Next Article in Journal
Rice Terrace Experience in Japan: An Ode to the Beauty of Seasonality and Nostalgia
Next Article in Special Issue
Comparison of the Analytic Network Process and the Best–Worst Method in Ranking Urban Resilience and Regeneration Prioritization by Applying Geographic Information Systems
Previous Article in Journal
The Role of Vegetation Monitoring in the Conservation of Coastal Habitats N2000: A Case Study of a Wetland Area in Southeast Sicily (Italy)
Previous Article in Special Issue
Soybean (Glycine max) Cropland Suitability Analysis in Subtropical Desert Climate through GIS-Based Multicriteria Analysis and Sentinel-2 Multispectral Imaging
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:

Analysis of Land Suitability for Maize Production under Climate Change and Its Mitigation Potential through Crop Residue Management

Department of Hydraulics, Soil Science and Agricultural Engineering, School of Agriculture, Aristotle University of Thessaloniki, 541 24 Thessaloniki, Greece
Department of Earth and Environmental Sciences, Faculty of Bioscience Engineering, KU Leuven, 3000 Leuven, Belgium
Flemish Institute for Technological Research, 2400 Mol, Belgium
Department of Meteorology and Climatology, School of Geology, Aristotle University of Thessaloniki, 541 24 Thessaloniki, Greece
Author to whom correspondence should be addressed.
Land 2024, 13(1), 63;
Submission received: 19 November 2023 / Revised: 29 December 2023 / Accepted: 31 December 2023 / Published: 4 January 2024
(This article belongs to the Special Issue Geospatial Data in Land Suitability Assessment)


Land Suitability Analysis (LSA), under the impact of climate change, is a fundamental approach to the design of appropriate land management strategies for sustainable crop production and food security. In this study, the FAO framework was used to assess the impact of climate change on land suitability for maize in Flanders, Belgium. The current LSA revealed the marginal suitability for maize cultivation, characterizing most of the agricultural land in Flanders and identifying precipitation as the most limiting factor for maize suitability. The LSA, under two climate change scenarios, was based on climate projections from several CMIP5 Global Circulation Models, transformed into future land suitability projections and assembled into a multi-model ensemble (MME) of projected suitability changes. The results indicate an average reduction in projected suitability of approximately 7% by 2099 under the high-emission scenario. The potential of the Soil-Improving Cropping System (SICS) to mitigate the impacts of climate change on land suitability was statistically significant under both low- and high-emission scenarios. This research provides valuable insights into the MME modeling of climate change impacts on land suitability and its associated uncertainty, with the application of SICS as a potential long-term mitigation measure to promote sustainable agricultural practices.

