Mysuru is a historical city and erstwhile capital and is situated in the southern part of Karnataka state. The district of Mysuru shares its borders with Tamil Nadu to the southeast, Kodagu district to the west, To the north, Mandya district is situated, while the northwest is bordered by Hassan district, and the northeast is adjacent to Bangalore district. The city's geographical coordinates span from 12° 14' 41" to 12° 22' 25" N latitudes and 76° 34' 20" to 76° 43' 23" E longitudes (Figure 1). The Cauvery River drains the northern region of the city, while the Kabini river drains the southern region. In addition, there are numerous tanks and lakes situated throughout Mysuru city and its outskirts. Additionally, Mysuru boasts numerous tanks and lakes, both within the city and its peripheries, further enhancing its ecological and hydrological significance. The distinctive spatial and environmental attributes of the Mysuru study area establish it as a perfect context for delving into the repercussions of swift urbanization on alterations in land use and land cover. It is a historically significant city known for its rich cultural heritage and is often referred to as the Cultural Capital of Karnataka. Mysuru is governed by a Municipal Corporation and plays a crucial role as an administrative, educational, and economic centre in the region. The city is located approximately 146 km southwest of the state capital, Bengaluru. Mysuru is renowned for its magnificent palaces, including the iconic Mysuru Palace, which attracts tourists from around the world. In recent years, Mysuru has experienced rapid urban expansion and development, leading to various land use and land cover changes (Aithal, Setturu, Sreekantha, Durgappa, & Ramachandra, 2012). The expansion of residential areas, commercial establishments, and infrastructure projects has transformed the city's landscape. These changes have consequences for the environment, socio-economical aspects, and general sustainability of the city. Considering the unique characteristics and ongoing urbanization dynamics, Mysuru provides an excellent case study for examining the impacts of urban expansion, land use changes, and the need for sustainable planning. The present study, an area encompassing 15 km from the CBD of Mysuru has been chosen for studying the LULC change dynamics for a period of 17 years.

In this study, a blend of primary and secondary data was employed. The primary data was collected through on-site observations and surveys conducted in the study area. Secondary data, on the other hand, was obtained from various sources. Regarding satellite imagery, a collection of five distinct Landsat images, encompassing TM, ETM+, OLI, and TIRS imageries, was procured from the United States Geological Survey (USGS) (https://earthexplorer.usgs.gov/) to cover the years 1995, 2001, 2007, 2014, and 2022 (Table 1). The selection of these images was grounded in their accessibility, absence of cloud cover, and alignment with seasons that offered optimal visibility of the analysed features. In addition, the secondary data were acquired from various sources, including both physical offices and online platforms. High-resolution images were acquired from Google Earth, while toposheets, city maps, plans, and boundaries were procured from pertinent administrative offices. The demographic data, crucial for the analysis, were collected from the Census of India. Ground observations were undertaken in the study area to collect primary data. This entailed conducting on-site visits to gather GPS reference points, which were subsequently utilized for satellite image classification and precision evaluation. The processing and classification of satellite imagery, along with the examination of changes in land use and land cover and urban expansion, were executed using software including ERDAS Imagine, ArcGIS, and QGIS.

These pre-processing techniques are essential to improve the quality and accuracy of the data obtained from satellite images (Asokan et al., 2020; Sowmya, Shenoy, & Venugopal, 2017). By applying pre-processing methods, the inherent limitations and distortions in the images can be corrected, allowing for more reliable and meaningful analysis. Landsat collection-2 level-1 datasets are of the highest standard and are considered suitable for time series analysis (EROS, 2020). In this study, atmospheric correction of the Landsat images was conducted by employing the Semi-Automatic Classification Plugin (SCP) within the QGIS 2.6.1 software. The SCP tool facilitated the correction of atmospheric influences present in the satellite images, ensuring more accurate and reliable data for further analysis. To create a comprehensive analysis, the Landsat images were processed by combining their respective bands to generate composite images for each year under investigation (1995, 2001, 2007, 2014, and 2022). Afterward, a segment of the composite image was isolated by extracting the 15 km buffer zone surrounding Mysuru's CBD for each Landsat image, thereby concentrating the analysis on the designated study region.

