordination indicated good description of the global environmental gradient by the EFTs, supporting the understanding of phenological and productivity dynamics of global ecosystems. Climatic constraints of vegetation growth explained 50% of variation in the phenological data along the EFTs showing that part of the variation in the global phenological gradient is not climate related but is unique to the Earth Observation derived variables. DCA demonstrated good correspondence of the EFTs to global climate and also to land use classification. The results show the great potential of Earth Observation derived parameters for the quantification of ecosystem functional dynamics and for providing reference status information for future assessments of ecosystem changes.

Environmental change in the Sahel: reconciling contrasting evidence and interpretations

The Sahel has been the object of intensive international research since the drought of the early 1970s. A considerable part of the research has focused on environmental change in general and land degradation, land cover change and climate change in particular. Rich and diverse insights from many different scientific disciplines about these three domains have been put forward. One intriguing feature is that an agreement on the overall trends of environmental change does not appear to emerge: questions such as whether the Sahel is greening, cropland is encroaching on rangelands, drought persists remain contested in the scientific literature, and arguments are supported by contrasting empirical evidence. The paper explores the generic reasons behind this situation in a systematic manner. We distinguish between divergences in interpretations emerging from (1) conceptualizations, definitions and choice of indicators, (2) biases, for example, related to selection of study sites, methodological choices, measurement accuracy, perceptions among interlocutors, and selection of temporal and spatial scales of analysis. The analysis of the root causes for different interpretations suggests that differences in findings could often be considered as complementary insights rather than mutually exclusive. This will have implications for the ways in which scientific results can be expected to support regional environmental policies and contribute to knowledge production.

Very high CO2 exchange fluxes at the peak of the rainy season in a West African grazed semi-arid savanna ecosystem

Abstract Africa is a sink of carbon, but there are large gaps in our knowledge regarding the CO2 exchange fluxes for many African ecosystems. Here, we analyse multi-annual eddy covariance data of CO2 exchange fluxes for a grazed Sahelian semi-arid savanna ecosystem in Senegal, West Africa. The aim of the study is to investigate the high CO2 exchange fluxes measured at the peak of the rainy season at the Dahra field site: gross primary productivity and ecosystem respiration peaked at values up to −48 μmol CO2 m−2 s−1 and 20 μmol CO2 m−2 s−1, respectively. Possible explanations for such high fluxes include a combination of moderately dense herbaceous C4 ground vegetation, high soil nutrient availability and a grazing pressure increasing the fluxes. Even though the peak net CO2 uptake was high, the annual budget of −229 ± 7 ± 49 g C m−2 y−1 (±random errors ± systematic errors) is comparable to that of other semi-arid savanna sites due the short length of the rainy season. An inter-comparison between the open-path and a closed-path infrared sensor indicated no systematic errors related to the instrumentation. An uncertainty analysis of long-term NEE budgets indicated that corrections for air density fluctuations were the largest error source (11.3% out of 24.3% uncertainty). Soil organic carbon data indicated a substantial increase in the soil organic carbon pool for the uppermost .20 m. These findings have large implications for the perception of the carbon sink/source of Sahelian ecosystems and its response to climate change.

Global Ecosystem Response Types Derived from the Standardized Precipitation Evapotranspiration Index and FPAR3g Series

Observing trends in global ecosystem dynamics is an important first step, but attributing these trends to climate variability represents a further step in understanding Earth system changes. In the present study, we classified global Ecosystem Response Types (ERTs) based on common spatio-temporal patterns in time-series of Standardized Precipitation Evapotranspiration Index (SPEI) and FPAR3g anomalies (1982-2011) by using an extended Principal Component Analysis. The ERTs represent region specific spatio-temporal patterns of ecosystems responding to drought or ecosystems with decreasing severity in drought events as well as ecosystems where drought was not a dominant factor in a 30-year period. Highest explanatory values in the SPEI12-FPAR3g anomalies and strongest SPEI12-FPAR3g correlations were seen in the ERTs of Australia and South America whereas lowest explanatory value and lowest correlations were observed in Asia and North America. These ERTs complement traditional pixel based methods by enabling the combined assessment of the location, timing, duration, frequency and severity of climatic and vegetation anomalies with the joint assessment of wetting and drying climatic conditions. The ERTs produced here thus have potential in supporting global change studies by mapping reference conditions of long term ecosystem changes.

Assessing drivers of vegetation changes in drylands from time series of earth observation data

This chapter summarizes methods of inferring information about drivers of global dryland vegetation changes observed from remote sensing time series data covering from the 1980s until present time. Earth observation (EO) based time series of vegetation metrics, sea surface temperature (SST) (both from the AVHRR (Advanced Very High Resolution Radiometer) series of instruments) and precipitation data (blended satellite/rain gauge) are used for determining the mechanisms of observed changes. EO-based methods to better distinguish between climate and human induced (land use) vegetation changes are reviewed. The techniques presented include trend analysis based on the Rain-Use Efficiency (RUE) and the Residual Trend Analysis (RESTREND) and the methodological challenges related to the use of these. Finally, teleconnections between global sea surface temperature (SST) anomalies and dryland vegetation productivity are illustrated and the associated predictive capabilities are discussed.

Assessment of Vegetation Trends in Drylands from Time Series of Earth Observation Data

This chapter summarizes approaches to the detection of dryland vegetation change and methods for observing spatio-temporal trends from space. An overview of suitable long-term Earth Observation (EO) based datasets for assessment of global dryland vegetation trends is provided and a status map of contemporary greening and browning trends for global drylands is presented. The vegetation metrics suitable for per-pixel temporal trend analysis is discussed, including seasonal parameterisation and the appropriate choice of trend indicators. Recent methods designed to overcome assumptions of long-term linearity in time series analysis (Breaks For Additive Season and Trend(BFAST)) are discussed. Finally, the importance of the spatial scale when performing temporal trend analysis is introduced and a method for image downscaling (Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM)) is presented.

Massively-parallel break detection for satellite data

The field of remote sensing is nowadays faced with huge amounts of data. While this offers a variety of exciting research opportunities, it also yields significant challenges regarding both computation time and space requirements. In practice, the sheer data volumes render existing approaches too slow for processing and analyzing all the available data. This work aims at accelerating BFAST, one of the state-of-the-art methods for break detection given satellite image time series. In particular, we propose a massively-parallel implementation for BFAST that can effectively make use of modern parallel compute devices such as GPUs. Our experimental evaluation shows that the proposed GPU implementation is up to four orders of magnitude faster than the existing publicly available implementation and up to ten times faster than a corresponding multi-threaded CPU execution. The dramatic decrease in running time renders the analysis of significantly larger datasets possible in seconds or minutes instead of hours or days. We demonstrate the practical benefits of our implementations given both artificial and real datasets.