Machine Learning Application for Identifying The Agricultural Land Use Change to Support Regional Food Security Estimation in Southern Area of Kulon Progo
Bangkit Fatwa Pratama, Liana Ni^Mathus Sholikah, Zulfa Khoirun Nisa, Ansita Gupitakingkin Pradipta, Ngadisih, Sahid Susanto, Akram Sripandam Prihanantya, Rose Tirtalistyani, and Sigit Supadmo Arif

Department of Agricultural and Biosystems Engineering, Faculty of Agricultural Technology, Universitas Gadjah Mada

Department of Geodetic Engineering, Faculty of Engineering, Universitas Gadjah Mada

Abstract

Population and the massive construction of public facilities increased land demand. The southern region of Kulon Progo is one area that has experienced a lot of land conversion due to population growth, the construction of the New Yogyakarta International Airport (NYIA), and the construction of Southern Cross Road (JJLS). The purpose of this study was to determine the conversion rate of land in the form of paddy fields to the condition of food security and to predict the limit of food self-sufficiency (rice) based on the availability of paddy fields in the southern area of Kulon Progo. Identification of land-use change was carried out using the Google Earth Engine (GEE) platform based on five land cover classes: water bodies, paddy fields, vegetation, built-up land, and open land with variations in 2005, 2010, 2015, and 2020. Data processing was conducted gradually, including composite image process, NDVI calculation process, supervised classification, field validation, calculation of food security based on food production, and prediction of food self-sufficiency limit in rice. This study revealed the rate of conversion of paddy fields in the period 2005-2010, 2010-2015 was 45.43 ha/year, -79.90 ha/year, and in the 2015-2020 period, there was a significant increase to 85, 81 ha/yr due to new paddy fields. The research locations were in food-secure conditions for the four sub-districts, namely Galur, Panjatan, Wates, and Temon in 2005, 2010, 2015, and 2020, except for Wates in 2005 showed food insecurity conditions. Prediction of food self-sufficiency limit in the form of rice may occur in the next 48,48 years (2068) with an available agricultural land area of 1345,04 ha. The consideration used was the availability of food only comes from the research area. Besides, there is no effort to increase the intensification or extensification of paddy fields. Another assumption was the rate of conversion of agricultural land and population growth is constant.

Keywords: machine learning, agriculture, land use, change, food security

Topic: Geospatial Technologies in Agriculture