Introduction

Urban sprawl is a process that is dynamic and socioeconomic in nature whereby rural areas are transformed into urban areas (Jat et al. 2008). This process does not only manifest itself through an increase in land ceiling but also gradually changes the identity of the social and physical landscape from a rural to an urban character. India is expected to undergo major rural–urban transitions in the coming years. Numerous studies have quantified and analysed the scope and consequences of rural–urban changes. (Mushore et al. 2017; Tang et al. 2007). The national spatial policy of India does not qualify nor specify where urban boundaries are located or where and how cities may expand (Fazal 2000; Jiang et al. 2007). The immediate result of this absence of boundary specifications is an unsustainable urban growth at the expense of surrounding rural territories and/or vacant lands, leading to unplanned and uncontrolled urban sprawl (Zhao et al. 2019). Additionally, this leads to an exponential increase in what can be referred to as peri-urban city boundaries characterized physically by decreasing areas of cropland and increasing areas of urban built-up and land ceiling (Anane 2022; Anugya et al. 2017). These physical changes also inflict considerable changes in the economic opportunities and possibilities, which are often manifested in loss of employment for farmers or changing socio-economic status of individual and communities in these areas of transition.

Presently, the majority of Indian cities are undergoing this shift, which is eventually having a negative impact on the rural–urban income disparity throughout the nation. (Bhatta et al. 2010; Stone et al. 2010). In other words, there is an increase in spatial inequality, visible in the amount and types of economic opportunities and dependencies in rural areas, a decrease in public services in rural areas, and decline in rural identities (Guo and Zhong 2023; Martinuzzi et al. 2007). On top of that, food security is at risk, given the huge decline in arable land and the rural workforce (Ren et al. 2013). In order to address and get a grip on these critical developments, there is an urgent need to continuously monitor and identify the growth of the transition areas, and the effect of the urbanization (Kassouri and Okunlola 2022; Sudhira et al. 2004; Zhang et al. 2023). Although it is obvious that these rural–urban transitions have negative effects on the various aspects of both the city and the rural areas, it is currently unclear how significant these effects are and what kind of effects they are specifically. (Cheng and Masser 2003; Dhanaraj and Angadi 2022). The current knowledge is that the transformations tend to result in spatial ambiguity in areas that are unplanned and where fragmented LULC patterns emerge (Ji et al. 2006; Kumar et al. 2007).

Some of the outlying rural communities transform into (peri) urban areas or urban villages as urbanization spreads, while others retain their rural character. (Balha and Singh 2018). However, the amount and direction of this spatial difference are not always obvious. As a result, additional spatiotemporal data are necessary for quantifying, mapping, and monitoring large-scale land modification. (Lillesand and Kiefer 1979; Niyogi et al. 2018). It has proven effective for improving monitoring and analysis of urban land transitions and changes, as well as urban land dynamics (Tsai 2005). Based on the findings of these studies, it is evident that urban heat islands (UHIs) result from the increased land surface temperature linked to the ceiling of rural lands (including agricultural and forestry lands) (Jeganathan et al. 2016). Recent advances in thermal remote sensing, GIS and statistical methods have helped researchers carry out studies to analyse, monitor and map the location and severity of these UHIs. These insights have significantly contributed in policy formulation procedures for a more sustainable urban development (Chen et al. 2006; Kumari and Sarma 2017; Roy and Kasemi 2021).

Given this context, this study aims at critically analysing both the physical and spatial rural–urban transitions as well as the environmental parameters associated with it them, for a particular study area. The spatio-temporal characteristics of both types of transitions relied on geospatial and statistical methods. Therefore, the primary focus of the study is:

  1. (1).

    To evaluate the area change associated with the urban growth of the GBN district between 2001 and 2016.

  2. (2).

