Modeling Rice Growth and Yield using Integrated Remote Sensing Data on Google Earth Engine

Abstract

Assessing the growth stages, health, and yield of rice is crucial for agriculture, economy, sustainability, and food security. Such assessments provide valuable insights for farmers to optimize agricultural practices, effectively manage pests and diseases, and enhance crop management, leading to improved yields and efficient resource management. This study focuses on paddy fields in Buak Khang sub-district, San Kamphaeng district, Chiang Mai, Thailand, with the objective of developing models to assess rice growth stages, evaluate rice health using the NDVI, and model rice yield using Sentinel-2 MSI and Sentinel-1 SAR (VV and VH polarizations) satellite imagery. The study was conducted during the rainy season of 2023 (June-November). Various parameters were identified to establish correlations and develop models, with field data collected to validate the models. The study yielded three models for rice growth stage assessment, with the best model achieving an R² of 0.67. Monitoring rice health revealed that on September 21, the NDVI values ranged between 0.51 and 0.91, indicating optimal growth conditions and healthy crops. Similarly, three models were developed for yield estimation, with the best model achieving an R² of 0.51. Validation showed that the growth stage model had a Kappa coefficient of 0.80, while the yield estimation model had a RMSE of 0.887 kg/m². These models, demonstrating high accuracy, provide a robust framework for agricultural agencies to develop effective agricultural policies.