1. Introduction

With the onset of an era in which climate change significantly influences the suitability of land for agricultural activities, altering temperature patterns, precipitation levels, and environmental conditions, it has become imperative to estimate and spatially quantify the possible impacts under different climate scenarios, while proposing mitigation measures to address the potential impacts. The Land Suitability Analysis (LSA), i.e., the assessment process of land suitability for certain agricultural crops in areas under evaluation, plays a crucial role, especially when future climate change scenarios must be considered to achieve sustainable agriculture. Based on the LSA, agricultural land use can be optimized by identifying appropriate land management strategies for current and future land use. By considering various factors, such as soil quality, climatic conditions, and topographic features, the LSA helps to ensure that land resources are efficiently managed, promoting sustainable agricultural practices, and preventing land degradation [1].
The analytical approach of LSA enables the identification of areas that may experience shifts in agricultural productivity due to climate change. By recognizing areas that may become less suitable for certain agricultural activities, adaptation strategies can be prioritized to address possible problems [2]. Moreover, LSA enables the evaluation of agricultural management strategies that increase productivity, reduce agricultural inputs, conserve natural resources, and mitigate climate change impacts. These practices encompass soil carbon sequestration methods, like compost and manure application, cover crops, reduced tillage, and Crop Residue Management [3], and have a direct effect on farming and cropping systems, aiming to make land use and agricultural production more sustainable. Cropping systems encompass the specific crop type and the agronomic management practices applied by farmers, while Soil-Improving Cropping Systems (SICSs) are those cropping systems that result in a lasting increase in the soil’s capacity to accomplish its fundamental functions and, in particular, its production of food and biomass, along with the provision of several other ecosystem services [4]. The adoption of SICS enhances soil functions and prevents degradation, thereby reducing risks like soil organic matter (SOM) decline, which degrades structure, impairs water/nutrient retention, and reduces productivity. Amid the most aspiring SICS, Crop Residue Management (CRM) can reduce mineralization of net SOM and enhance organic matter contribution into the soil [5]. Crop residues provide crucial organic matter and nutrients that are typically removed during harvest. CRM is established by returning both aboveground and belowground biomass to the field after harvest and is considered a beneficial management practice, enhancing soil quality and increasing productivity. CRM can improve soil structure, enhance soil organic carbon (SOC) sequestration capacity, and substitute fertilizer input to an extent, leading to an increase in yield and productivity of agricultural cropping systems in a sustainable manner [6].
The decision-making process of LSA requires interdisciplinary collaboration, incorporating inputs from soil and plant sciences, climatology, socioeconomic sciences, and local knowledge of crop production. Assessing land suitability for specific crops entails a comprehensive analysis of natural and socioeconomic factors within the geographical context [7]. LSA employs the multicriteria evaluation (MCE) approach, considering a subset of globally significant factors based on available data, expressed as a function of spatial factors or criteria affecting the suitability based on their individual weights. The Food and Agriculture Organization (FAO)’s approach, commonly used in LSA [8], matches land utilization types to land requirements across land units, employing a Boolean mapping technique that overlooks continuous soil variation and uncertainties. To resolve this, LSA is usually established under MCE techniques on spatial data related to physical and socioeconomic attributes of the agricultural environment of crop production [9]. The MCE procedures involve the incorporation of expert preferences on the relative importance of input factors on land suitability. Expert opinions are transformed into factor weights, using pairwise comparisons through the Analytic Hierarchy Process (AHP) and integrated into the LSA. Fuzzy logic techniques are also employed in the process of mapping input environmental attributes into a common suitability scale, using membership functions [10]. Fuzzy inference has the advantage of formulating this mapping by using soft class boundaries in the form of a transition zone or partial membership, taking into consideration the uncertainty involved.
Another important aspect to consider is that the impact of climate change on agricultural activities varies across spatial dimensions and may be different depending on the geographical location. The resulting effects depend on multiple factors, including the existing climate and soil conditions, the direction of climate change, and the capacity of the affected areas to withstand and adapt to the forthcoming changes [2]. Suitable areas for ‘key crops’ for a specific area may be affected because climate change is expected to disrupt water availability levels through changes in rainfall, evaporation, runoff, and soil moisture storage, in addition to significant temperature fluctuations [11]. Global crop stress is anticipated due to extreme seasonal temperature variations caused by climate change, resulting in negative yield responses for wheat, maize, and barley. Identifying resilient agro-ecological zones and realizing climate change implications for land suitability are crucial for sustainable crop production [12]. Integrating future climate projections into the LSA process enables the assessment of climate change impacts on land suitability. However, research in this field remains limited [2]. Climate projections represent a potential evolution of future climate characteristics (e.g., temperature), based on hypotheses about the future state of global greenhouse gas (GHG) emissions [13]. Impact studies on land suitability under climate change are mostly based on climate model data. GCMs simulate future climate characteristics under anthropogenic forcing, i.e., current, and projected future emissions of greenhouse gases. GCM climate projections exhibit considerable uncertainties, stemming either from an insufficient understanding of Earth’s climate system or from an inadequate representation of the complex climate system processes. Disagreement between GCMs on the direction and extent of the climate response to the same forcing increases the uncertainty that ultimately cascades to LSA studies that use GCMs projections [14]. Similarly, representative concentrations pathways (RCPs), which are scenarios demonstrating different possible climate changes under various socioeconomic storylines [15], introduce their own uncertainty, corresponding to the ambiguous future of GHG emissions. For these reasons, most studies use several GCM projections, combined with multi-model ensembles (MMEs), in order to reduce the uncertainty related to climate modeling and increase the corresponding confidence level [16].
In LSA under climate change, efforts focus on statistically downscaling GCMs to provide several climatic variables, including temperature and precipitation with enhanced spatial resolution, that are suitable for agricultural impact studies. These climate projections are usually integrated into the processing chain of multiple environmental factors, using standard LSA methods, including AHP combined with MCE or multicriteria decision making (MCDM) and fuzzy inference, or even machine learning (ML) methods [2], that, despite limitations from computational costs, have been proven useful in identifying the most important environmental factors that govern land suitability [10]. Hybrid LSA approaches have also emerged [17], attempting to combine standard LSA methods with mechanistic crop modeling, which has been extensively used in impact studies to assess the influence of climate change on agricultural production and specifically on crop yield [18]. Crop simulation models (CSMs) dynamically determine plant phenology-related sensitivities to climate change, providing more accurate information on the effects of climate change during crop development [19]. CSM simulation results are integrated into the empirical land evaluation process, thereby improving the evaluation approach. The development of CSMs relies primarily on data availability, requiring long time series and high-resolution data, usually only available at the local level [1]. Incorporating CSMs into LSA shifts the focus from land to crop, i.e., yield as a response of physiological processes of crop growth and development to future climate change, integrating complex and nonlinear effects of climate on crops. Maize CSMs have been widely used to estimate future yields under climate change, with a range of possible outcomes for crop growth and yield, in several contrasting maize production sites [20]. Results from Lusignan, FR, based on ensembles of several CSMs simulations, indicate a negative yield response to rising temperatures, up to 4.5% by 2050, due to a shortening of the growing cycle period, affecting biomass accumulation and ultimately yield. With a similar focus on crops, but following an eco-physiological growth modeling approach, mechanistic species distribution models (SDMs) assess climatic suitability of crops, using biologically meaningful crop response functions, projecting the distribution of crop potential into future climate conditions [21]. In a global-scale mechanistic SDM attempt [22], using future climate projections from two specific GCMs (CSIRO Mk3.0 and MIROC-H) under a high-emissions scenario, all European countries are projected to be suitable for maize to some extent, while, by the year 2100, climatic suitability for maize is expected to expand towards the northern latitudes of the European continent.
For LSA at regional-to-national scales, the physical environmental factors considered fall within the three general categories, namely soil, climate, and topography [2]. Crop growth is governed by the spatial and temporal heterogeneity in climatic conditions, which, along with soil–topography interactions, determine the suitability of a location for crop production. In addition to this well-established approach, socioeconomic factors could be included in LSA for a more realistic assessment of land suitability. However, the assessment of several socioeconomic criteria is a challenging process, carried out usually by agricultural investment planners and policy makers [23]. The objective of this study was to evaluate the application of a SICS as a long-term climate measure influencing land suitability for maize in Flanders (BE), based on future land suitability projections under different climate scenarios, while identifying the principal limiting factors that determine maize land suitability.

