Development of Agriculture Land Evaluation Model Based on Environmental Risk

Abstract

Karawang district is one of the largest rice granaries in Indonesia. The existence of agricultural areas is one of the important things that must be maintained to support the national food security program. However, industrial development and rapid population growth are factors in the conversion of agricultural land to non agricultural functions. The process of land conversion occurs massively on the productive area with a small scale and is difficult to identify. Based on the descriptions, it is necessary to evaluate agricultural land to support decision making, especially in the conservation and management of agricultural areas in Karawang district. The objectives of this study are to introduce the alternative method for improved image classification analysis based on hybrids time series normalized difference index for rapid mapping of agriculture area and developing the spatially explicit model to evaluate the potential of agricultural land to be converted based on environmental risks. The method introduced is a hybrid method by integrating the quantification of the Normalized Difference Tillage Index (NDTI) and the characterization of the Tillage Normalized Difference Vegetation Index (NDVI). The Normalized Difference Tillage Index (NDTI) has been widely used to identify biophysical parameters and agricultural practices. While the Normalized Difference Vegetation Index (NDVI) is optimized for the characterization and analysis of agricultural phenology. The hybrid method that integrates the quantification and characterization of the Normalized Difference Tillage Index (NDTI) and the Normalized Difference Vegetation Index (NDVI) promises a fast and precise method for image classification. From the results of the accuracy assessment using the Kappa coefficient, Producer accuracy is 95%, User accuracy is 71%, and Overall accuracy is 72% with a coefficient of 31.4%. The use of Sentinel-2 data with good spatial resolution can identify a unit area of less than 1 hectare, where that the classification results can support precision agriculture activities. The output of the Agricultural Land Evaluation model produces spatial information that can explicitly describe the effect of each environmental risk according to its weight. the agriculture area in Karawang district is dominated by areas with high risk (39.66%), very high (32.95%), and moderate (24.21%). Agricultural land objects with the highest and very high risk are adjacent to the road network. Based on validation using serial very high-resolution data from 2017 to 2020 in West Karawang and Purwasari sub-districts, it was found that agricultural areas with high and very high environmental risks have a high potential to be converted into non-agricultural areas. The advantageous information would be useful for decision-makers in determining priorities, management strategies, and alternative scenarios for agricultural land conservation.