Identification of environmental anomaly hot spots in West Africa from time series of NDVI and rainfall
Introduction
The West African Sahelian belt witnessed a dramatic food crisis in the 1970s–1980s, which was caused by prolonged drought and was locally exacerbated by socio economic instability. The progressive environmental degradation and food shortage during those years have stimulated the international attention towards Sahel, which materialised in the United Nations Conference on Desertification (UNCOD), organised in 1977 and 1978. The conference has launched a debate, which is still open, on the causes of drought and its impact on food security for the local population. The scientific interest focused on (a) seeking climatic or anthropogenic driving factors for the extended drought and (b) assessing its impact on the environment and the agro-ecosystems (Herrmann et al., 2005). For what concerns the interpretation of the causes of drought, there are two different schools of thought: one adheres to the hypothesis of a human-induced drought (Ibrahim, 1978) and the other one to the hypothesis that drought was caused by climatic variability on which human have a minor role (Nicholson et al., 1998, Olsson et al., 2005). These opposing hypotheses derived from the difficulty in identifying root causes of long term climate fluctuations. Some authors point to the relationship between rainfall and the anomalies of the ocean surface temperature, which is a phenomenon similar to El Niño occurring in the Pacific Ocean (Nicholson and Grist, 2001, Giannini et al., 2003), whereas others point to the relation between the monsoon season and the cold temperatures in the southern part of the Atlantic (Brooks, 2004).
Similarly, investigations on the consequences and the impact of drought have seen different interpretations. In the late 1980s, Lamprey (1988) stated that the Sahelian region was witnessing an “irreversible and progressive desertification”, but more recent studies based on satellite data show an increase in vegetation cover since the mid-1980s (Eklundh and Olsson, 2003), which is correlated to an increase in large-scale precipitation (Herrmann et al., 2005, Heumann et al., 2007). This phenomenon, commonly referred to as “greening” or “re-greening” has been interpreted as a recovery from the great Sahelian drought of the 1960s and 1970s (Anyamba and Tucker, 2005, Olsson et al., 2005, Giannini et al., 2008). Although the extent of “greening” has not yet been established with ground data, it questions previous theories about the “irreversible damage inflicted on the ecosystem Sahel” (Dregne, 1986, Middleton and Thomas, 1997). New technologies, like remote sensing, have allowed the scientific community to further analyse the climatic phenomena in their spatial and temporal dimensions and their relation with vegetation dynamics. Herrmann et al. (2005) compared trends of rainfall and Normalized Difference Vegetation Index (NDVI), a widely used proxy of vegetation productivity, and found a general increase of vegetation abundance in the Sahel between 1982 and 2003, thus refusing the presence of a large-scale irreversible desertification. The same authors pointed out that, locally, the NDVI-rainfall relation differs from the expected trend, thus suggesting that other factors have driven vegetation regrowth such as the improvement of agricultural practises. Huber et al. (2011) highlighted that differences in vegetation patterns are also related to water availability into the soil and not only to rainfall quantity. Some areas resulted especially interesting, because “…the NDVI trends […] are opposed to those observed in the rainfall time series”, underlining local anomalous response of vegetation to changes in the precipitation amount. In particular, the authors found that Western Sahel (Senegal) shows a significant greening trend despite stable rainfall while Eastern Sahel (Sudan) shows an opposite behaviour.
Several studies based on time series of satellite data and, where available, on ground data, confirmed the dynamics of the Sahelian ecosystem and their tendency to change with time, although there is no general agreement on the entity of such changes and on the major driving factors (Herrmann et al., 2005). Capecchi et al. (2008) verified the positive effect of increased rainfall on crop production and especially on the production of millet between 1986 and 2000. However, due to the lack of detailed information, it is hard to assess the real extent of the “greening” phenomenon in terms of ecosystem production. For example, the herbaceous savanna of the Sahel has shown a strong greening when analysed with satellite data, however in field information indicates that locally the vegetation composition of these systems is changing. FAO indicates that rangelands have shown a phytosociological response of grassland to prolonged droughts: loss of perennial herbaceous species as Cymbopogon giganteus and Aristida sieberiana and the expansion of species with shorter phenological cycle (thus more resistant to drought) such as Cenchrus biflorus and Aristida funiculata. In conclusion, a global greening could be due not only to ecosystem recovery but also to changes of herbaceous species composition. Some of these species are also palatable to animals but their low productivity (about 400 kg/ha compared with 2000 kg/ha of perennial herbaceous species) affects agro-pastoral activities (Geesing and Djibo, 2001). If a general regional “greening” is highlighted by time series of satellite data, it does not necessarily imply a recovery of the carrying capacity of these ecosystems. Several authors (Herrmann et al., 2005, Seaquist et al., 2009, Fensholt and Rasmussen, 2011) state that it is not possible to exclude local processes of desertification and degradation and therefore it is necessary to conduct analyses at a finer spatial scale to highlight possible anomalous situations.
