Application of Remote Sensing for Agricultural Land Use Mapping in Adamawa State

This study applies satellite remote sensing techniques to agricultural land use mapping in Adamawa State, North East Nigeria, demonstrating the operational feasibility and accuracy of remote sensing-based agricultural mapping for planning and policy applications. Accurate and current agricultural land use information is essential for irrigation planning, food security monitoring, and agricultural investment targeting, but conventional field-based land use surveys are costly and infrequent. Remote sensing offers a cost-effective approach to generating current, spatially comprehensive agricultural land use maps. This study uses Sentinel-2 multispectral imagery and Landsat 8 data for Adamawa State, processed using supervised maximum likelihood and random forest classification algorithms in Google Earth Engine. Training and validation data are collected from ground truth points across 300 locations representing 10 land use classes including rainfed cropland, irrigated farmland, fallow land, and natural vegetation types. Classification accuracy is assessed using confusion matrix analysis. The map is validated against ADP crop area survey data. Findings reveal that the random forest classifier achieves an overall classification accuracy of 87.3 percent with a Kappa coefficient of 0.83. Irrigated farmland along the Benue River tributaries is clearly distinguished from rainfed cropland. Cropland area estimates from remote sensing are within 8 percent of ADP survey estimates. The study concludes that satellite remote sensing provides a highly accurate and cost-effective agricultural land use mapping tool for Adamawa State. It recommends annual remote sensing-based land use updates integrated into the state agricultural information system.

Keywords: remote sensing, land use mapping, Adamawa State, agricultural planning, satellite imagery.