    To determine the urban footprint in view of built-up densities and its relation with land surface temperature of the area. These outcomes establish a framework for additional examination on rural–urban transition in an applied manner

Study area

The district of Gautam Buddha Nagar (GBN), which has geographic coordinates of longitude 77°17′–77°45′ E and latitude 28°5′–28°41′ N, is one of the major cities in the National Capital Region (NCR). The study area is 1442 square kilometres and is located in the Indo Gangetic Plain at an altitude of about 200 m above mean sea level. With an average precipitation of about 790 mm, the climate varies from sub-humid with hot summers to cold winters (Fig. 1). The most common crops are wheat, rice, mustard, barely, sugarcane and maize. From 8,38,469 in 1991 to 16,48,115 in 2011, the population of the GBN has significantly increased over the past 20 years.

Fig. 1
figure 1

Location map of study area

Full size image

Materials and methods

Data sources

The temporal Landsat satellite images with the spatial resolution of 30 m acquired in 2001, 2010 and 2016 were used for thematic layers’ preparation along with the field data collection. The accuracy assessment was done through waypoints using Oregon 550 GPS handheld receivers. The utility of thermal band for deriving the temperature was utilized to investigate the variation in temperature of the area due to expansion in urban settlement. The UTM (WGS-84) coordinate system was used to reproject all the images, harmonizing the various data sets and making them comparable for spatial analysis. Hyperspherical shading space (HCS) fusion and low-pass spatial filtering were used to improve the images (Somvanshi and Kumari 2020). The details of the data are displayed in Table 1.

Table 1 Satellite data specification
Full size table

LULC characterization

Satellite images from 2001, 2010, and 2016 were classified under supervision using the Mahalanobis classifier algorithm (Somvanshi et al. 2020). This research concentrated on three main classes: built-up, waterbody, and others. Built-up areas include all surface areas that have been developed, including all paved or rocky surfaces, buildings, and other infrastructure. The waterbody category comprises of flowing waterbodies (rivers and streams) and stagnant waterbodies (ponds, lakes and waterlogged). The agriculture, vegetation, fallows land, forest, and open spaces were counted in others class. To validate the accuracy of LULC maps of 2001, 2010 and 2016 (Fig. 2), the accuracy of classified images was assessed using error matrix at producer, user and overall accuracy levels (Congalton 1991; Stabursvik 2007).

Fig. 2
figure 2

LULC classification map a 2001, b 2010 and c 2016

Full size image

Urban landscape quantification

To analyse the spatial expansion of the urban landscape classification was done for three time periods, namely 2001, 2010 and 2016. The quantification was done according to rule-based categorization scheme (Angel et al. 2007). Urban densities were analysed grid wise by dividing the entire study area into grids. Each grid comprises of an area 4 km sq. (2 km × 2 km). The developed (built-up extent and density) and non-developed (rural area) category were assigned on the basis of the urban density values. Urban density estimation parameters were derived from the compactness and association of urban pixels. Furthermore, for the footprint mapping, the sub-categories (urban built-up area, suburban built-up area, rural built-up area, urbanized open land (UOL), captured open land (COL), rural open land (ROL) and waterbody) were assigned on the basis of extent and density of built in each grid, or in other words, the percentage of total built-up area in each grid. The developed and undeveloped lands were calculated using Eqs. (1, 2).

$${U}_{\mathrm{b}}=\left[{UB}_{i}/\mathrm{TA}\right]*100$$
(1)

where Ub is extent and density of the built-up area, Ubi is the total urban built-up area in a grid, and TA is the total area of the grid.

The dynamics of urban landscape is estimated on the basis of value of Uud.

$${U}_{\mathrm{ud}}=\left[{U}_{\mathrm{udi}}/\mathrm{TA}\right]*100$$
(2)

where Uud is extent and density for the undeveloped area, Uudi is the total undeveloped area in a grid, and TA is the total area of the grid. Parameter used for identifying unsustainable sprawl is decreasing spatial extent and density of developed land as the physical fragmentation of built-up areas increases and also an increase in isolated urbanized patches.