Following the pre-processing stage, the pre-processed satellite data was put under a supervised classification procedure to classify the pixels into distinct LULC classes. Supervised classification involves an expert overseeing the process of pixel assemblage using training sites that provide spectral statistics for each feature type. These training sites are used to guide the computer algorithm in assigning labels to the spectral signatures, categorizing them according to their respective land cover classes (Thenkabail, 2016). The objective of this classification process was to categorize each pixel based on its spectral characteristics and assign it to the appropriate LULC class 20, 21. The supervised classification technique employed in this study involved the utilization of known reference samples for training the classification algorithm (Jensen, 2005; Lillesand, Kiefer, & Chipman, 2015). The reference samples were meticulously chosen relying on the ground truth data gathered through field observations. By associating the spectral signatures of the reference samples with their corresponding LULC classes, the maximum likelihood classifier algorithm used has identified similar spectral patterns in the satellite imagery. Following an initial visual interpretation of the images and taking into account the sensor's spatial resolution, a total of five Land Use and Land Cover (LULC) Level I classes, as outlined by Anderson, Hardy, Roach, & Witmer in 1976, were discerned. Then a signature file was created by selecting pixels that exhibited uniform surface types across the study area (Alberti, Weeks, & Coe, 2004). The signature file was then used to assign the pixels to five distinct land use and land cover classes, viz., agriculture land, vegetation, water bodies, built-up and open/fallow land.

Afterwards, the model was executed using ERDAS Imagine software to generate the LULC map consisting of five thematic classes. To enhance the thematic accuracy of the classified map, post-classification error correction techniques were employed (Harris & Ventur, 1995; Manandhar, Odeh, & Ancev, 2009). These methodologies were employed to detect and rectify any misclassifications or inaccuracies that could have arisen during the initial classification procedure. To rectify any incorrectly classified pixels, manual corrections were carried out, aided by field data and supplementary information. The final land use and land cover maps, comprising five distinct classes, were generated by refining the initial generalized maps for the years 1995, 2001, 2007, 2014 and 2022.

An accuracy assessment is a method employed to determine the level of accuracy of LULC classes showed in a thematic map concerning the real classes observed on the ground 23, 27. The error matrix serves as a valuable tool for quantifying and expressing the accuracy of the map. By comparing the assigned LULC classes in the map with the corresponding ground truth information, the error matrix provides a comprehensive and efficient means of evaluating the overall accuracy and consistency of the thematic map. To conduct the accuracy assessment, reference data were obtained through a combination of field visits and digital sources. For the 2022 image, ground truth data was obtained during field visits using a handheld GPS (Global Positioning System). To acquire reference images for the past imageries (1995, 2001, 2007 and 2014), high-resolution Google Earth images and toposheets belonging to the study area were utilized. In the accuracy assessment, a stratified random sampling method was used, ensuring the inclusion of a diverse range of sample points for each image 1, 28. With over 200 sample collection points allocated for each classified image, this sampling approach allowed for a robust evaluation of classification accuracy. The accuracy assessment includes four key measures: Users' accuracy, Producers' accuracy, overall accuracy and Kappa accuracy (Congalton & Green, 2008) (Table 2). Although the accuracy assessment of the classified image using the error matrix technique is a valuable tool, it is not without limitations (Foody, 2002). Errors introduced during the assessment phase have the potential to influence the thematic accuracy of the map.

The technique of cross-tabulation and dynamic change was employed to detect and analyse the changes in LULC during two periods. This method facilitated the comparison and examination of the spatial dispersal and transition of various land use classes. The change matrix method involved overlaying the LULC maps of the two periods, enabling pixel-to-pixel comparison (Hassan et al., 2016). To create the Land Use and Land Cover (LULC) change, the overlay tool within ArcMap 10.3 and the pivot table function in Excel were employed. The objective was to analyse changes in LULC between the years 1995-2001, 2001-2007, 2007-2014 and 2014-2022. Four separate intervals were created, and for each interval, new thematic layers were generated to represent the changes in LULC. Additionally, a change matrix was also constructed to offer statistical information on the transitions or variations that happened from one LULC class to another during the specific period. The analysis has discovered the increases and losses in each LULC class. This information was then used to create a change map, which visually represented and analysed the spatial changes that occurred between the two periods.

