How to identify rice-growing areas using satellite remote sensing data

Rice is one of Indonesia's most important food crops, with approximately 90% of the population relying on it as a staple food. Its production directly impacts national food security and social stability. As the world's third-largest rice producer, Indonesia cultivates over 14 million hectares of rice, primarily in Java, Sumatra, and Bali. However, due to climate change, arable land loss, and diversified planting patterns, accurately monitoring rice acreage and growth status presents significant challenges. Accurate and efficient extraction of rice acreage information is crucial for agricultural monitoring, farmland management, food yield assessment, and policy formulation.

In recent years, with the increasing abundance of satellite remote sensing data and the development of cloud computing platforms, crop classification methods based on multi-source data have been widely applied. Among them, the Normalized Difference Vegetation Index (NDVI) and Radar Backscattering Coefficient (VH/VV) can effectively reflect the phenological characteristics and growth status of rice. Combined with machine learning algorithms (such as random forests), they can achieve high-precision automated extraction of rice planting areas. This study takes northern East Java Province as an example to explore methods for extracting rice planting areas based on time-series NDVI and VH data, aiming to provide a scientific basis for regional food security monitoring and agricultural policy formulation.

I. Regional Overview

The study area is located in the northern part of East Java Province, Indonesia, at longitude 111°21'31.1"E, latitude 7°26'26.7"S, as shown in Figure 1. The area has a tropical climate, with a rainy season typically from November to April of the following year and a dry season from May to October. The main crop is rice, supplemented by other crops such as maize and soybeans.

Figure 1 Study Area

II. Data Preprocessing 

Data was preprocessed using the GEE platform, covering the period from June 1, 2024 to May 31, 2025. Monthly VH and NDVI data were generated. Monthly VH data were synthesized using median filtering to reduce noise in the polarization data. NDVI is affected by image acquisition time, cloud cover, and aerosols; therefore, monthly NDVI was generated using maximum value synthesis to reduce noise impact. Furthermore, the different acquisition times from various data sources made data misalignment in the temporal dimension; this issue was resolved by synthesizing monthly data.

III. Phenological Information Analysis

By analyzing the relationship between the rice growth cycle and NDVI in the target area, the rice growth cycle can be determined. Therefore, this paper selects some feature points and obtains the NDVI time series values ​​as shown in Figure 2(a), Figure 2(b), Figure 2(c), and Figure 2(d). Based on the time series values, it can be determined that the rice in this area is grown three times a year, and the rice planting cycle is approximately 4 months.

Figure 2(a) NDVI phenological information (111°26'35.40"E, 7°28'11.37"S)

Figure 2(b) NDVI phenological information (111°26'12.41"E, 7°27'35.42"S)

Figure 2(c) NDVI phenological information (111°21'47.56"E, 7°24'50.01"S)

Figure 2(d) NDVI phenological information (111°23'41.21"E, 7°26'58.21"S)

IV. Extraction Results

Features were constructed based on the NDVI time series dataset and the VH time series dataset, and then the oil palm region in the study area was extracted using random forest. The extraction results are shown in Figure 3 below.

Figure 3. Results of rice planting area extraction

The results extracted in this study are consistent with those of historical studies. For the provided target area, a total rice-growing area of ​​76.51 square kilometers was calculated. Rice is grown in this area three times a year, with a growth cycle of approximately four months. However, the sowing times vary within this region, making it difficult to accurately extract the planting area using traditional methods.

Based on the research findings extracted from the aforementioned rice-growing areas, we possess the capability to continuously provide high-precision agricultural remote sensing monitoring services. Leveraging advanced GEE platform data processing technology and multi-source time-series data analysis methods, we can efficiently complete dynamic monitoring tasks across large-scale crop-growing areas and provide the following extended services according to customer needs: 1) multi-season crop rotation pattern analysis; 2) prediction of planting area change trends. Our team has mature experience in optimizing random forest algorithms and NDVI/VH multi-feature fusion technology, ensuring long-term stable output of crop distribution data and providing continuous technical support and data update services to agricultural management departments and research institutions.