Detection of irrigated crops using Landsat 8 images: A methodology based on fieldwork and spectral reflectance analysis

Abstract

The mapping and quantification of agricultural surfaces using remote sensing (RS) data at different scales and environmental conditions have become essential to ensure the implementation of a sustainable water resource management policy. On a global scale, the steady increase in publications over the last decades reflects the significance of optical satellite images in studying land use (LU). In the present study, we suggest a methodology to identify the most suitable dates and spectral bands for mapping irrigated crops in the Guigou depression Landsat 8 satellite images. The methodology relies primarily on fieldwork and spectral reflectance (SR) analysis. The extraction of irrigated crops is carried out using the Support Vector Machine (SVM) classification algorithm. The combination of SR data, fieldwork, and vegetation indices has indicated that August is the most favorable month for studying irrigated crops. Thus, we concluded that the Near Infrared band is the most effective for discriminating agricultural surfaces. Results obtained from the processing of L8SI reveal that the classification accuracy varies based on the different land use (LU) classes. The mapping of major LU classes indicates a high level of agreement between the classified image and ground truth, with a Kappa index of 0.95 (95%). The classification of crop types shows low accuracy for potatoes and carrots, with a User's Accuracy of 0.78 and 0.76 and a Producer's Accuracy of 0.74 and 0.73, respectively. Based on the classification accuracy level, we observed that the combination of SR, fieldwork, and legend selection criteria has a high potential for distinguishing irrigated crops from other LU classes. The approach developed in this work has highlighted the importance of Landsat OLI images in mapping and quantifying agricultural surfaces in the GD. This approach could be valuable in other regions to select periods favorable to the study of irrigated crops.