The dynamic degree method, on the other hand, measures the extent of disparities in the area of land use over a specific time interval. To quantify the rate of change in LULC classes, the land use dynamic index was calculated using the following equation (Eq. 1) (Sowmya et al., 2017).

K=Ub-UaUaX1TX100

The magnitude of Land Use Land Cover change for a particular category, designated as K, is computed using the initial area 'a' and the final area 'b' of land use or land cover over the selected time period. The total duration of the study is denoted as T.

Finally, to measure the spatial range of built-up growth in and surrounding Mysuru, the built-up density has been calculated using a multi-ring buffer of 1 km intervals. The built-up layer of different years has been clipped using a buffer distance of 1 km intervals to measure the compactness of the built-up area concerning the buffer area.

The land use and land cover maps for the study area in the years 1995, 2001, 2007, 2014 and 2022 were derived through supervised classification of the Landsat images. The classified image (Figure 3 ) reveals the presence of five distinct categories of LULC in the study area. These categories include built-up areas represented in red, open land in brown, agricultural land in yellow, vegetation in green and water bodies in blue. The visual analysis of the classified image depicts a consistent pattern of land use and land cover transformation taking place in Mysuru from 1972 to 2018. This observation highlights the dynamic nature of land utilization and underscores the significance of monitoring and understanding LULC changes over time.

The accuracy assessment intends to assess the degree to which the classes in the classified image align with the reference image or ground truth. To achieve this objective, a confusion matrix was generated to evaluate the precision of the classified image, illustrated in Table 2 . The overall accuracy of the classified images for the years 1995, 2001, 2007, 2014 and 2022 was determined to be 91.74%, 92.04%, 92.42%, 92.13% and 91.15% respectively. These accuracy measures indicate the overall reliability and correctness of the classification outcomes.

The Kappa coefficient values for the images range between 0.88 and 0.90. The Kappa coefficient is a statistical metric that evaluates the concurrence between the classified image and the reference data, considering the extent of agreement that might arise by random chance. In land use and land cover classification, an accuracy level of at least 85% is commonly accepted as the minimum threshold for acceptable classification results (Anderson, Hardy, Roach, & Witmer, 1976).

From Table 2 , it can be observed that the accuracy of the classification varies across different years and classes. Generally, the user's and producer's accuracies were high, ranging from 81.63% to 100%, indicating a good level of classification performance. The overall accuracies range from 91.15% to 92.42%, indicating a high level of overall classification accuracy. The Kappa coefficients also indicate a substantial level of agreement amongst the classified images and the location data, with values ranging from 0.88 to 0.90. During the study period from 1995 to 2022, it was observed that the accuracy of identifying built-up land steadily improved. Similarly, the accuracy of detecting water bodies consistently remained at a high level, demonstrating a reliable and consistent performance. Conversely, the accuracy of agricultural land has diminished due to the challenge posed by mixed pixels within the study area. Open/fallow land mixed up with the agricultural land and vegetation has noticeably reduced the accuracy of agricultural area. The classification accuracies were deemed satisfactory as all recorded accuracy values exceeded 90% for all the years, indicating a high level of accuracy across the board. In summary, the classification was deemed practical as it achieved an agreed-upon value for both overall accuracy and overall Kappa coefficient, indicating a satisfactory level of performance.

Land use and land cover area of Mysuru and the surrounding area for the year 1995-2022 has been presented in Table 3 . The land use and land cover of Mysuru were observed to undergo persistent changes over time. This change is especially noticeable in the case of built-up land, which exhibits a clear upward trend in terms of growth. The built-up land class demonstrates consistent growth throughout the entire period. Starting at 6,063.90 ha in 1995, it steadily expands to 8,226.47 ha in 2001. The rate of growth rate intensifies, with a significant leap to 12,944.63 ha in 2007. Further, the trend continues, reaching 20,312.57 ha by 2022. This growth can be attributed to the demographic shift and the diversification of economic activities within the city. The nearness of Mysuru to the city of Bangalore has played a noteworthy role in driving these changes. The built-up area has expanded spatial from the CBD of the city to the surrounding area (Figure 3 ). The expansion of urban sprawl becomes particularly evident after the year 2007, with a significant upsurge in built-up lands. The present expansion is largely facilitated by the development of transport infrastructure, which contributes to the quick growth of urbanization and the spread of built-up land.

