Estimation of Rice Productivity in South Sulawesi, Indonesia Using Meteorological Data Nur Azizah (a), Masayuki Matsuoka (a*)
a) Department of Information Engineering, Mie University
1577, Kurimamachiya, Tsu, 514-8507, Japan
*matsuoka[at]info.mie-u.ac.jp
Abstract
Accurate estimation of rice productivity is critical for ensuring food security in South Sulawesi, Indonesia, where rising consumption and declining yields have increased reliance on imports. This study aims to develop a predictive model of rice productivity in South Sulawesi by analyzing the relationship between meteorological factors and reported yields. We hypothesize that key climatic variables precipitation, temperature, solar radiation, and related parameters may have significant explanatory power for estimating annual rice productivity at the district level.
The research focuses on six districts with varying production levels. Meteorological data from January 2022 to December 2024 are being sourced entirely from global reanalysis and satellite-based datasets (ERA5, CHIRPS, etc.) to ensure comprehensive spatial and temporal coverage. Production reports from official agricultural statistics will serve as the dependent variable. Data preprocessing and exploratory analysis are underway to identify seasonal patterns and anomalies. The next phase is to apply machine learning models, including Random Forest and Gradient Boosting, to evaluate the predictive contribution of each climatic factor.
While the analysis is ongoing, the expected outcomes include insights into the relative influence of meteorological variables on rice productivity and a validated, scalable prediction model. These findings aim to inform adaptive planning strategies and improve agricultural sustainability in Indonesia.
