1 Introduction

Due to anthropogenic activities, the earth’s climate system has experienced a lot of changes. The increase in extreme climatic events has already been detected as severity and frequency of the occurrence of drought, complex seasonal variation of rainfall and intense heat waves (US Climate Change Science Program (USCCSP 2008)). Climate change (CC) has altered the hydrological process which results in uneven distribution of rainfall worldwide. Such change in rainfall patterns may keenly affect the productivity and phenology of growth and reproduction in arid and semiarid regions. The seasonal variability over many parts of arid and semiarid regions implies increasing uncertainty in the intensity of rainfall and its arrival time and duration of each rainfall event (Feng et al. 2013). India being an agricultural country is the main source of 17% of GDP (Jamil et al. 2018). The Sone command of Bihar state in India is one of the agriculturally prosperous regions, and the majority of the area is available for agricultural purposes (Kumar and Roshni 2019). Continuous land degradation and intense population growth towards the north of Sone command have put the land on enormous pressure. Demand of food increases and efficient agricultural productivity should be ensured and should be at par with the population increase (Johnson 1999; Rosegrant et al. 2001; FAO 2003; Patel and Markand 2014). Antecedent land suitability analysis is mandatory for agriculture sustainability (Bandyopadhyay et al. 2009; Collins et al. 2001; Hopkins 1977). Hence, proper management of land resources is essential for agricultural sustainability.

To ensure food security, cropland suitability assessment becomes imperative to determine the appropriate type of crop in a region (Kurtener et al. 2004; Sarkar et al. 2014; Bonfante et al. 2018). Suitability of land depends on the requirement of crops and the features of the land (FAO 1976; Mustafa et al. 2011). There are many factors which need to be assessed for determining the suitability of the land (Belka 2005). Several studies on cropland suitability have been conducted worldwide (FAO 2007). Recent advances in GIS, Remote Sensing (RS) and Decision Support System (DSS) have allowed more precision in the field of agriculture specifically the cropland suitability (Akpoti et al. 2019). The investigation for land suitability employing GIS does not fully incorporate decision maker’s preferences as it lacks a well-defined mechanism (Malczewski 2006). This problem can be solved by the integration of GIS with multi-criteria evaluation (MCE) methods. MCE is generally used as an effective tool for multi-criteria decision making. MCE integrated with GIS is useful, as variables can be evaluated and weights assigned depending on the importance of the parameters (Perveen et al. 2007). In the study of Kanlaya et al. (2009), the land suitability of sugarcane and cassava crops is achieved by the integration of multi-criteria decision making (MCDM) with FAO framework (1976). Mustafa et al. (2011) have integrated soil physical and chemical parameters using a multi-criteria decision making and GIS to generate cropland suitability for various crops. De La Rosa et al. (2004) have reviewed the analysis of qualitative and quantitative land evaluation methods and are compared by the methods of Manna et al. (2009). Their analysis suggests that to estimate the vulnerability, suitability of land, change in land use and management practices must be based on the results of land evaluation. A tool which enables the assessment of localised impacts of land related biophysical impacts has been developed by Duveiller et al. (2020). One of the key tools for ensuring sustainable agriculture and for attaining the current global food security goal in line with the Sustainability Development Goals (SDGs) of the United Nations is agricultural land suitability analysis (ALSA) (Akpoti et al. 2019).

The increase in the carbon dioxide and other greenhouse gases emissions are increasing due to population growth, industrialization and urbanization. Increase in greenhouse gases has altered the overall magnitude and seasonal distributions of rainfall in a greater extend (Feng et al. 2013). The extremity of rainfall due to the climate change has direct impact on agricultural production as timely sufficient rainfall is not available to the crops, where water availability and timing are the key factors for agricultural production. Soil texture changes very slowly and hence is not affected much due to climate change. Determination of soil pH is based on parent material, vegetation, weathering rate and climate. Thus, it is not expected to change much due to climate change (Gelybó et al. 2018). Also, due to climate change, the organic carbon does not suffer much loss in carbon (Gottschalk et al. 2012).

