Analysis of water demand for tertiary crops based on planting phase identification using sentinel-2 imagery and high-resolution near-infrared airborne data

Abstract Several regions in Indonesia suffer from insufficient water supply for irrigation due to climate conditions, often leading to crop failures. Situbondo Regency is one of the areas experiencing these issues, with limited water availability from irrigation systems. Prior research by Kurniawan (2020) utilized the Normalized Difference Vegetation Index (NDVI) algorithm applied to Sentinel-2 satellite imagery to assess planting phases, but the data processing required advanced computational resources and significant storage capacity, making it time-consuming. This study, however, leverages Google Earth Engine, a cloud-based platform, to process and store data more efficiently. By using Sentinel-2 imagery and the NDVI vegetation index, the research identifies different rice planting phases and creates maps that guide the calculation of water demand for tertiary irrigation plots in the Sampean Irrigation Area of Situbondo Regency. The highest NDVI value was recorded on January 26, 2022, during the Mid-Season Stage 1, covering an area of 749.87 hectares, while the lowest value was on March 11, 2023, marking the end of the planting phase, with 489.09 hectares of rice fields. Classification using the Random Forest machine learning method demonstrated excellent accuracy, achieving an Overall Accuracy (OA) of 99.52% and a kappa coefficient of 0.9935365 for the Sentinel-2 imagery dated March 11, 2023. The study also calculated the irrigation water needs, finding that the highest demand occurred on February 25, 2023, at a rate of 253.20 liters per second in study area, while the lowest demand coincided with the harvest on March 11. The findings suggest that rainfall during December and January was sufficient to meet the irrigation requirements, eliminating the need for supplementary irrigation from channels during that period.