Introduction

Large scale land-use land-cover (LULC) dynamics leading to the decline of vegetation cover is one of the drivers of global climate changes and alteration of biogeochemical cycles. Global warming and consequent changes in the climate has given momentum to investigate the causes of LULC by mapping and modelling landscape patterns and dynamics and evaluate these in the context of human-environment interactions in the rapidly urbanizing landscapes. Human induced environmental changes and consequences are not uniformly distributed over the earth. However their impacts threaten the sustenance of human-environmental relationships. Post-independence period in India, particularly during the globalization era in 1990’s, the government facilitated the interactions of global industries with in-house industries. Large scale industrialization paved way for major LULC changes, caused by migration of people from different parts of the country, also from other parts of the globe and country for the employment opportunities. These led to intense urbanisation of major metropolitan cities with spurt in human population due to migration and also sprawl in peri-urban pockets. Unplanned urbanisation are characterized by the loss of diversity, with changes in the coherence and identity of the existing landscapes. The drastic landscape changes are a threat or a negative evolution, as it affects the sustenance of natural resources. Urbanisation process leads to conversion of ecological land use (such as vegetation. Open area, cultivable lands, water) into impervious layers on the earth surface. Increasing unplanned urbanisation is an important cause for depletion of resources species extension, hydro-geological alterations, loss of crop lands [1,2]. Unplanned urbanisation has various underlying effects such as dispersed growth or sprawl.

Urban Sprawl refers to an uncontrolled, unplanned, scattered urban growth as a consequence of socio economic infrastructural development leading to increase in traffic, deficit of resources by depletion of the locally available resources while creating demand for more resources [3], often exceeding the carrying capacity of the land. Sprawl causes a major imbalance between urban spatial expansion and the underlying population growth [4]. The dispersed growth or sprawl occurs basically in the periphery and the outskirts and these regions are devoid of any basic amenities or infrastructure. Sprawl can be in the radial direction encircling the city center or in linear direction along the highways, ring roads, etc. This necessitates visualization of urban trajectory for an effective urban planning.

Urban area currently with about 4 billion population, is projected to reach 8 billion by 2050 [4], which would be about 72% of the global population. Megacities, large agglomerations, are main consumers of natural resources (energy, food, etc.) with the generation of waste [5] beyond assimilative capacity of the region, continue to evolve and grow [6] with further loss of biodiversity, environmental degradation, affecting human health [1]. This phenomenon is most prevalent in developing countries [1,6] especially the rapidly developing regions in India and other Asian countries [3]. This development may be due to various factors such as political, geographical, shortage of viable land for development etc., based on region and national scale [1,2]. Urban sprawl with lack of appropriate infrastructure and basic amenities, affects urban space with due to the loss of agricultural and rural land, degradation of natural ecosystems, etc. The major causes of sprawl are attributed to huge growth of population, migration from rural to urban areas and unplanned developments. The urbanisation of core region also fuels the growth at outskirts as the population tends to move outskirts due to their lack of affordability.  Demographic change not only imply the shift from high to low rates of fertility and mortality and is also associated with the development of households and features of their life cycle. The family or life-cycle features relate mainly to labour availability at the level of households, which is linked to migration, urbanization, and the breakdown of extended families into several nuclear families. At longer timescales, the increase of population also has a large impact on land use in a region. Hence there is a need for better planning and administration. For better land use planning changes in current land use patterns temporally is essential.This necessitates the analysis of land use changes and the prediction of likely changes in the future.

Availability of spatio-temporal data with the advancement of remote sensing technologies [7] has enabled unbiased land use analysis. Analysis of land use dynamics has attained research attention both at global and Indian contexts focusing on dynamically evolving cities [8]. Temporal land use changes at regional levels have s been attempted by various researchers [9,10].  Several studies have assessed urban growth in various megacities around the world [1,11,12]. These studies though mapped and focused on temporally evolved current land use across various cities, have not addressed the likely growth required for the regional planning. Prediction of future growth are essential to control the uncontrolled development and plan for sustainable cities. Predictive models become very significant as they foresee spatial changes based on the historical land uses, which helps the decision makersin planning the growth including sprawl across the city periphery.

Urban growth models can be broadly grouped as (a) statistical models, based on regression and Markov chain [13] (b) dynamic evolving models, such as Cellular Automata (CA) [14]. Dynamic models are better suited to predict land use changes. Dynamical models coupled with agents of changes based on elements of different modelling techniques will help in better understanding of past land use changes for modelling land use dynamics.

A  Multi-layer perceptron (MLP) based CA-Markov model with a capability to incorporate the agents of spatial changes is a powerful tool [15] to predict the growth.  MLP helps in calibrating the agents and its relationship with land use changes. Markov chain helps in generating transition probability matrices based on the understanding of land use changes [16]. CA with markov considering spatial context based on neighbourhood configuration generates transition potential maps [17].  CA-Markov model is effective to model urbanisation [15]. However, for models to be effective there is a need for incorporating the agents such as social factors, economic factors, geography of an area which have decisive role in the urban process of a region. This has been demonstrated through incorporation of socioeconomic data into CA-Markov to predict land use changes [15]. This highlights the need for considering agents of changes, which still remains a research challenge.

The  objective  of  this  study  is  to  simulate  future  land  use  changes in Bangalore, India based  on  the  MLP-CA-Markov model  considering the agents of current changes.  MLP was used to calibrate the agents considering the transition of land use changes.  Transition  matrix  is  computed using the transition potential sub models based on the land  use  maps  (2008, 2010, 2012)  using  the  Markov  chain module in Land use change modeller.  Finally, spatial  distribution  of  land  uses  from  2012  to 2020 are simulated through CA model with transition  matrix and  transition  potential  map.