This study aims to identify hot spots of environmental anomalies where vegetation dynamics are not fully explained by rainfall variability suggesting that other factors are interfering with the phenomena. We analysed the correlation between SPOT–VGT NDVI and FEWS–RFE (Famine Early Warning Systems Network – Rain Fall Estimates) over the period 1998–2010 and we studied the temporal dynamics of NDVI to identify areas where vegetation presents a statistically significant change over the 13-year period. The spatial distribution of these hot spots of environmental anomaly is analysed as a function of the land cover and it is discussed with other ancillary information. Finally, a set of multi-temporal Landsat TM images, with higher spatial resolution (30 m) compare to 1 km SPOT–VGT, is used to evaluate the method and to interpret locally the results.
Section snippets
Study area
The study area covers sub-Saharan West Africa between 4°N and 18°N of latitude and −18°E and 25°E of longitude (Fig. 1). The area of interest is the Sahel: a transition zone, which is marked by a steep north–south gradient in mean annual rainfall, and is located between the Sahara desert and the sub-humid tropical savannas. The climate of the Sahel is characterised by a strong seasonality and variability of rainfall, with a long dry season in winter and a short summer rainy period. The effect
Satellite data time series
Satellite data are particularly suitable, and often represent the only available data source, for the analysis of large scale phenomena such as the vegetation dynamics of the sub-Saharan West Africa. The analysis of Sahelian vegetation dynamics showed that the boundaries of this area may change even 150 km a year following rainfall pattern (Tucker and Nicholson, 1999). Thus, the analysis of desertification and land degradation processes can be carried out only with long-term spatial analysis at
Pre-processing and standardization of the time series
The autocorrelation (or serial correlation) of temporal data is a well know problem that can strongly bias the detection of trend in NDVI time series analysis (Eklundh, 1998, de Beurs and Henebry, 2005). de Jong and de Bruin (2012) underlined that the managing of time series has to be done with care in order to avoid spurious results; the serial correlation affects the null hypothesis of a trend test producing false rejections that leads to an overestimation of temporal trend detection even
Spatial correlation and temporal dynamics analysis
The autocorrelation analysis conducted on ∑NDVI time series, before performing the subsequent statistical analysis, indicated that, when lag-1 to lag-3 condition is considered, all time series resulted temporally independent (i.e. no autocorrelation). Spatial distribution of the Kendal tau rank correlation coefficient between ∑NDVI and Y-Rain is shown in Fig. 2a and areas where correlation is significant are presented in Fig. 2b. The Sahelian belt shows high positive correlation (blue regions)
Conclusions
This study performs a preliminary screening at regional scale using time series of 1 km satellite data in order to identify local anomalous environmental hot spots where changes in vegetation cover are likely to occur. The information that is obtained by this study is particularly relevant for strategic orientation of environmental resource management, not only at national level, but also for regional and continental institutions such as ECOWAS (Economic Community of West African States) and the
Acknowledgements
This research was conducted as a part of the GEOLAND-2 project (www.gmes-geoland.info), which is a Collaborative Project (2008–2012) funded by the European Union under the 7th Framework Programme (Project Number 218795). The authors would like to thank the anonymous reviewers for their valuable comments and suggestions to improve the research results presented in this paper.
References (74)
- et al.
Analysis of Sahelian vegetation dynamics using NOAA-AVHRR NDVI data from 1981–2003
Journal of Arid Environments
(2005) - et al.
A 20-year study of NDVI variability over the Northeast Region of Brazil
Journal of Arid Environments
(2006) - et al.
Monitoring 25 years of land cover change dynamics in Africa: A sample based remote sensing approach
Applied Geography
(2009) - et al.
Analysis of monotonic greening and browning trends from global NDVI time-series
Remote Sensing of Environment
(2011) - et al.
Analysis of trends in the Sahelian “rain-use efficiency” using GIMMS NDVI, RFE and GPCP rainfall data
Remote Sensing of Environment
(2011) - et al.
Evaluation of satellite based primary production modelling in the semi-arid Sahel
Remote Sensing of Environment
(2006) - et al.
