An evaluation of multiple land-cover data sets to estimate cropland area in West Africa. Issue 22 (16th November 2016)

• Record Type: Journal Article
• Authors: Adhikari, Pradeep; de Beurs, Kirsten M.
• Is Part Of: International journal of remote sensing. Volume 37:Issue 22(2016)
• Page Start: 5344
• Page End: 5364
• Publication Date: 2016-11-16
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2016.1232869
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: West Africa is one of the fastest growing regions of the world and depends heavily on rain-fed agriculture for its food production. This study evaluates 12 freely available land-cover and land-use (LCLU) data sets at the ecoregion, country, and pixel levels in West Africa to estimate croplands. The selected data sets are primarily derived using remote-sensing data, representing different time periods and using various classification schemes. The result shows a very high variability of the estimated cropland at all levels. Despite this variability, data sets having a finer spatial resolution and representing a similar time period – specifically data from the International Institute for Applied Systems Analysis-International Food Policy Research Institute (IIASA−IFPRI), Global Land Cover−Share (GLC−Share), Moderate Resolution Imaging Spectroradiometer–University of Maryland (MODIS−UMD), Global Cropland Extent, Moderate Resolution Imaging Spectroradiometer−International Geosphere Biosphere Programme (MODIS−IGBP), and GlobCover version 2.3 (GlobCover V23) – estimated comparable cropland areas at the ecoregion and country levels. The countrywide cropland area, obtained from the selected data sets, when compared with the sum of arable land and permanent crop area obtained from the Food and Agriculture Organization (FAO), showed a high coefficient of determination ( R 2 > 0.95) for IIASA–IFPRI and GLC−Share. At the pixel level, at the original resolution, the newer data sets have a comparable user's accuracy (UA>53%) and producer's accuracy (PA>46%), except for the Global Cropland Extent data. Overall, two data sets – IIASA−IFPRI and GLC−Share – performed better in the region to estimate the cropland area at all levels.