Machine Learning Based Land Use Land Change Classification for Remote Sensing Data

Abstract

While developed land in cities plays a key role in land use and land cover, developing land in urban areas is crucial to land use and land cover. Land change is taking place in the built-up area of a region is a critical indication of urban expansion. While traditional methods treat land cover (LC) and land use (LU) separately from remote sensing imaging, which misses the interrelated hierarchical and layered linkages that exist between them, remote sensing imagery sees the entire picture at once. In this paper, the authors develop a successful hybrid neural network for the classification of both LCS and LUs for the first time. The proposed learning methodology uses iterative updating with a and a convolutional neural network, helps the learner build knowledge over time (CNN). In the proposed method, multilayer perceptron (MLP)is employed with CNN approach for LU classification. To derive the dynamic character of the built-up region in the urbanizable area of Jaipur city, data and software were used to collect remote sensing information and to integrate that data with geographic information systems (GIS). Using the land surface reflectance data product, it was estimated that the total built-up area will increase by 25 percent by the year 2021. A study states that built-up covered an additional 45.35% of the Earth's surface over the period 2000–2011, and another 66.53% of the surface was built up during the 2013–2021 time frame. 106.63 square kilometers, which is a lot, and there have been dramatic changes in land use, mostly due to suburban development during the last two decades (a 96 % change). The findings quantified built-up area change patterns and demonstrated how remote sensing and GIS techniques may be applied to correctly and cost-effectively map and evaluate changes in the built-up area over time in metropolitan settings.