2. Materials and Methods

2.1. Study Area

The study area is Flanders (BE), located in the northern part of Belgium, bordering France, and the Netherlands (Figure 1). Geographically, Flanders is mainly flat, occupying part of the North Sea coast. The total area exceeds 13 K km2 and is densely populated, with soils formed mainly on sedimentary and loess deposits. The main arable crops are wheat, potato, maize, grass, and sugar beet [24].
Flanders has a mild–temperate Western European climate, influenced by North Sea conditions, characterized by high precipitation in winter and water scarcity in summer. The high temporal variability of precipitation complicates the attempt to associate the frequency and amplitude of rainfall extremes with global warming [26]. Nevertheless, monthly mean temperatures and the likelihood of heatwave occurrence have risen considerably over the past two decades, and future changes in weather conditions due to climate change are expected to have significant impacts on agricultural activities [24]. Extensive masking of non-agricultural areas was carried out, based on the annual 100 m global land cover maps available from Copernicus Global Land service. The masking process limited the total agricultural area in Flanders to approximately 11 K km2. Of the 7 agricultural regions of northern Belgium (Figure 1), the three main agricultural regions that were considered in this study are the Sandy Loam, Campine, and Sandy regions, covering a total of 9272 km2. According to the soil map of the Flemish region and following the World Reference Base for Soil Resources (WRB) soil classification system [27], the dominant Reference Soil Groups per agricultural region considered here are as follows. (1) Sandy region: Cambisols, i.e., soils with a moderately developed profile due to the limited age of the soil material and which can be very productive, especially in loess areas; Anthrosols, i.e., soils under intensive agricultural use, with substantial additions of mineral and organic fertilizers; and Arenosols or sandy soils with limited soil formation. (2) Campine region: Podzols, which are acidic, mostly coarse-textured soils with a bleached horizon underlain by an accumulation of organic matter, aluminum, and iron; and Anthrosols. (3) Sandy Loam region: Luvisols, i.e., soils with a subsurface horizon of high-activity clay accumulation and high base saturation; Retisols, which are soils with a clay-enriched subsoil; and Fluvic Cambisols formed on sediment deposits in alluvial valleys.