Dell et al. (2014) investigated agricultural models that utilised rainfall, temperature and water as the factors for productivity. Rainfall forecasted using general circulation models, which have been used to analyse the impact of climate change on rainfall (Shashikanth and Sukumar 2017; Kumar et al. 2019a, b). Global climate models (GCMs) have been used widely for predicting future climate parameters (Tesfaye et al. 2017: Amanambu et al. 2019). Quantifying the effects of past and future cropland suitability helps the policy makers for adopting proper management measures and the selection of suitable crop species. A grid-based performance evaluation approach, as well as intercomparison framework to evaluate the uncertainty of climate models for rainfall production, has been proposed (Mehr and Kahya 2017).

Accurate projection of rainfall is very important as the agricultural sector is immensely dependent on rainfall. Therefore, it becomes imperative to detect the cropland suitability analysis with spatio-temporal variation of rainfall, which is of utmost importance for the crop growth. Although several studies have focused on the MCE-GIS-integrated approach for crop suitability analysis, however to our knowledge, past literature work have not been reflected on the climate change impact on cropland suitability for agricultural crops. As Bihar is a major agricultural produce of rice and wheat crops, this work concentrates only on these two crops. The main purpose of the study is to develop the cropland suitability map of rice and wheat crops in view of the changing climatic conditions and to identify the area for expansion of the production of these crops. Cropland suitability for wheat and rice crops has been estimated for the period 2011–2100 and compared with the baseline period 1985–2011. The analytical hierarchy process (AHP) has been utilised for the multi-criteria evaluation for cropland suitability of rice and wheat crops. Four GCMs (CAN-ESM2, MPI-ESM-MR, CSIRO, CMCC-CMS) for three future time slices (2011–2040, 2041–2070, 2071–2100) of representative concentration pathways (RCP) RCP 2.6, RCP 4.5 and RCP 8.5 scenarios have been utilized for projecting rainfall.

2 Materials and methods

2.1 Study area and data acquisition

The study area covers the Sone command in Bihar, India (Fig. 1). It lies at latitude 24° 48′ N and longitude 84° 07′ E. The total command area of Sone command is 8600 km2, which covers the districts Buxar, Bhojpur, Rohtas, Bhabua and Aurangabad. Soils are alluvial in nature and vary from light to heavy-textured clays in the top layer with coarse substrata (Kumar et al. 2019a, b). The plain alluvial soil is fertile and constitutes a wide spectrum of sand, silt and humus rich clay. Recent alluvium soil is found along the banks of river Ganga. The soil is medium to highly fertile. The command area selected is almost plain in topography and it uniformly slopes towards the Ganga river. All these aspects make the area ideal for irrigated agriculture. Rice is the main crop grown in the area in the Kharif (monsoon) season and wheat in Rabi season. Crops other than rice and wheat occupy less than 2% of the cultivable command area. The irrigation system has been operative since 1871; however, irrigation in an organized manner started in 1879 (Ahmad et al. 2015).

Fig. 1
figure 1

Location of study area: Sone command in Bihar, India, and distribution of 8 rain gauge stations

Full size image

The input data used for the current study are slope map, soil pH, soil organic carbon content, rainfall and soil texture. The rainfall data of 8 stations Jagdishpur, Buxar, Inderpuri, Bhabua, Dumrawan, Durgavathy, Rohtas and Koilwar of the Sone River command for the period 1985–2011 have been used for cropland suitability analysis (Fig. 1). The observed rainfall data have been obtained from IMD Pune while the future rainfall data have been obtained from World Climate Research Programme (WCRP) CMIP5 model. The data used for model development and the source of data are detailed in Table 1. The area receives about 1100 mm of rainfall, over 80% of which occurs over the monsoon season (June to September) (Kumar and Roshni 2019).

Table 1 Description of the data used
Full size table

2.2 Crop suitability analysis using AHP technique

Multi-criteria decision evaluation (MCDE) is a method designed to facilitate decision-making. The fundamental purpose is to consider a number of options in the light of various requirements and competing priorities. The primary issue in MCDE is concerned with how to combine the information from several criteria to form a single index of evaluation. The analytic hierarchy process (AHP), introduced by Saaty (1980), is one of the effective tools widely used for dealing with complex decision making, and may aid the decision makers to set priorities and make the best decision. In short, it is a method to derive ratio scales from paired comparisons.

