Visual Image Design Based on Multi-sensor Machine Learning for Monitoring Plateau Lake Dynamics and Pasture Change

The economic value of ecosystem services is precious to the community and changes with land use/land cover (LULC). The main aim of this study is to examine LULC changes in the Sahiwal area between 1994 and 2024 using a Landsat time series, and to determine ecosystem service values (ESV) and the sustainability of water and vegetation resources. The innovation lies in incorporating Random Forests (RF) into Google Earth Engine (GEE) to systematically measure the dynamics of multi-decadal LULC and quantify their implications for ecosystem services and resource sustainability. Our outcomes indicated that forest and vegetation areas declined by 0.229% and 3.9%, respectively, from 1994 to 2024 in the study area. The built-up area increased by 7.83% to 16.53% from 1994 to 2024. The overall accuracy (OA) values were 96.7%, 94.2%, 91.9%, and 87.7% for 1994, 2004, 2014, and 2024, respectively. Similarly, overall kappa (OK) values were 94%, 88%, 86%, and 82% for 1994, 2004, 2014, and 2024, respectively. Results showed that vegetation ESV rose from 8669.69 to 8934.13 million USD/year from 1994 to 2004, but declined to 8726.207 million USD/year in 2024, while water bodies showed variability, rising from 210 to 231.5 million USD/year from 2004 to 2024. The findings of this study call for improved land and water management planning, ecosystem restoration, and policy interventions to promote nature-based solutions for more sustainable water in semi-arid regions.