Soumare S., Andrieu J., Fall A., Diaw E.H.B., Descroix Luc. (2022). Spatial and temporal analysis of rain-NDVI relationship in lower and middle Casamance from 1982 to 2019. Remote Sensing in Earth Systems Sciences, 5, 246-262. ISSN 2520-8195.
Titre du document
Spatial and temporal analysis of rain-NDVI relationship in lower and middle Casamance from 1982 to 2019
Année de publication
2022
Type de document
Article
Auteurs
Soumare S., Andrieu J., Fall A., Diaw E.H.B., Descroix Luc
Source
Remote Sensing in Earth Systems Sciences, 2022,
5, 246-262 ISSN 2520-8195
Contrary to claims of widespread irreversible degradation of vegetation and landscapes in West Africa, a recent increase in seasonal vegetation indices of Sahelian areas has been observed, which has been interpreted as a consequence of the rainfall recovery after the major droughts (1968-1994). This paper aims at revealing the climatic drivers of such greening in Casamance. A multi-scalar and multi-satellite remote sensing approach was implemented to study the temporal trends of vegetation activity and their relationships with rainfall in Lower and middle Casamance. The datasets used are the NOAA NDVI (GIMMS) from 1982 to 2015, the MODIS NDVI (MOD13Q1) from 2000 to 2019, and rainfall data from the Ziguinchor station from 1982 to 2017. These two NDVI time series were studied, on the one hand, with a classification method to discretize the different vegetations according to the rhythms and intensities of their vegetation activity throughout the year and, on the other hand, with Mann-Kendall's correlation to reveal the trends. Almost three-quarters (72.5%) of the pixels show a significant positive trend (regreening) between 1982 and 2015. The simple correlation between NDVI and rainfall is very low (r2=0.17) but both lagged correlation (r2=0.86) and the correlation between NDVI and cumulated rainfall of longer periods (r2=0.75) are strong. In other terms, after 1998 stronger rainfall in July and August give stronger NDVI in October and November. While a rainfall positive trend since the 1980s appears to be the main causal factor for the increase in vegetation indices, negative trends were also locally observed that are not explained by the rainfall-vegetation relationship and thus hypothetically a human-induced change.
