Evaluating pasture cover density mapping: a comparative analysis of Sentinel-2 and Spot-5 multispectral sensor images

• Published in: Applied water science 2024-09, Vol.14 (9), p.188-9, Article 188
• Main Authors: Mansouri, Taha, Varvani, Javad, Toranjzar, Hamid, Abdi, Nourollah, Ahmadi, Abbas
• Format: Article
• Language: English
• Subjects: Accuracy; Aerial photography; Algorithms; Aquatic Pollution; Artificial neural networks; Band composition; Classification; Comparative analysis; Comparative Law; Composition; Density; Digital Elevation Models; Digital imaging; Digital mapping; Earth and Environmental Science; Earth Sciences; Effectiveness; Ground truth; Hydrogeology; Image processing; Industrial and Production Engineering; International & Foreign Law; Multispectral imaging; Nanotechnology; Neural networks; Original Article; Overall accuracy; Parallelograms; Pasture; Pasture management; Private International Law; Remote sensing; Remote sensors; Satellite imagery; Satellite photography; Satellites; Sensors; Sentinel-2; Spectral resolution; Spot-5; Supervised classification; Vegetation; Vegetation type; Waste Water Technology; Water Industry/Water Technologies; Water Management; Water Pollution Control
• ISSN: 2190-5487; 2190-5495
• DOI: 10.1007/s13201-024-02250-y

Vegetation density extraction is influenced by the characteristics of satellite images, vegetation type, classification algorithm, and region, but there is little information about multispectral imaging (MSI). Studying the compatibility of the information content of sensors to replace sensors in areas where there is no easy access to their data is necessary for remote sensing (RS) studies. This study aims to assess the suitability of MSI from Sentinel-2 and Spot-5 satellites for generating pasture density maps. The Middle Kashkan watershed in the Lorestan Province of Iran was the study area. Geometric correction of the images was performed using ground control points (GCP) and the area's digital elevation model, achieving an accuracy of 0.21 pixels or better. Supervised classification techniques including parallelogram, minimal distance, maximum likelihood, and artificial neural network (ANN) algorithms were applied to the primary MSI of each satellite, as well as the integrated image of Spot-5 and the resulting pasture density map. Three density categories were considered: 5–25%, 25–50%, and over 50%. To validate the accuracy of the classification, a ground truth map of the region was created by interpreting a referenced official digital orthophotomosaic image at a scale of 1:40,000. Comparative analysis of the classified images revealed that the Sentinel-2 image with PCA-2-8 band composition and ANN classification algorithm yielded superior results, with an overall accuracy of 65.72% and a kappa coefficient of 0.08, compared to the Spot-5 image with PCA-3-1 band composition and the ANN classification algorithm. Spot-5 satellite images demonstrated greater effectiveness in generating pasture cover maps across the three density categories. These findings suggest that satellite images with suitable spatial and spectral resolution can be effectively utilized for generating accurate pasture density maps and monitoring long-term pasture preservation, particularly in regions characterized by high aerial photography altitudes in pasture areas. This approach holds the potential for effective pasture management and conservation efforts on a global scale.