Urban footprint mapping

The Urban footprint map of the study area was generated by classifying into seven sub-categories which were the urban area, Sub-urban area, Rural area, UOL, COL and ROL (Table 2).

Table 2 Urban footprint classification
Full size table

Pixel values were assigned based on the extent and density of urban areas. Ub > 50% was an urban built-up area, 10–50% was a suburban one, and 10% was a rural one (Fig. 3). Similar to how pixel value was assigned for developed land, the extent and density of undeveloped land were taken into account. Areas with a Uud value greater than 50% were classified as UOL, while areas with a Uud value of less than 200 ha that were surrounded by urban and suburban patches were classified as COL.

Fig. 3
figure 3

Urban footprint maps of 2001, 2010 and 2016

Full size image

Land surface temperature estimations

The methodology for deriving the LST was in accordance with LANDSAT Science Data Users Handbook.

Spectral radiance (Lλ)

Lλ = gain × DN + offset

Where Lλ is the spectral radiance, DN is the digital number of a given pixel, gain is the slope of the radiance/DN conversion function, and offset is the intercept of the radiance/DN conversion function

Brightness temperature

\(T_{B} = {{K_{2} } \mathord{\left/ {\vphantom {{K_{2} } {\left( {\ln \left( {\frac{{K_{1} }}{{L_{\lambda } }}} \right) + 1} \right)}}} \right. \kern-0pt} {\left( {\ln \left( {\frac{{K_{1} }}{{L_{\lambda } }}} \right) + 1} \right)}}\)

In this equation, TB represents the brightness temperature in kelvin, K1 represents calibration constant 1, K2 represents calibration constant 2, and L represents the spectral radiance of thermal band pixels

Land surface emissivity (LSE)

\(LSE=0.004{P}_{v }+0.986\)

Where Pv is proportionate vegetation

Land surface temperature (LST)

\({\text{LST}} = \frac{{T_{B} }}{{1 + \left( {\frac{{\lambda T_{B} }}{\rho }} \right)\ln \varepsilon }}\)

Where λ is the wavelength of the emitted radiance; ρ = h* c/σ (1.438 9 10 − 2 Mk); σ is Boltzmann’s constant (1.38 9 10–23 JK−1); h is Planck’s constant (6.626 9 10–34 Js), and c is the velocity of light (2.998 9 108 ms−1)

Results and discussion

LULC characterization

According to the spatio-temporal analysis of LULC, the built-up area was 396.59 sq km in 2001 and then rose to 594.82 sq km in 2016. The built-up area increased from 27.51 per cent of the total area in 2001 to 41.25 per cent in 2016. Over the course of 15 years, this increase in built-up area resulted in a 13.13 per cent decline in agricultural land and other classes. As shown in Table 3, the conversion of arable land took place during the development phase. Along the main roads and highway network, significant urban expansion was seen.

Table 3 Spatio-temporal LULC classification statistics
Full size table

Accuracy assessment

In order to evaluate the effectiveness of Mahalanobi's classifier in urban mapping and the accuracy of categorized images, an accuracy evaluation was conducted using an Error matrix. Table 4 depicts the accuracy assessment of classified images. Both producer and user accuracy of each class in all three years were more than 80 per cent. The overall accuracy was highest in 2016 with accuracy of 92.84% followed by 88.61% in 2010 and 87.10% in 2001. Similar trend was seen for kappa coefficient with 0.94 in 2016, 0.89 in 2010 and 0.86 in 2001, respectively.

Table 4 Accuracy assessment of LULC maps
Full size table

Urban footprint analysis

  • The urban footprint, i.e. the influence that the expansions have on the conversion and transformation of character and identify of the surrounding areas is persuasive. It becomes increasingly easy to destroy green and rural areas in favour of urban functionalities and socio-physical identities.

  • The rate of expansion is increasing. This is proven by the remote sensing analysis. If this is not stopped it is likely that other rural villages in the vicinity will be swallowed by this urban monster.

