Detection of recent changes in Gambia vegetation cover using time series MODIS NDVI

MODIS-NDVI, de


Introduction 1
In West Africa, particularly in The Gambia, vegetation cover has undergone significant changes over the past five decades (Ariori and Ozer 2005;FAO, 2015;CILSS, 2016;Diedhiou et al., 2020) in relation to climate variability and human actions. The former is commonly indexed as the main driver of vegetation change in West Africa (Herrmann et al., 2005). It determines the zoning of the vegetation cover and the amount of plant biomass. Its role in changes in vegetation cover has been demonstrated several times (Philippon et al., 2008;Hountoudji, 2008;Cisse, 2016). However, climatic variability marked by decreasing rainfall and increasing temperature has been noted in The Gambia since the drought years, that is to say the years 1970-80 (Sanneh et al., 2013Diedhiou, 2019). For the second, it is about agricultural land clearing, firewood and charcoal cutting, overgrazing, and the advance of the urban front (Diedhiou, 2019;Bah et al., 2019;Fent et al., 2019). 2 Thus, in order to participate in the production of knowledge in achieving Goal 15 of the Sustainable Development Goals on the preservation and restoration of terrestrial ecosystems, monitoring of changes in vegetation cover is necessary. For a State such as The Gambia and for the United Nations Development Program (UNDP), such monitoring is a major interest for better resource and environmental management. This monitoring takes on a particularly interesting dimension through the identification, description and analysis of the ongoing process. It is all the more interesting considering that the literature on the issue throughout the Republic of The Gambia is very little, if not scarce. Most of the studies conducted there have focused on adaptation to climate change and agricultural production (Akon-Yamga et al., 2011;Yaffa, 2013;Sanneh et al., 2014;Sonko et al., 2019;Jarju et al., 2021). A few rare studies have focused on the dynamics of land use and changes in vegetation cover both in wetlands and land, either at the zonal, regional or national scale (Andrieu and Mering, 2008 ;CILSS, 2014;Diedhiou, 2019;Bah et al., 2019;Fent et al., 2019;Diedhiou et al., 2020;Dieye et al., 2021). Studies carried out at the West African or Sahel scale have paid little attention to the analysis of results at The Gambia scale (Herrmann et al., 2005;Dardel et al., 2014;Leroux et al., 2014;Brandt et al., 2016).

3
Techniques for monitoring recent changes in vegetation cover at the national or West African band level includes remote sensing, through the analysis of trends in time series of MODIS images, is positioned as an effective means (Andrieu, 2018;Zoungrana et al., 2018;Gansaonre et al., 2020). These time series provide the most appropriate data for analyzing the trend of changes related to natural and human phenomena (Hamimina et al., 2013). Using the Normalized Difference Vegetation Index (NDVI), they identify land degradation and regeneration (Eckert et al., 2015;Solly et al., 2021). This index provides information on photosynthetic activity, leaf water content, soil moisture, phytosanitary quality, primary productivity, biomass, etc. It can detect changes in vegetation productivity through changes in vegetation cover (Guo et al., 2018). In addition, it is strongly correlated with rainfall in many West African countries (San Emeterio et al., 2011), especially in areas that record less than 1000 mm per year (Nicholson et al., 1990). Differences or changes in species types, the presence of disturbed vegetation in an area with a similar vegetation type, and evolutionary factors may also be reflected through NDVI (Yengoh et al., 2015).

4
The objective of this study is therefore to determine changes in vegetation cover in The Gambia, through the use of time series of NDVI MODIS images over the period 2000-2019.

Study area 5
The Republic of The Gambia is located in the West African zone between latitudes 13°30 and 13°49 N and longitudes 16°48 and 13°47. Except on the Atlantic coast, the country is bordered on all sides by the Republic of Senegal. The country is about 480 km long and varies in width from 48 km in the river estuary to 24 km inland (Ministry of Environment, 2018). It covers an area of 11,259 sq.km spread over six administrative regions (Fig. 1). Land use and land cover are composed of 14 main classes: forest, gallery forest, woodland, mangrove, savannah, wooded to shrub savannah, cultivation areas, irrigated crops, plantations, settlements, bare soils, sandy surfaces, water bodies and swampy and floodplain grasslands (CILSS, 2016;Ministry of Environment, 2018). From a climatic point of view, the year is characterized by two seasons. The dry season, which usually starts from November to May, and the rainy season from June to October. Average annual rainfall varies from 600 to 900 mm, and seasonal temperatures vary between 32 C and 34°C and are generally higher in the eastern part of the country (Loum and Fogarassy 2015;Bojang et al., 2016). The population is estimated at 1,867,000 inhabitants in 2013 and is mainly concentrated in the Greater Banjul Area, with 1,390,000 inhabitants in 2016 (CIESIN, 2005). The main economic activities revolve around agriculture (rice, maize and peanut), cutting and marketing of firewood and service, and fishing. These are also the main activities that affect the development of vegetation cover in The Gambia. Data used 7 To detect changes in vegetation cover in The Gambia, we used MODIS NDVI data. The MODIS satellite has the advantage of providing images over 16 days available since the second half of February 2000 with a spatial resolution of 250 m. In terms of the long NDVI time series (>10 years), it is the most spatially resolved series that exists today. Compared to NOAA AVHRR data, MODIS data showed better performance for such a study (Fensholt, 2004). The images are uploaded to https://search.earthdata.nasa.gov/ as granules. The products are provided in the Hierarchical Data Format for NASA's Earth Observing System (HDF-EOS) format and are composed of several bands, including NDVI. The series used here runs from 2000 to 2019, from version 6 of the vegetation indices of the Terra medium resolution imaging spectrometer (product MOD13Q1).

