An early investigation of spatial correlation between Sentinel-2 based rice growth stages maps with satellite-based precipitation data to support digital agriculture development in Indonesia

Abstract

Sustainability of rice production is a critical issue to ensure food security and needs to be monitored on time. Therefore, the online monitoring system using Sentinel-2 has been introduced to monitor rice fields in Indonesia. However, the system needs to be coupled with precipitation to improve user usage. In this study, the spatial correlation using Pearson’s correlation analysis and linear regression between the floods and the vegetative stages area of the rice monitoring maps and the precipitation data from CHIRPS was investigated. The analysis was conducted with the two datasets with 439 regencies in Indonesia on a monthly basis from December 2019 until May 2020. The result shows that 96 regencies have a highly positive correlation (r>0.6, p>0.05, n=6) with 57 regencies have a high R2 (R 2 >0.6). Also, there are 83 regencies has a profoundly negative correlation (r≤-0.6, p<0.05, n=6) with 32 regencies with high R2 (R 2 >0.6), On the other hand, there are 79 regencies have medium R2 (0.4≤R 2 <0.6), and 271 regencies have the lowest R2 value (R 2 ≤0.4). These early-stage results show an opportunity to combine two datasets to produce early warning systems or recommended cropping calendar in a timely and accurate manner to the stakeholders or the farmers.