2.2.1 Selection of parameters for crop suitability

The parameters that affect the crop suitability analysis are selected based on the past literature studies and suggestions from the experts (Ahamed et al. 2000, Worqlul et al. 2017). The parameters selected for the Sone river command are slope map, soil pH, soil organic carbon content, soil texture and rainfall. Crop development, suitability, nutrient availability and activity of microorganisms are affected by soil pH (USDA-NRCS 1998). High organic carbon is a symbol of good fertility conditions for crop production (FAO 1976). The productivity of crop is significantly affected by soil texture. The yield of rice is favourable in soil with higher water retention, while wheat requires soil with high oxygen content and better drainage. The spatial distribution of rainfall maps has been made in ArcGIS 10.3 by importing the point rainfall data and then interpolating using IDW technique (Gong et al. 2014; Plouffe et al. 2015; Yang et al. 2015).The slope map has been generated by utilising the digital elevation model (DEM) of Shuttle Radar Topographic Mission (SRTM) having resolution of 90 m. A level surface favours equal distribution of water and is highly appropriate for rice cultivation (Samanta et al. 2011). The entire parameter maps have been georeferenced to UTM Zone 45 N in ArcGIS 10.3. The development of an adequate hierarchy structure is an ultimate objective for AHP (Fig. 2).

Fig. 2
figure 2

Hierarchy for crop suitability

Full size image

For calculating factor weights by comparing two factors together, pairwise comparison matrix introduced by Saaty (1980) is shown in Table 2. The scale varies from 9 to 1/9. A rating of 9 indicates that in relation to the column factor, the row factor is more important. On the other hand, a rating of 1/9 indicates that relative to the column factor, the row factor is less important (Kumar et al. 2016; Mustafa et al. 2011). A rating value of 1 is assigned in case the factors are equally important. Based on the previous literature observations, a pairwise comparison matrix is prepared and utilised in this study for the weight generation and is shown in Table 4. In the diagonal, elements are assigned the value of unity. Priority vector is obtained by normalising principal Eigenvector. The priority vector is obtained by calculating the mean of the rows. The summation of all values of the priority vector is 1.

Table 2 The Saaty scale (2004) for pairwise comparison
Full size table

Consistency ratio (CR) is used to suggest the probability of random generation of matrix judgements (Malczewski 2006). The formulation of CR (is given in Eq. 1) in the AHP system is ratio between the consistency index (CI) of the matrix and the random index (RI) (Table 3) (Saaty 1980). The suitability levels of the parameters (Table 4) were finalised for CR values equal to or less than 10%. Table 5 details the normalised matrix.

$$ CR=\frac{CI}{RI} $$
(1)
Table 3 RI values
Full size table
Table 4 Suitability levels of the parameters used for AHP
Full size table
Table 5 Normalised matrix of the parameters
Full size table

The CI refers to

$$ CI=\frac{\lambda max\hbox{--} n}{n-1} $$
(2)

where n refers to the order of matrix and λmax is the greatest Eigenvalue of preference matrix. In this study, the CR value of the matrix between 5 parameters is 0.0796 which is reasonable.

2.2.2 Standardization of criteria layers

Standardization is essential for suitability analysis as different parameters are measured in different units (Voodg 1983). The level of suitability for each parameter has been rated as highly suitable-S1, moderately suitable-S2, marginally suitable-S3, not suitable-N, based on the FAO 1976 guidelines for land suitability classification. All the parameters considered for suitability analysis have been reclassified using the guidelines for agricultural suitability by FAO (Table 6).

Table 6 Parameters for crop suitability for rice and wheat
Full size table

2.2.3 Weighted overlay analysis

The weights obtained by AHP process are assigned to the reclassified thematic maps of different variables. The weighted sum overlay of spatial analyst toolbox in ArcGIS 10.3 is then used for combining the different weighted maps (Eq. 3) to produce the final crop suitability map (S).

$$ \mathrm{S}={\sum}_{i=0}^n\left({W}_i\ast {X}_i\right) $$
(3)

where S is the final suitability, Wi is the weight of parameter i and Xi is the scale factor of parameter i.

The overall flowchart of the methodology that has been followed in this study is shown in Fig. 3.

