Pan-Sharpening Analysis for Improved Detection Accuracy and Estimation of Coffee Plantation Land Area (Case Study: South OKU Regency, South Sumatra Province)

The use of remote sensing technology in monitoring coffee plantations is becoming increasingly important considering the vital role of coffee in the economy as an export product that increases state revenue. However, challenges remain, especially regarding the low resolution of satellite imagery which hinders accurate and efficient monitoring of coffee fields. This study aims to improve the accuracy of coffee plantation land analysis in South OKU Regency, South Sumatra Province, by using a pan-sharpening method consisting of IHS, Brovey, and Gram-Schmidt and assisted by a composite index. Satellite image sampling data from Landsat-8 was carried out at 1800 points divided into six classes. The results of the study show that the characteristics of coffee plantation land have NDVI, EVI, and ARVI values that tend to be lower, but the NDBI and NDWI values tend to be higher than the non-coffee plantation and forest classes. This study also compares the data from the pan-sharpening method using machine learning and deep learning methods to get the best classification model. The results showed that the SVM model machine learning method on the pan-sharpening brovey data gave the best results with an ACCURACY value of 83.49 and an F1-score of 83.59 percent. Keywords: Coffee Plantations, Deep Learning, Machine Learning, Pan-sharpening, Remote Sensing.