Comparison of Sugarcane Drought Stress Based on Climatology Data using Machine Learning Regression Model in East Java

• DOI: 10.29207/resti.v9i2.6159
• Publication Date: 2025-04-26T06:33:31Z
• Authors: Aries Suharso; Yeni Herdiyeni; Suria Darma Tarigan; Yandra Arkeman
• Keywords: Drought stress

Crop Water Stress Index (CWSI), derived from vegetation features (NDVI) and canopy thermal temperature (LST), is an effective method to evaluate sugarcane sensitivity drought using satellite data. However, obtaining CWSI values ​​is complicated. This study introduces a novel approach estimate climatological data, including average air temperature, humidity, rainfall, sunshine duration, wind speed obtained the local weather station BMKG Malang City, East Java, for period 2021-2023. Before estimating CWSI, we analyzed water stress phenology, examined strength of correlation between looked at potential adding lag features. Our proposed prediction model uses with additional Lag in machine learning regression 5-fold cross-validation training-testing data split help optimization hyperparameters. Different models are implemented compared. The evaluation results showed that performance SVR achieved best accuracy R2 = 90.45% MAPE 9.55%, which outperformed other models. These findings indicate effects can effectively predict conditions rainfed if appropriate model. main contribution this utilization easier obtain collect than sophisticated CWSI. application shows accurately sugarcane. In drought-prone areas, strategy farmers make better choices about land management irrigation.