Crop Identification and Analysis Based on Spatial Features by Using Geospatial Technology

Abstract

The foremost objective of this research is to improve crop classification and status interpretation analysis by leveraging picture features received from space. Crop mapping and knowledge of crop conditions are critical for precision agriculture and making resource-efficient decisions. In this work, Sentinel-2A multispectral data is utilized to acquire spatial information on canopy structure, field patterns, and vegetative cover in the Sillod region of Maharashtra, India, in order to investigate the potential of remote sensing information to discriminate important crops. Such spatial information is crucial for detecting crop kind and assessing health condition over time. newlineIn order to conduct an integrative analysis of crops, the use of several methodologies were deployed, including classical vegetation cover indices such as NDVI, EVI, and GNDVI. In order to determine the best performance in terms of classification of crops and detection of variation in health, both traditional machine learning algorithms (Random Forest, SVM, KNN) and deep learning algorithms (DNN, CNN, LSTM, and CNN-LSTM) were used for crop type classification. On top of that, the estimation of chlorophyll, which is seen as an important physiological measure to capture crop health, was performed by correlating the indices obtained from satellite with ground truth data collected in the field. The combination of these dimensions provided additional detail on vegetation health and crop dynamics. newline newline