SPATIAL MACHINE LEARNING FOR MONITORING TEA LEAVES AND CROP YIELD ESTIMATION USING SENTINEL-2 IMAGERY - A Case of Gunung Mas Plantation, Bogor - Remote sensing and GeoAI for tea yield prediction

Aug 5, 2026

Upload by Logic

Indonesia’s tea production and export volumes have shown a declining trend over the last five years, partly driven by more competitive global quality requirements. Accurate crop yield estimation is essential to optimize tea plucking management, but it faces practical constraints that limit speed and effectiveness. Remote sensing combined with machine learning has been used in precision agriculture, and cloud platforms enable large-scale processing. This study proposes GeoAI using Sentinel-2A imagery, machine learning, and Google Collaboratory to predict ready-for-plucking tea leaves and estimate crop yield for Gunung Mas Plantation, Bogor.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Tea production trends and challenges

Crop yield estimation needs for tea management

Limitations of sampling-based yield estimation

Role of plucking time and management

Remote sensing and cloud-based AI approaches

Method Overview

Data source: Sentinel-2A bands and feature extraction

Labeling with tea block boundaries and plucking data

Classifiers: Random Forest and Support Vector Machine

Yield estimation from classification outputs

Results and Discussion

Overall accuracy of RF vs SVM

Accuracy at optimally aged tea blocks

Comparison of RF and SVM performance

SPATIAL MACHINE LEARNING FOR MONITORING TEA LEAVES AND CROP YIELD ESTIMATION USING SENTINEL-2 IMAGERY,(A Case of Gunung Mas Plantation, Bogor) Dini Nuraeni1,2, Masita Dwi Mandini Manessa1 1 Department of Geography, Faculty of Mathematics and Natural Sciences, University of Indonesia 2Geospatial Information Agency of Indonesia, Bogor, Indonesia e-mail: [dini.nuraeni@ui.ac.id / dini.nuraeni@big.go.id](dini.nuraeni@ui.ac.id / dini.nuraeni@big.go.id) Received: 30.12.2022; Revised:31.12.2022; Approved: 31.12.2022 Abstract. Indonesia's tea production and export volume have fluctuated with a downward trend in the last five years, partly due to the increasingly competitive world tea quality. Crop yield estimation is part of the management of tea plucking, affecting tea quality and quantity. The constraint in estimating crop yields requires technology that can make the process more effective and efficient. Remote sensing technology and machine learning have been widely used in precision agriculture. Recently, big data processing, especially remote sensing data, machine learning, and deep learning have been carried out using a cloud computing platform. Therefore, we propose using GeoAI, a combination of Sentinel-2A imagery, machine learning, and Google Collaboratory, to predict ready for plucking tea leaves at optimal plucking time at Gunung Mas Plantation Bogor. We used selected bands of Sentinel-2A and extracted more features (i.e. , NDVI) as a training set. Then we utilized the tea block boundaries and tea plucking data to generate labels using Random Forest (RF) and Support Vector Machine (SVM) . The classification results were further used to estimate the production of crop tea yield. The RF classifier is able to achieve an overall accuracy of 51% and SVM of 54% . Meanwhile, accuracy at optimally aged tea blocks is able to achieve 75.62% for RF and 52.88% for SVM. Thus, the SVM classifier is better in terms of overall accuracy. Meanwhile, the RF classifier is superior in predicting ready for plucking tea at optimally aged tea. Keywords: GeoAI, Sentinel-2, machine learning, crop tea yield estimation. 1 INTRODUCTION Indonesia is the eighth world's largest tea-producing country after China, India, Kenya, Argentina, Sri Lanka, Turkiye, and Vietnam, with a total production of 138,323 tons in 2020 (FAO 2022) . During the period 2016– 2020, Indonesian tea production and volume of tea exports fluctuated with a downward trend (BPS 2020) . In 2020, Indonesia ranked tenth with 46,265 tons of total exports and fourteenth with 96.3 million USD of export value (BPS 2020, FAO 2022) . Low crop production of tea is caused by decreased plant performance, fertilization below the target, lack of support crop for infrastructure, limited quantity and skill of picking workers, and inadequate supervision (PTPN VIII 2019) . The sales volume of tea in 2020 also decreased due to the over-supply in producing countries which are also potential buyers, and the scarcity of containers in Indonesia (PTPN VIII 2020) . In addition, the competitive quality of tea from the world’s tea-producing countries has resulted in low market absorption in Indonesia. The quality and quantity of tea are influenced by the management of plucking tea (Dewi et al. 2019) . It starts from harvest planning to post-harvest, which requires optimal preparation to increase tea productivity. Preparing work plans for harvest and post-harvest activities requires information on estimated crop yields. Crop yield estimation is an activity that estimates the potential production for the following harvest season. The estimated yield must reflect the actual production or be close to accuracy because it will affect the time, cost and resources needed (Junaedi 2020) . PTPN VIII uses a sampling method on a specific area in several tea blocks to get the average tea yield and estimate tea production each month. There are constraints in estimating tea yields, such as the time requ

FAQs

What problem does the study address in tea cultivation management?

It targets the need for more effective and efficient crop yield estimation and prediction of ready-for-plucking tea leaves to support tea quality and production planning.

Which data and tools are used for the GeoAI approach?

The method uses Sentinel-2A imagery, selected spectral bands and extracted features such as NDVI, and performs modeling with machine learning in Google Collaboratory.

How are tea leaf readiness and crop yield labels generated?

Tea block boundaries and tea plucking data are used to generate labels for training the machine learning models.

Which classifier shows better performance overall, and when does RF excel?

SVM achieves higher overall accuracy (54% vs 51% for RF), while RF is superior in predicting ready for plucking tea at optimally aged tea blocks (75.62% vs 52.88%).

Aug 5, 2026

Upload by Logic