Fig. 3
figure 3

Outline of the methodology of cropland suitability

Full size image

2.2.4 Effects of climate change on rainfall

Effects of climate change on crops have been communicated a lot of times by several researchers and considered for modelling crop productivity (Tubiello and Ewert 2002). Four GCM models (Table 7) and their corresponding representative concentration pathways (RCP) RCP 2.6, RCP 4.5 and RCP 8.5 scenarios have been utilised in this study to project the rainfall for the period 2011–2100. As mentioned by Yan et al. (2015), rainfall simulation using GCMs exhibits good performance and relative independence. The projections have been carried out for 8 rainfall stations in the Sone river command. GCM data have various spatial resolutions. Hence, the data have to be brought to a common resolution for application. In this study, bilinear interpolation technique has been carried out to derive the rainfall data of the particular station for the projected period (2011–2100) located between midpoints of the four neighbour grids.

Table 7 Details of the GCMs used for rainfall simulation for the period 2011–2100
Full size table

The climatic parameters obtained from GCMs are not always accurate. Bias correction is generally adopted method for developing relationship between the observed parameters and modelled parameters. In this paper, linear bias correction has been utilised (Eq. 4) for rainfall correction (Lenderink et al. 2007; Chen et al. 2013). This corrected rainfall data is utilised for further computations in this work.

$$ P\mathrm{corrected}=P\mathrm{GCM}\ \mathrm{x}\ \frac{\mu \left(P\mathrm{observed}\right)}{\mu \left(P\mathrm{GCM}\right)} $$
(4)

where Pcorrected is the corrected annual rainfall, PGCM is the annual rainfall obtained using GCM, Pobserved is the observed annual rainfall and μ is the mean of the rainfall data.

2.2.5 Mann-Kendall test

This test is frequently utilised to detect monotonic trend in weather data. The chance of detection of series as having a trend is more. The null hypothesis, H0 denotes that the data are spread identically. Let x1, x2, … xn represent n data points where xj represents the data point at time j. Then, the Mann-Kendall statistic (S) is defined as follows (Salas 1993)

$$ S={\sum}_{k=1}^{n-1}{\sum}_{j=k+1}^n\mathit{\operatorname{sgn}}\left({x}_j-{x}_k\right) $$
(5)

where the number of sample points is denoted by n, xj and xk which are the time-varying sample values where j > k and assuming xj − xk  = θ, sgn (θ) is the sign function as,

$$ \mathit{\operatorname{sgn}}\left(\theta \right)=\left\{\begin{array}{c}+1\ if\ \varTheta >1\ \\ {}\kern0.75em 0\ if\ \varTheta =1\ \\ {}-1\ if\ \varTheta <1\end{array}\right. $$
(6)

where a very high positive1value of S denotes an increasing trend and a very low negative value denotes decreasing trend. P value is Mann-Kendall statistical parameter symbolizing the probability value. This value tests the hypothesis whether it is correct or not. If the hypothesis is wrong, the trend is significant and it is declined. P value < 0.05 signifies that there is trend and if Kendall’s tau value is positive, it signifies an increasing trend and if tau value is negative, the trend is decreasing. P value > 0.05 signifies that there is no monotonic trend. In this study, Mann-Kendall test has been carried out using XLSTAT 2018 which uses exact method for the calculation of p value.

2.2.6 Impact of climate change on cropland suitability

The water availability and timing are the key features for agricultural production. Studies have proven that climate change has altered the magnitude of rainfall and the seasonal distribution globally. Hence, it is important to analyse the change in precipitation pattern in the projected period for cropland suitability. The other parameters like soil texture and pH change very slowly, and hence, the changes due to climate change are not considered in this study. The determination of soil pH is based on parent material, vegetation, weathering rate and climate. Thus, it is not expected to change much due to climate change (Gelybó et al. 2018). Also, due to climate change, the organic carbon does not suffer much loss in carbon (Gottschalk et al. 2012). Keeping these parameters like soil pH, carbon content, clay and slope of the area same, the cropland suitability maps for the future period have been generated utilising the projected rainfall obtained from the GCMs. The projected rainfall using GCMs (3 time slices i.e. 2011–2040, 2041–2070 and 2071–2100) is used for understanding the changes in a better way. The variation in the cropland suitability for rice and wheat has been examined with respect to the baseline period.

