PREDICTING URBAN GROWTH OF KATHMANDU VALLEY USING ARTIFICIAL INTELLIGENCE

Abstract

Globally, urban areas are growing at a tremendous speed. The goal of regulating urban growth has become one of the crucial challenges of the twenty-first century. The Kathmandu Valley illustrates the rising urbanisation trend that is conquering the Himalayan foothills. Kathmandu Valley landscapes have seen tremendous transformation in the last four decades, resulting in significant changes in land use and cover (LULC). My research first detected a pixel-based unsupervised classification technique to classify these images into Four LULC categories using four Landsat images from 1992, 2003, 2013, and 2021 through Remote sensing and GIS technique. LULC noticed the change and then was analysed in light of proximate causes and factors driving those changes. The built-up area was only 71.11 sq. km in 1992 and 80.53 sq. km in 2003. 143 sq. km in 2013 while 156.64 sq. km in 2021. This built-up expansion has occurred with the conversion of agricultural land. The majority of urban growth happened from 2003 to 2013 and still growing along the major roads in a concentric pattern altering the cityscape of the valley. The centrality feature of Kathmandu valley and the massive surge in rural-to-urban migration are the primary proximate causes of the fast expansion of built-up areas and rapid conversions of agricultural regions. Using the QGIS 2.8.3 version MOLUSCE plugin (MLP-ANN) model, this research intends to detect land-use changes in the Kathmandu Valley between 2013 and 2021 as forecast and establish potential land-use changes in the years 2031 and 2041 to create urban growth scenarios in Kathmandu Valley. The prediction model was able to forecast the 2031 and 2041 LULC maps with an overall accuracy of 93.82 percent. The valley has minimal land resources for new development because of its rugged mountainous terrain. Therefore it becomes necessary to comprehend the spatial process of urban growth and forecast future urban growth scenarios. This model optimises the spatial patterns of future urban growth allocation under three scenarios: Business as usual or baseline scenario, environment-friendly, and resource-efficient sustainable development Scenario. The expected spatial patterns reflect where and how the valley's urban growth is likely to go by 2041, and they provide important information on land availability for future land development projects. Compared to 2021, the forecast model shows a maximum rise of 8.32 per cent in the built-up area by 2041. As a result, the study's findings show that employing the LULC and CA-ANN models can assist in identifying future trends, which can help governments, planners, and stakeholders estimate and evaluate the likely repercussions of future policy options. Before making decisions, what-if scenarios for policy implementation are created, which aids in the valley's long-term growth.