Treatment method 8
The methodological approach began with the creation of the time series (TSF) from the module "Earth Trends Modeler" on Idrisi TerrSet, using all the images (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019). After this step, we corrected the images by replacing the erratic values (especially related to clouds and missing lines) with interpolated values using the Preprocess function (Denoise). . Two parameters were considered here. This is the correlation coefficient "tau" and the significance "p". The correlation coefficient (S) highlights the trend in the vegetation index. It is given by equation 1 and 2: al., 2020), and according to the estimator used (p, slope TS, Z, TS Intercept). The 0.1% threshold is used here because, at 0.05% which is the most used, we have almost no significance in the trends. In this study the estimator used is p. 17 The result made it possible to detect the different trends (positive and negative) of the NDVI and their significance, presented in the rest of the work. A positive value indicates an upward trend while a negative value indicates a downward trend. The trend is significant positive when ≥0.1; significant negative when ≤-0.1.

Vegetation trends in The Gambia between 2000 and 2009
18 During the decade 2000-2009, the average Kendall correlation generally indicates a positive trend of NDVI on 98.37% of the study area. However, this trend is not significant (Fig. 2). Correlation values range from 0.2 to -0.15. 19 From a spatial perspective, positive correlation values are observed in all administrative regions. The southern part of the North Bank Region and the northern part of the West Coast Region and the Lower River Region had the most significant positive correlations with a significant trend at p≥0.1. Otherwise, negative but not significant correlations values were also observed. In these three regions, the positive trends represent 20.33%, 16.21% and 13.82% respectively (Table 1). Total all trends 100 20 However, more than a quarter of the positive trends in vegetation productivity in the Republic of The Gambia during this decade are recorded in the Central River Region, although this is also the region with the most pixels of significant negative trends. 21 The percentage of significant positive trend (at the 10% threshold) is 1.57%. The significant negative trend is 0.04% and is mainly observed in the northeastern part of the Central River Region.

Vegetation trends in The Gambia between 2010 and 2019
22 During the decade 2010-2019, we generally see a decreasing trend in vegetation productivity (Fig. 3a) compared to the decade 2000-2009, but not significant in the pValue test (Fig. 3b). Only a few pixels with a significant positive trend are observed; they are dispersed throughout all administrative regions except the West Coast Region. This region was also the one with the largest increase in NDVI with 13.87% of the study area (Table 2).  Vegetation trends in The Gambia between 2000 and 2019 24 During these two decades, the average of the Kendall correlation indicates that 61.86% of the national territory is marked by a positive trend and 38.14% by a negative trend ( Fig. 4a; Table 3). However, the southern part of the North Bank Region, and the northern part of the West Coast Region and the Lower River Region, are the ones that have recorded the strongest trends in increasing vegetation productivity. These trends are also significant overall at the p test (figure 4b). The Central River Region also recorded a significant positive trend in NDVI.   It is also significant from north to south along the river. In contrast, in the decade 2010-2019, there was a general downward trend in productivity in 44.01% of the study area, but less pronounced in the western part of the Lower River Region, West Coast Region and Banjul. The calculation of the Kendall correlation over the period studied indicated that 59.2 per cent of the national territory was marked by a positive trend and 40.8 per cent by a negative trend. Over the entire study period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019), the Kendall correlation calculation indicated that 59.2% of the national territory is marked by a positive trend and 40.8% by a negative trend. The percentage of significant positive trend is 10%, the negative one is 4%.
33 This study is important in that it brings new knowledge to decision makers in relation to the existing knowledge on changes in vegetation cover in The Gambia. It allows for more accurate identification of the different trends that are taking place and the main determinants of these trends, which are necessary for the development of effective forest resource management policies. However, despite the effectiveness of the methodological approach used, it would be useful in the future to include, field data, land use and land cover data, and spatialized precipitation data in the analysis.

ABSTRACTS
In The Gambia, the vegetation cover has undergone significant changes over the past two decades. To assist policy decision-making, this study seeks to detect trends in changes over the Detection of recent changes in Gambia vegetation cover using time series modi...