The surface of the earth is used by people mainly for living, cultivation, and transportation. Rivers, forests, and hilly areas are the natural users of the earth's surface. The use of the earth's surface has been altering day by day due to the necessity of people. This alteration to the land surface on earth is referred to as “Land Use and Land Cover (LULC) Change” (Gasirabo et al., 2023). Over the last few decades, the LULC has been changed due to population growth and economic development (Gasirabo et al., 2023). This population growth and economic improvement tend to expand the number as well as the area of cities. These are considered to be human and natural factors for land use changes. A number of detrimental effects result from LULC change, including deforestation, diminished agriculture and streams, conversion of grassland to urban areas, and Urban Heat Island (UHI). Hence, the factors of LULC change like population growth, rapid urbanization, and industrialization are becoming the major challenges against the sustainable development of an area. Based on the research of Garcia-Ruiz et al. (García-Ruiz et al., 1996) and Musa et al. (Ibrahim Musa et al., 2018), sustainable development has been affected due to the LULC change. To reduce the interconnected land use challenges in the upcoming future for sustainable development, observation on LULC variations data is very crucial. Hence the LULC monitoring is considered to be an important part of achieving sustainable development against these challenges (Hakim et al., 2020).
Developing countries in the South Asian regions have faced these challenges frequently (Kotharkar et al., 2018). Bangladesh like others developing countries has also faced these challenges now a days. The increased population growth in different cities of Bangladesh is accelerating the reduction of vegetation, water bodies and the increase in urban areas (Rashid et al., 2022). The capital city of Bangladesh, Dhaka was experienced a rapid change in LULC between 1975–2003 i.e., an increase in urban areas about 10553 hectares and a substantial reduction in water bodies, vegetation, crop lands and, wetlands (Dewan & Yamaguchi, 2009). Due to urbanization, the Chittagong City Corporation faced an increase of built-up area by 13.72% and a reduction in water bodies, fallow land and hilly vegetation areas (Hussain et al., 2016). In Rajshahi City Corporation region Kafy et al. (Kafy et al., 2020) showed that there was an increase of urban area almost about 16% since 1999 and a decrease of vegetation (19%) and water bodies (4%), which leads to an increase of average Land Surface Temperature (LST) by 9.83°C. On the same way, Narayanganj is considered to be a potential area in Bangladesh for its rapid industrialization (Rashid et al., 2022). Rashid et al. (Rashid et al., 2022) evaluated urban heat island effects based on numeric parameters i.e., LULC, land surface temperature (LST), normalized difference vegetation index (NDVI) for Narayanganj city area with increasing LST of 1.8°C during 2011 and 2019. Islam & Haque (Islam & Haque, 2022) identify the LULC change on Narayanganj Upazila over the year 2001 and 2021. However, there is no existence of district scale (bigger scale) quantitative analysis of LULC change and most of the previous studies had been conducted based on downscale of Narayanganj district to access the UHI effect so far. Focusing on this background, to achieve a sustainable development, this research aims to quantify the LULC change over last 20 years (2000 ~ 2020) on district scale.
LULC change pattern analysis by direct field visits are very time-consuming, laborious and error-prone (Kafy et al., 2020). Therefore, various geographic models and publicly available remote sensing data are thought to be highly useful tools for managing and keeping track of the state and evolution of LULC (Gasirabo et al., 2023). Now-a-days researchers are using various Geographic Information System (GIS) technologies and machine learning algorithms for the thematic mapping of LULC classification and LULC change (Dr. C. Pande, 2022). To classify a LULC image, multiple functions like supervised and unsupervised classification, machine learning programming, fuzzy grading, and Google Earth Engine (GEE) platform are used as the GIS and Remote Sensing (RS) technology (Dr. C. Pande, 2022). GEE is promising platform for big geospatial data processing with comprehensive analysis (Dr. C. Pande, 2022). Moreover, machine learning approach in the GEE platform is quicker to access, process, and analyze large imagery data within a very few seconds comparing with RS and GIS software.(Dr. C. Pande, 2022).
GEE is a cloud-based computing platform of Google and was officially released in 2010 with substantial computational facilities (Zhao et al., 2021). In this platform, various supervised and unsupervised algorithms are available for LULC classification task. K-means clustering, ISODATA (Iterative Self-Organizing Data Analysis) clustering, artificial neural network, Fuzzy C-means clustering
are available unsupervised classification approaches. On the other hand, K-nearest neighbor (KNN), support vector machine (SVM), random forest (RF) classifier, maximum likelihood classifier (MLC), minimum distance classifier are available supervised classification approaches (Jangid et al., 2023). Among these, Random Forest (RF) classifier is an ensemble classifier (Pal, 2005) which provide better accuracy than any other single classifier (Giacinto & Roli, 1997). Pal (Pal, 2005) suggested that in terms of classification accuracy and training time, the RF classifier is just as effective as SVMs but the RF classifier required a smaller number of user-defined parameters. Hence, the RF supervised classification approach is more widely used by researchers (Amani et al., 2020). Also in this study, the RF classifier is used to classify the LULC due to its popularity.