Prediction of Land Use Changes for the Mae Soi River Sub-basin, Lampang Province, Using a Neural Network Model

Abstract

The purpose of this research was to measure and forecast the land use changes in the Mae Soi River Sub-basin, Lampang Province. Landsat satellite imagery data during the years 2001, 2011, and 2021 were used to detect such changes and make future predictions. All three years of satellite imagery data have been geometrically revised for geographic compatibility. The image data was classified and the accuracy was determined using the kappa statistic, while the prediction of land use change was made using a neural network model combined with factors influencing the changes in land use.

The study found that there was a change in land use between 2001 to 2011, with the most reduction of approximately 30,011 rai (4,801.76 ha) occurring in agricultural land, which was converted to forest land. Between 2011 and 2021, the watershed quality forests under class 1 and 2 in the Chae Son sub – district experienced the greatest reduction of approximately 2,762 rai (441.92 ha). A neural network model was used to predict the land use change by the 2031. The Jupyter Notebook platform provides the best estimate with an accuracy of 79.9 percent. Based on the predicted change between 2021 to 2031, it was observed that the forest land would decrease continuously, with approximately 13,268 rai (2,122.88 ha) being converted to agricultural land. The study area should be a defensive measure and allocate suitable areas for agriculture. In order to prevent the ingress of agricultural land into the watershed quality forests class 1 and 2, especially in the Chae Son subdistrict. Agroforestry systems can be promoted in the study area in order to promote sustainable use.