2.3 NDVI and LULC

NDVI stands for normalised difference vegetation index. It utilises multispectral remote sensing data to generate vegetation index. Near-infrared and red bands are the bands of satellite which are responsive to vegetation. The algorithm defined for calculating NDVI (shown in eq. 7) is the ratio of the difference of the red reflectance values from the near-infrared and the summation of near-infrared and red bands.

$$ \mathrm{NDVI}=\left(\mathrm{NIR}-\mathrm{RED}\right)/\left(\mathrm{NIR}+\mathrm{RED}\right) $$
(7)

where NIR and RED are the reflectance in the near-infrared and red region wavelengths respectively of the electromagnetic spectrum. The NDVI maps have been developed utilising OCM 2 filtered product downloaded from BHUVAN portal.

Landsat 8 satellite images (spatial resolution 30 m) have been obtained from Earth Explorer in the form of tiles (Table 8). Land use/land cover map for the year 2019 covering three classes water, built up and agriculture have been generated using supervised classification in ArcGIS 10.3.

Table 8 Satellite data specifications
Full size table

3 Results and discussions

3.1 Cropland suitability analysis for rice and wheat crops for baseline (1985–2011)

As detailed in section 2.2, each layer feature for the cropland suitability has been analysed. The ability of each land feature has been assessed using symbols S1 (highly suitable), S2 (moderately suitable) and S3 (marginally suitable).

For the analysis, the slope map of the Sone river command is prepared and is shown in Fig. 4. Of the total area, 46.5% has very less slope of < 5.1% making the land highly suitable for the growth of crops. Moderately sloping 5.1–10% is 52.8% of the total area. Steeper terrain is 0.7% of the total area. The rice crop cultivation is highly favoured in flat and level surfaces (Samanta et al. 2011).

Fig. 4
figure 4

A plot of slope map of the Sone river command

Full size image

The spatial distribution of the observed average rainfall for the baseline (1985–2011) is plotted and is shown in Fig. 5. It is seen that 10% of the area which lies in the upper part of the command area has received the average annual rainfall more than 1150 mm. Majority of the study area (77%) experienced average annual rainfall in the range 950–1150 mm thus making the study area suitable for the growth of crops like rice and wheat.

Fig. 5
figure 5

Spatial distribution of average annual rainfall (mm) for the baseline 1985–2011 in Sone river command and the proportion of rainfall distribution in pie chart

Full size image

Using AHP technique, the weights of the selected parameters have been generated and are assigned to the reclassified maps of soil pH, carbon-content, slope, rainfall and clay content. The weighted maps are then overlaid in ArcGIS for generating the final cropland suitability maps of rice and wheat crops. The spatial distribution of the cropland suitability maps for rice and wheat crops of Sone river command and % suitability for different margins for the baseline is shown in Fig. 6. Rice crop requires high rainfall and suitable soil with high water retention properties. The crop suitability map shows that the regions of Rohtas and Koilwar districts are highly suitable for rice production. The marginally suitable part is unevenly spread across the whole basin. The high suitability is accounted for favourable soil properties, climatic conditions and topography. The pie chart of the baseline period shows that 31% of the total area is highly suitable, 66% is moderately suitable and 3% is marginally suitable for rice crop. For wheat crop, the suitability map is predominantly of moderately suitable type. Highly suitable area is uneven for the baseline period covering the central part extending towards east and west. The pie chart of the suitability map of wheat symbolises that of the total area. The 17% area is highly suitable, 80% is moderately suitable and 3% area is marginally suitable for the crop production.

Fig. 6
figure 6

a A plot of cropland suitability map for rice and wheat crop and b the distribution of area for highly, moderately and marginally suitable for rice and wheat crops in pie chart for the baseline (1985–2011)

Full size image

3.2 Impact of climate change on cropland suitability for rice and wheat crops

As mentioned, the rainfall is considered the only variable feature due to climate change in the present study. The rainfall data from four GCMs and three RCPs for the projected time period 2011–2100 has been selected and interpolated for the selected gauge sites. The whole data were segregated into three time slices (2011–2040, 2041–2070 and 2071–2100). The variation in rainfall as a result of changing climatic conditions is illustrated in Table 9. The results reveal a very significant variation in rainfall for most of the GCMs when compared with the baseline.

