A High-Resolution Cropland Map for the West African Sahel Based on High-Density Training Data, Google Earth Engine, and Locally Optimized Machine Learning

Abstract

The West African Sahel Cropland map (WASC30) is a new 30-m cropland extent product for the nominal year of 2015. We used the computing resources provided by Google Earth Engine (GEE) to fit and apply Random Forest models for cropland detection in each of 189 grid cells (composed of 100 km2, hence a total of ~1.9 × 106 km2) across five countries of the West African Sahel (Burkina Faso, Mauritania, Mali, Niger, and Senegal). Landsat-8 surface reflectance (Bands 2–7) and vegetation indices (NDVI, EVI, SAVI, and MSAVI), organized to include dry-season and growing-season band reflectances and vegetation indices for the years 2013–2015, were used as predictors. Training data were derived from an independent, high-resolution, visually interpreted sample dataset that classifies sample points across West Africa using a 2-km grid (~380,000 points were used in this study, with 50% used for model training and 50% used for model validation). Analysis of the new cropland dataset indicates a summed cropland area of ~316 × 103 km2 across the 5 countries, primarily in rainfed cropland (309 × 103 km2), with irrigated cropland area (7 × 103 km2) representing 2% of the total cropland area. At regional scale, the cropland dataset has an overall accuracy of 90.1% and a cropland class (rainfed and irrigated) user’s accuracy of 79%. At bioclimatic zones scale, results show that land proportion occupied by rainfed agriculture increases with annual precipitation up to 1000 mm. The Sudanian zone (600–1200 mm) has the highest proportion of land in agriculture (24%), followed by the Sahelian (200–600 mm) and the Guinean (1200 +) zones for 15% and 4%, respectively. The new West African Sahel dataset is made freely available for applications requiring improved cropland area information for agricultural monitoring and food security applications.

Lowland Rice Mapping in Sédhiou Region (Senegal) Using Sentinel 1 and Sentinel 2 Data and Random Forest

“…The statistical estimation of crop area is a complex issue that should theoretically be kept distinct from crop mapping. The direct use of classified satellite images for a statistically sound estimation of crop area is, in fact, usually hampered by the insufficient accuracy and biasedness of the maps obtained [53]. This was also the case for the current investigation, which implies that the area estimates produced by RF classification are likely not unbiased.…”

“…This combined use of dense optic and microwave datasets implies the use of a massive amount of data which, especially for very large areas, can become dramatically time-consuming and computationally difficult. From this point of view, the use of the Google Earth Engine platform, although not implemented in this study, can be utilized to overcome these limits and has already been effectively exploited in other studies [50,53] based on the use of the RF classifier.…”

“…This was also the case for the current investigation, which implies that the area estimates produced by RF classification are likely not unbiased. These estimates, however, can be corrected into unbiased values through the application of an estimator belonging to one of the two major families, i.e., calibration and regression [53,54]. The calibration estimator relies on the use of error matrices and can contemporarily correct the area estimates of all classes considered.…”

The Asia‐Pacific Biodiversity Observation Network: 10‐year achievements and new strategies to 2030

“…The 3D modeling and CT scanned data would also contribute to the development of morphology and taxonomy (see recent data in “ffish‐asia”; Kano et al, 2013, Table 1). Image recognition and analysis is also essential for biodiversity monitoring by deep/machine learning (e.g., satellite map, Samasse, Hanan, Anchang, & Diallo, 2020; seagrass, Yamakita, 2019). In addition, data representation techniques such as online dashboard are also expected for more efficient use of the data.…”

The Detecting of Irrigated Croplands Changes in 1987–2015 in Zhangjiakou

“…The overall accuracy of irrigated croplands in the Indo-Ganges Basin, the map of which was drawn by Murali et al based on MODIS data with 250 m resolution, reached 73% and the kappa coefficient reached 0.71 [59]. Five countries in west Africa were classified by Samasse K et al using GEE (Google earth engine) and a random forest model, and the accuracy of farmlands (rainfed and irrigated) types was 79% [60]. Yanhua Xie et al rapidly mapped irrigated croplands in the continental United States at a 30 m resolution, with an average Kappa value of 0.88 and an overall accuracy of 94% [57].…”