The area dedicated to agriculture experienced fluctuations throughout the years. It saw an increase from 3186.09 hectares in 1995 to 5728.33 hectares in 2001. However, there was a decline in agricultural land from 2001 to 2007, subsequently, it started increasing again in 2014 and 2022. The agricultural area was also classified under the fallow land category, as most of the agricultural area during summer (March) was left uncultivated in Mysuru. The reduced agricultural land is also substantiated by the fact that the majority of farmers hold agricultural land of less than one hectare (Uma et al., 2021). This has also contributed to the extravagant increase in the open/fallow land group in the study area. The total area of open/fallow land has decreased from the year 1995 to 2022. In 1995, open/fallow land covered 90744.21 ha. By 2022, this number had decreased to 42745.47 ha. This decrease is likely due to a combination of factors, including urbanization, population growth, and changes in agricultural practices. The vegetation class demonstrates fluctuations over time. The area increases from 22760.54 ha in 1995 to 25511.06 ha in 2001. However, it declines to 32869.55 ha in 2007 and further decreases to 19092.64 ha in 2014. Surprisingly, there is a substantial rebound with a significant expansion to 54858.25 ha in 2022. These fluctuations may be due to factors such as changes in land management practices, conservation efforts, or ecological restoration initiatives.

The total land under water bodies, such as lakes, streams, and reservoirs, shows relatively minor fluctuations. It decreases slightly from 2867.36 ha in 1995 to 2229.76 ha in 2022. These variations could be inclined by factors such as water management practices, climate variations, or human interventions affecting water bodies. Figure 4 displays the area share of various LULC categories for different years in per cent for Mysuru. The River Cauvery and KRS reservoir in the northern and Kabini River in the southern directions and several small lakes in the city represent the water bodies in the study area. The agricultural lands can be seen mainly concentrated surrounding these waterbodies both in the north and south of the city (Figure 3 ). The vegetation area was seen fluctuating, sometimes overlapping with open/fallow land due to natural growth in the abandoned agricultural area. This can be evident in Figure 4 , where the drop in open/fallow land area is compensated by increasing vegetation. One possible reason for the abandonment of agricultural fields is the changing nature of economic activities. This shift may lead to a decreased focus on agricultural practices and a subsequent decline in the utilization of agricultural fields.

To delve deeper into the examination of land use and land cover alterations between 1995 and 2022 in Mysuru, the dynamic degree index of change is utilized. This index quantifies the magnitude of shifts in land use categories over a defined period. A positive value for the dynamic degree of land use signifies progressive change, while a negative value indicates regressive change. The dynamic degree of LULC variation for Mysuru (Table 4 ) shows that the built-up land use class indicate progressive changes in all periods analysed. The built-up class has the maximum dynamic degree change, with an overall increase of 8.70%. Although the periodical rate of built-up area growth has decreased since 2007, the absolute increase remains highly noticeable. The growth is particularly visible towards the southern and northeastern directions of the city. Between 1995 and 2022, the vegetation area increased by 5.22%. This increase has been driven by several factors, including reforestation efforts, the desertion of agricultural land, and the natural succession of vegetation.

The agricultural land class has contracted the most, with a dynamic degree of -2.66. This is likely due to a combination of factors, including the decline of traditional farming practices, the increase of built-up areas, and the conversion of agricultural land to other uses. The open/fallow land class has experienced negative growth, with a dynamic degree of -1.96. This is likely due to the fact that this land class is often used for grazing or other agricultural purposes, which tend to fluctuate with the seasons. There has been a slight decrease in the extent of water bodies, which has reduced by 0.82%. The marginal decrease in the dynamic degree of water bodies is mostly accredited to the presence of rivers, which constitute the largest proportion of the water body category. Overall, the landscape of the region has experienced important variations over the last 27 years.