Table 9 Changes in annual average projected rainfall (2011–2100) of different climate models with respect to the baseline
Full size table

For the time slice 2011–2040, all the models projected a decrease in rainfall with respect to baseline. CAN-ESM2 RCP 8.5 shows the highest decline of 14.33% while CAN-ESM2 RCP 2.6 shows the least decline. The period 2041–2070 also followed similar trend with all the stations showing a decline in rainfall from the baseline. CAN-ESM2 RCP 2.6 is displaying the highest decline of 14.537% while CSIRO RCP 4.5 is showing the least decline. For the period 2071–2100, 6 models are showing a decrease while the model CAN-ESM2 RCP 4.5 is showing a minor increase in rainfall. Overall, there is a decreasing trend in the average annual rainfall for most of the models from 2011 to 2100 in the Sone river command owing to the influence of climate change.

The rainfall is projected for all the selected gauge locations and the checked the bias corrections. To examine the spatio-temporal variation of the projected rainfall, the projected rainfall is plotted in Fig. 7. The changes in the projected rainfall are clearly visible in the figures for respective models and time slices. The rainfall obtained from the model CAN-ESM2 RCP 4.5 (2071–2100) shows the highest rainfall (> 1150 mm) spread of 11.4% of the total area and CAN-ESM2 RCP 8.5 (2011–2040) displayed the least area of 7.92% under rainfall > 1150 mm. Also, models CAN-ESM2 RCP 8.5, CSIRO RCP 4.5 (2011–2040) and CAN-ESM2 RCP 2.6 (2041–2070) display the highest area 47.44%, 44.95% and 48.55% respectively under the least rainfall of < 850 mm.

Fig. 7
figure 7

Spatial distribution of the average annual rainfall for the projected time slices 2011–2040, 2041–2070 and 2071–2100

Full size image

As the variations in the spatial distribution of rainfall are evident for each GCMs, the use of average downscaled values minimises the possible inhomogeneity or unrepresentativeness of station data (Wójcik 2015, Amanambu et al. 2019). In the present work also, as the differences of model results are very evident and hence, for each time slices, an average projected rainfall of all GCMs is calculated and is plotted in Fig. 8i. It is seen from figure that there are changes in the projected average rainfall with respect to the baseline. For the period 2011–2040, 2041–2070 and 2071–2100, the rainfall at 5 stations Jagdishpur, Buxar, Inderpuri, Bhabua and Durgavathy is showing an increasing trend while at 3 stations Koilwar, Dumrawan and Rohtas, the rainfall is showing a declining trend. The current study also projected the rainfall deficit (Fig. 8ii ), which is calculated as the difference in the future rainfall (which is the average of all the ensembles) with the base period and plotted for three time slices. For all the three time slices, rainfall deficit of <− 225 mm is observed for the western part of the command with the highest area coverage of 1.04% for the time slice 2011–2040. The topmost part of the command is showing positive deficit or rainfall surplus of > 75 mm for all the three time slices with the highest area of 7.7% for the time slice 2071–2100.

Fig. 8
figure 8

(i) Projected average rainfall (mm) and (ii) rainfall deficit for the period a 2011–2040, b 2041–2070 and c 2071–2100

Full size image

For understanding the pattern of the future rainfall trend analysis, using Mann-Kendall test has been carried out.

Trend analysis using Mann-Kendall test (Table 10) for the projected rainfall suggests that of the 8 stations, only 2 stations Dumrawan and Bhabua are showing increasing trend. No significant trend is observed for the rest of the stations.

