Towards an Annually Updated Oil Palm Age Database (OPAD) with Multimodal Remote Sensing Features: Design and Preliminary Result

Abstract

The cultivation of oil palm (Elaeis guineensis) is globally significant, particularly in tropical regions, due to the diverse applications of palm oil and palm kernel oil. This necessitates precise monitoring for sustainable management amidst escalating demand. This study aims to develop an annually updated Oil Palm Age Database (OPAD) using multisource remote sensing data to predict the oil palm age, a critical factor affecting productivity and environmental sensitivity. By comprehensively analyzing multisource remote sensing data from 1982 to 2023, this study establishes OPAD and validates its reliability using a random forest classifier. The findings reveal the relationship between remote sensing data features and changes in the oil palm age, offering insights into sustainable management practices. The results indicate an overall accuracy of 86.86 ± 0.92% for grouped age (i.e., seed (1-2 years), young (3-8 years), teen (9-14 years), mature (15-25 years), and post-mature (over 25 years)), with features such as GEDI forest height, ScanSAR HH SAVG, Planet NIR, and Planet Blue being instrumental in age prediction. Integrating optical, synthetic aperture radar (SAR), and canopy height data enables effective age prediction, which is essential for optimizing management strategies, assessing yield potential, and mitigating environmental impacts.