Comparing Chlorophyll Content Reflectance in Irrigated Parcels: A Study Using the Random Forest Algorithm and Chlorophyll Content Indexes

Chlorophyll content in vegetation is a key variable for assessing the physiological condition and photosynthetic activity of plants. This study aim to examine the relationship between reflectance and chlorophyll content in irrigated parcels (IPs) in the Mostaganem plateau aquifer land (Northwestern Algeria), using Chlorophyll Content Indices (ChICI) and a land cover classification approach. The latter approach uses unsupervised clustering to segment the pixel, followed by a Random Forest (RF) model to classify pixels into land cover types. The proposed methods are implemented in the Google Earth Engine (GEE) platform, utilizing Sentinel-2 satellite imagery. The RF model was more effective at identifying and mapping IPs areas than approaches relying solely on chlorophyll-sensitive spectral bands and indexes derived from them. The confusion matrix analysis showed high-accuracy statistics. The overall accuracy was 0.91 and the Kappa coefficient was 0.83. For the IP class specifically, the sensitivity was 0.86.