Table 10 Trend analysis of projected rainfall (2011–2100) of different stations employing average of GCM products
Full size table

Based on the projected rainfall data of different models, cropland suitability map for rice is generated and is plotted in Fig. 9 for the period 2011–2100. The results of crop suitability for different time slices fluctuated due to the non-uniform rainfall for different GCMs. It is observed that the suitability map of the study area differs with the GCMs and time. Highly suitable area for rice production is characterised by the central part and the northern part of the study area. The figure also depicts that the area with high rainfall has higher suitability for rice production. For the time slice 2011–2040, CSIRO RCP 4.5 is exhibiting the least area of 13% highly suitable for rice crop, while CAN-ESM 2 RCP 2.6 is showing maximum area of 21% highly suitable for rice cultivation. CAN-ESM 2 RCP 2.6 is showing least value of 10.5% and CAN-ESM 2 RCP 8.5 exhibits the highest area of 23% highly suitable for rice cultivation for the time slice 2041–2070. The period 2071–2100 projected an average area of 21% highly suitable for rice farming, which is high compared with the other two time slices. The results of MPI ESM MR RCP 2.6 show the least area of 19% while CAN ESM 2 RCP 4.5 is showing the highest area of 25% highly suitable for rice growth.

Fig. 9
figure 9

Spatial distribution of cropland suitability map of rice crop for the future time slices obtained from GCMs

Full size image

Figure 10 shows the projected average suitability map of rice crops for the periods 2011–2040, 2041–2070 and 2071–2100. It is seen that the land appropriate for rice production for the future time frame is evolving, which is ascribed to the rainfall which has indicated unevenness across the command. From the pie chart (Fig. 10), for the period 2011–2040, 19% area is most suitable, 72% area is moderately suitable and 9% area is marginally suitable. For the period 2041–2070, 20% is most suitable, 71% is moderately suitable and 9% is marginally suitable. The period 2071–2100 is showing higher value of most suitable area of 24%, 69% is moderately suitable and 7% is marginally suitable. Compared with the results of the observed cropland suitability, the average area of rice crop for the whole command will experience decrease in highly suitable area. During the period 2011–2040, the highly suitable area decreases by 38.7%, for the period 2041–2070, the area decreases by 35.5% and for the period 2071–2100, the decrease is 22.6% forming an overall decrease of highly suitable area from 31% to 21%.

Fig. 10
figure 10

Average of projected cropland suitability map for rice crop and pie chart showing the distribution of highly suitable, moderately suitable and marginally suitable for the period

Full size image

The change in cropland area for rice crop suitability with respect to the baseline is shown in Fig. 11. Highly suitable area is decreasing for all the three time slices with the highest decrease of 12% for the period 2011–2040. Both moderately and marginally suitable areas experience an increase of 6%. For the time period 2041–2070, highly suitable cropland is showing a decrease of 11% while moderately suitable cropland is showing an increase of 5%. Marginally suitable area is showing an increase of 6%. The period 2071–2100 is showing a decrease of 7% for highly suitable cropland and an increase of 3% and 4% for moderately and marginally suitable cropland respectively.

Fig. 11
figure 11

Percentage changes in cropland suitability with respect to the baseline for rice crop

Full size image

Similarly, the cropland suitability map for wheat crop is also generated for the projected period 2011–2100 utilising 4 GCMs with 3 scenarios in Fig. 12.

Fig. 12
figure 12

Spatial distribution of cropland suitability map of wheat crop for the future period obtained from GCMs

Full size image

The cropland suitability map using different scenarios is showing uneveness for wheat cultivation. Highly suitable area for wheat production is extending from the central part towards the southern part. The topmost part of the study area is moderately to marginally suitable for wheat production. For the period 2011–2040, CAN-ESM2 RCP 8.5 is showing maximum area 56.8% for highly suitable cropland for wheat, while CAN-ESM2 RCP 2.6 exhibits the lowest area highly suitable for wheat production. Highly suitable area 56.93% is maximum for the model CAN-ESM2 RCP 2.6 while least 33.9% for CAN-ESM2 RCP 8.5. For the period 2071–2100, CSIRO RCP 4.5 is showing maximum highly suitable area of 56.4%, while the least area 35.8% is exhibited by CAN-ESM2 RCP 2.6.

Similarly, the cropland suitability map for wheat crop using average rainfall for the period 2011–2040, 2041–2070 and 2071–2100 is generated and shown in Fig. 13. The highly suitable area for wheat production is maximum for the period 2011–2040, while the period 2041–2100, the highly suitable area is decreasing. From the pie chart for the period 2011–2040, 40% of the total area is highly suitable, 58% is moderately suitable and 2% is marginally suitable. For the period 2041–2070, the highly suitable areas decrease to 36% and for the period 2071–2100, the cropland again decreases to 28% (Fig. 13). The highly suitable area for wheat production during the baseline is 17% which increases to 34% by the end of 2070. For the wheat crop, there is an increase of cropland for highly suitable area from 17 to 35% during the period 2011–2100.