A climate model-based review of drought in the Sahel: Desertification, the re-greening and climate change
Global and Planetary Change
(2008) - et al.
Regional desertification: a global synthesis
Global and Planetary Change
(2008) - et al.
Recent trends in vegetation dynamics in the African Sahel and their relationship to climate
Global Environmental Change Part A
(2005) - et al.
AVHRR derived phenological change in the Sahel and Soudan, Africa, 1982–2005
Remote Sensing of Environment
(2007)
Water availability as the driver of vegetation dynamics in the African Sahel from 1982 to 2007
Global and Planetary Change
Thermal remote sensing of water under flooded vegetation: new observations of inundation patterns for the Small Lake Chad
Journal of Hydrology
Vegetation dynamics from NDVI time series analysis using the wavelet transform
Remote Sensing of Environment
Farmer adaptation, change and “crisis” in the Sahel
Global Environmental Change
Fifty years of landscape evolution in Southwestern Mauritania by means of aerial photos
Journal of Arid Environments
A recent greening of the Sahel—trends, patterns and potential causes
Journal of Arid Environments
Mapping and monitoring net primary productivity with AVHRR NDVI time-series: statistical equivalence of cumulative vegetation indices
ISPRS Journal of Photogrammetry and Remote Sensing
A spatio-temporal analysis of forage availability and grazing and excretion behaviour of herded and free grazing cattle, sheep and goats in Western Niger
Agriculture, Ecosystems and Environment
Classification and change detection using landsat tm data: when and how to correct atmospheric effects?
Remote Sensing of Environment
Satellite remote sensing of total herbaceous biomass production in the senegalese Sahel: 1980–1984
Remote Sensing of Environment
Has the sahel a future?
Journal of Arid Environments
Proxy global assessment of land degradation
Soil Use and Management
The future is an ancient lake: traditional knowledge, biodiversity and genetic resources for food and agriculture in Lake Chad Basin ecosystems
Analysis of NDVI trends and their climatic origin in the Sahel 1986–2000
Geocarto International
Map of the administrative units
Land surface phenology and temperature variation in the International Geosphere-Biosphere Program high-latitude transects
Global Change Biology
A land surface phenology assessment of the northern polar regions using MODIS reflectance time series
Canadian Journal of Remote Sensing
Linear trends in seasonal vegetation time series and the modifiable temporal unit problem
Biogeosciences
Desertification of arid lands
Estimating relations between AVHRR NDVI and rainfall in East Africa at 10-day and monthly time scales
International Journal of Remote Sensing
Vegetation index trends for the African Sahel 1982–1999
Geophysical Research Letters
Cited by (60)
Analysis of forest fire patterns and their relationship with climate variables in Alberta's natural subregions
2024, Ecological InformaticsA dryness index TSWDI based on land surface temperature, sun-induced chlorophyll fluorescence, and water balance
2023, ISPRS Journal of Photogrammetry and Remote SensingA disease-specific spectral index tracks Magnaporthe oryzae infection in paddy rice from ground to space
2023, Remote Sensing of EnvironmentQuantifying spatial reallocation of land use/land cover categories in West Africa
2022, Ecological IndicatorsCitation Excerpt :LULC transition analysis is one of the best approaches to quantify deforestation and degradation of natural resources towards a better understanding of the re-greening observed in Africa (Anyamba and Tucker, 2005; Boschetti et al., 2013; Eklundh and Olsson, 2003; Herrmann et al., 2005; Leroux et al., 2017; Olsson et al., 2005).
A composite drought index developed for detecting large-scale drought characteristics
2022, Journal of HydrologyRemote sensed-based rainfall estimations over the East and West Africa regions for disaster risk management
2020, ISPRS Journal of Photogrammetry and Remote SensingCitation Excerpt :Precipitation and its associated negative impacts are natural stochastic phenomena that are unaffected by political boundaries. As seen in this study and others before it (Thorne, et al., 2001; Asadullh, et al., 2008; Dinku et al., 2011a,b, Boschetti et al., 2013; Dembélé and Zwart, 2016; Dinku et al., 2017; Maidment et al, 2017; Kimani et al., 2017; Seyama et al., 2019; Le Coz, & van de Giesen, 2020; Satgé et al., 2020) the satellite products have limitations and flaws that could be reduced if there was a common platform for the continent where all available gauged station datasets from government and private institutions (e.g. like TAHMO) and satellite products could be assimilated into a common analysis. Most of the countries in Africa are contiguous, and rainfall patterns can be considered homogenous from the regional-scale standpoint.