  • The amount of green and open areas is rapidly decreasing, posing a big threat to liveability of the area, and possibly also food security provided by the surrounding rural areas.

  • Agricultural areas are not replaced or re-located. Hence, there is going to be a problem finding suitable new agricultural areas.

  • Water bodies are decreasing. This may provide significant problems from drinking water availability and irrigation in the near future.

All-in-all the figure depicts a situation of a rapidly deteriorating responsible use of land.

Table 5 shows consistent rise in suburban area from 29.56 sq. km in 2001 to 32.97 sq. km in 2016. Similarly, open land was converted into other landuse for various purposes. In the span of fifteen years, it was observed that urban built-up has increased, while there was decline in rural area of GBN (Fig. 4).

Table 5 Area assessment of urban footprint classes
Full size table
Fig. 4
figure 4

Urban footprint analysis

Full size image

Temperature transition analysis

The advantage of thermal bands of satellite data is that LST can be estimated through the radiance values. During February 2001, the temperature ranged from 13.46 to 28.51 °C with a mean temperature of 20. 98 °C.

During the year 2010, the temperature ranged from 17.45 to 32.31 °C with a mean LST of 24.88 °C. Later, in the year 2016, LST ranged between 20.10 and 36.42 °C with a mean of 28.26 °C (Fig. 5). The LST over the point of rural–urban transition was analysed indicating a rise in the temperature in the vicinity of those locations.

Fig. 5
figure 5

Land surface temperature over a study area of 2001, 2010 and 2016

Full size image

The elevated temperature and density of imperviousness in the study area result in the emergence of urban heat islands (UHI) as shown in Table 6 and Fig. 6. The point with less or no change in the land use pattern indicated a very less increment in temperature. The population outburst and urban expansion have resulted in an overall increase in LST over the study area.

Table 6 Major transition locations over the study area that leads to the formation of urban heat islands
Full size table
Fig. 6
figure 6

Example of a grid (Kherli Bhav Mozumpur) depicting the formation of UHI in accordance with footprint category

Full size image

Driving factors of transition

The expansion and development of the study were done for the two time periods 2001–2010 and 2010–2016. By utilizing the results of urban footprint mapping and its attribute change, the transition points have been demarcated over the two time periods, as shown in Fig. 7. The unrestricted growth of core urban area towards the periphery or around a particular area leads to the transition of rural or open land into sub-urban and urban area. To assess the driving factors of transition the sub-category attributes were analysed and points of transition were demarcated.

Fig. 7
figure 7

Urban–rural transition point a 2001–2010 b 2010–2016

Full size image

The pattern of transition points indicates that within the study area, two types of urban expansion exist, i.e. a low density expansion that seems to be caused by an outward spreading of low-density sub-urban activities. The second type of urban expansion is a ribbon-shaped spatial sprawl, which can be characterized by enlargement of urban land ceilings along the major transportation lines. The low density expansion is prominent in the northern part of the study area, where controlled urban settlements were planned and built. This reflects the construction of planned residential areas as well as the gradual expansion or possibly redevelopment of once rural areas. In contrast, the urban socioeconomic activities and interdependencies along the Yamuna expressway roadways are reflected in the ribbon-shaped expansion.

Conclusion

The study examines spatial patterns and changes in urban areas, revealing a sprawling nature of urbanization. From 2001 to 2016, suburban and urban built-up areas increased, while rural built-up areas decreased. This expansion is influenced by two types: low-density fragmented expansion and ribbon-shaped expansion. Low-density expansion may reflect socio-economic changes, such as gentrification and illegal occupation. Ribbon-shaped expansion represents increased socio-economic activities, which require access to transportation networks. This suggests that GBN is accommodating population beyond its carrying capacity, causing environmental degradation and increasing land surface temperature (LST). The findings suggest that urban development and expansion should be planned and managed sustainably. This integrated method can provide planners and policymakers with detailed insights into long-term trends, aiding in informed spatial decisions and better managing rural–urban transitions.