Fig. 13
figure 13

Average of projected cropland suitability map for wheat crop and pie chart showing the distribution of highly suitable, moderately suitable and marginally suitable for the period

Full size image

The change in cropland suitability of wheat with respect to the baseline has been shown in Fig. 14. For the time period 2011–2040, the highly suitable cropland has increased by 23% with respect to the baseline. The moderately and marginally suitable cropland is displaying a decrease of 22% and 1% respectively. The 2041–2070 time period also exhibit an increase of 19% for highly suitable area while there is a decline of 18% and 1% for moderately and marginally suitable cropland respectively with respect to the baseline. For the period 2071–2100, there is an increase of 11% for highly suitable cropland and a decrease of 11% for moderately suitable cropland with respect to the baseline.

Fig. 14
figure 14

Percentage changes in cropland suitability with respect to the baseline for wheat crop

Full size image

The LULC map (Fig. Fig.15) and NDVI map (16) have been compared with the generated cropland suitability map. It is seen that generated crop suitability map follows both the maps. The percentage area of vegetation is calculated from the NDVI maps and found that 97.5% area is showing proper vegetation in the month of February and 96.4% area in the September month for the year 2012 which is close to the projected average crop land suitable for the period 2011–2040 for rice which is 91% for rice and 98% for wheat. Also, the LULC map follows the cropland suitability map. Table 11 represents the distribution of the three LULC classes observed for the year 2019 which demonstrates the accuracy of the crop suitability map.

Fig. 15
figure 15

NDVI for rice and wheat crops for the year 2012

Full size image
Table 11 Distribution of LULC (%)
Full size table
Fig. 16
figure 16

LULC map of the Sone command for the year 2019

Full size image

4 Conclusions

Rice and wheat crops are the major food crops for majority of population across the world. Due to water shortage, improper soil and land properties, and ineffective farming practices, the yield of rice and wheat is decreasing. Therefore, it is imperative to distinguish the cropland suitability for the production of rice and wheat crops in the Sone river command. This study confirms the capability of integrated GIS and MCDA for generating long-term cropland suitability maps of the Sone river command. The cropland suitability maps are prepared using observed and GCMs climatic data of rainfall for the past (1985–2011) and future period (2011–2100). The rainfall is projected for four GCMs and three RCP scenarios. The average of the projected rainfall for the three time slices is used for the generation of cropland suitability maps for rice and wheat crops. The results of the projected rainfall are showing an overall decreasing trend in the projected rainfall with respect to the baseline. Suitability analysis for the cropland suitability has been carried out using analytical hierarchical process for the two major crops rice and wheat. The study area has been divided into three classes i.e. highly suitable (S1), moderately suitable (S2) and marginally suitable (S3). The results signify that the largest area under highly suitable class is for wheat cultivation (35%) followed by rice cultivation (21%). These land features do not have any restrictions for crop cultivation. The projected cropland suitability signifies that most part of the study area is moderately suitable for rice and wheat crops. Highly suitable area for rice experiences a decrease in crop area from 31 to 21%, while the area under wheat production shows an increase in area from 17 to 35% of the total area which suggests that wheat can be the key crop. The trend analysis using Mann-Kendall test shows that of the 8 stations, only 2 stations are showing an increasing trend while the rest of the stations show no trend. Rice cultivation is strongly influenced by water. Since only 2 stations are showing an increasing rainfall trend, it can be concluded that rice cultivation will show a decline while wheat which does not require much water experiences an increase. The cropland suitability map closely follows the NDVI map thereby asserting the authenticity of the developed maps.

Right place for the right crop is essential for the sustainable yields. Consequently, it is inferred that most of the study area has a huge potential for the production of cereal crops and can be brought under agriculture with varying degrees of suitability depending on future changes in climate. The result of this study is helpful for the farmers who are profoundly dependent on agriculture for the efficient cropland management and sustainable agriculture. For further research in this field, it is suggested that more factors like financial, social criteria and change in land use/land cover pattern ought to be chosen that will refine the accuracy of the result.