The post-classification change matrix was formulated by spatially intersecting distinct land use and land cover maps to evaluate the transitions between various categories during different time intervals. This matrix provides a quantitative analysis of the fluctuations in land use and land cover over the years (Table 5, Table 6, Table 7, Table 8 ). The tables display the horizontal values indicating the additional area due to the alteration from one category to another, and the vertical values representing the area contracted by the conversion from one class to another. The transverse values in the table represent the unaffected area for each LULC zone.

From 1995 to 2001, open/fallow land contributed significantly to the built-up land trailed by vegetation and agricultural land. The open/fallow land, which typically consists of unused or uncultivated areas in Mysuru, often becomes the target for urban expansion and development. As the city grows, there is a higher demand for residential, commercial, and industrial spaces, leading to the transformation of open/fallow land into built-up areas. A comparable tendency can be seen to continue for the period of 2001-2007, 2007-2014 and 2014-2022 period as well (Table 6, Table 7, Table 8 ). From 2007 to 2014 the contribution of vegetation to built-up areas has increased significantly to 1552.77 ha, denoting increased investment in transportation infrastructure and large-scale projects in Mysuru. The exaggerated increase can also be credited to classification errors that might have crept in between agricultural and vegetation categories. The agricultural area has remained a marginal contributor to built-up growth as an agricultural activity constitutes the core traditional economic activity in Mysuru. Water bodies have also contributed to a minimal area of built-up land in Mysuru. The vegetation land, which has increased significantly throughout the study period has been contributed largely by open/fallow and agricultural land. The shift in economic activity has led largely to the fallowing of the land and the growth of natural vegetation in the empty lands. This can also be observed in the synonymous contribution of vegetation and open/fallow land towards agricultural areas in different periods. The water bodies have remained constant more or less during the study period, with a major portion of the transition towards agricultural and open/fallow land.

Figure 5 a, 5b, 5c and 5d represents the spatial conversion of land use and land cover in Mysuru for the year 1995 to 2001, 2001 to 2007, 2007 to 2014 and 2014 to 2022 respectively. Observing the figures reveals that within the city's central zone, the expansion of built-up areas predominantly originates from the vegetation class. Conversely, in the periphery of the city, the growth of built-up areas is primarily attributed to open/fallow and agricultural regions. The expansion of transport infrastructure and linear expansion is particularly visible since the year 2007 in Mysuru. During 2014 and 2022 (Figure 5 d) the transition into the vegetation category is dominant across Mysuru due to rapid land use and socio-economic variations. Increased concentration of built-up land in the south near Nanjangud is becoming apparent as well during the 2014-2022 period.

To measure the spatial dispersal of built-up density in Mysuru for the different years from 1995 to 2022, the concentric buffer-based built-up density is measured (Figure 6 a-b). It can be seen from the descending curves of Figure 6 b that the concentration of the built-up area generally declines towards the periphery. However, the rate of decrease is not constant. The central part of the city exhibits a dense concentration of built-up areas, with a density of over 75 per cent persisting within a range of 6 to 8 km from the CBD. Eventually, there is a gentle sloping of the density curves from a distance of 8 km towards the fringe of the city. From a distance of 14 km from the CBD of Mysuru the density is reduced to 10 per cent and the curves become flat. The distance between curves representing periodical built-up growth maintains a uniform distance except for the years 2001 to 2007, during which there was a substantial increase in the built-up land.

Figure 6 a demonstrates a constant accretion of the built-up land in Mysuru during the study period. The growth is particularly significant in closer buffer distances, indicating more concentrated urban development and higher population density near the city centre. As the buffer distance increases, the built-up density tends to spread out, reflecting urban expansion and suburban development. The rural-urban conversion at the margin of the city remains evident by increased built-up accretion at the periphery over the years. This information provides insights into the spatial dispersal and density of built-up areas in Mysuru, which can be useful for urban planning and land management purposes.