2.2. Data Sources

LSA for maize was based on publicly available geospatial data on land cover, soil, climate, and topography. All data were retrieved in the form of spatial raster datasets from the following online sources: (a) Copernicus Land Monitoring Service (ver.3), Land Cover Change Version 3.0 product at 100 m resolution; (b) European Digital Elevation Model (EU-DEM), version 1.1; (c) SoilGrids at 250 m resolution based on a global compilation of soil profile data (WoSIS) and environmental layers [28]; (d) Climate Data from the Copernicus Climate Data Store on the land component of the fifth generation of the European Re-Analysis (ERA5), referred to as ERA5-Land [29]; (e) TerraClimate dataset of monthly climate and climatic water balance for global terrestrial surfaces from 1958 to 2020 [30]; and (f) CCAFS-Climate data from the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) global and regional climate datasets of future high-resolution and bias-corrected CMIP5 projections [31].
The selection process of environmental variables as representative factors affecting land suitability for maize was based on an extensive review of published agricultural LSA studies [2], which highlights the usefulness and importance of a wide range of factors related to soil, climatic, and topographic characteristics for LSA applications. Data availability determined the final extraction of input factors from data sources, which included critical soil physicochemical properties, basic climatic variables, and the underlying topography. All the available geospatial data involved in LSA, processed in an R computational environment (, accessed on 11 December 2023), were harmonized to a common spatial resolution of 250m analysis cell, corresponding to 6.25 ha in local UTM projection, using standard resampling procedures (bilinear interpolation). For LSA purposes, SoilGrids were preprocessed by combining the first four layers of available information, integrating values over the soil depth up to 60cm, which corresponds to the effective rooting depth over an analysis cell to which the crop can take up water and nutrients, rather than the maximum rooting depth of a single plant [32]. The climatic variables of monthly precipitation and monthly average and minimum temperature during the local growing season of maize (May–August) were averaged over a 20-year period (2001–2020) from the available time series of climatic data.
The comprehensive list of geospatial data utilized for LSA, in addition to CCAFS-Climate data, grouped in the main factor categories of soil, climate, and topography, is presented in Table 1, together with their statistical distribution concerning the analysis mask of the study area, i.e., the agricultural land cover (additional information regarding current monthly mean temperature during the growing season, are provided as supplementary material).

2.3. LSA Application

Consecutive LSA applications were required to evaluate SICSs as a long-term measure to mitigate impacts due to climate change on land suitability. Initially, the LSA was used to assess current land suitability for maize and identify the main limiting factors. Subsequent LSA scenarios included the assessment of future land suitability under climate change scenarios and the application of SICSs. The entire LSA application focused on the three dominant agricultural regions (Sandy Loam, Campine, and Sandy regions) (Figure 1).

2.3.1. Current LSA

The current LSA was carried out to evaluate land suitability for sustainable maize production and identify the key limiting factors that compromise maize cultivation based on the criteria of soil, climate, and topography. Based on the LSA in the same region [33], expert knowledge from local information sources was incorporated through AHP, utilizing weighting schemes to express the relative importance of each parameter or category in the final assessment of land suitability. The FAO framework for LSA [8] was employed to assess current land suitability for maize production, classifying suitable land into three categories: highly suitable (S1); moderately suitable (S2), with additional inputs required; and marginally suitable (S3), with significant inputs needed. Threshold values for input parameters related to maize suitability were adopted from previous work in the region [33] for the selected soil properties from SoilGrids, namely CEC, SOC, pH, texture, and coarse fragments (see supplementary material). Soil textural classes were based on the USDA classification scheme. Climatic characteristics, i.e., monthly precipitation and temperature, were computed for the local growing cycle of maize. The LSA was applied using MCE techniques on spatial data regarding the physical environment of agricultural crop production, grouped as soil, climate, and topography factors, using R package ALUES [34]. The assessment of land suitability required the mapping of land characteristics into suitability classes of the target factor. Complete suitability scores were obtained using fuzzy rule inference based on different forms of membership functions (Triangular, Trapezoidal, and Gaussian), followed by Weighted Linear Combinations (WLCs) of individual scores, controlled by the AHP-generated weighting schemes. The LSA validation was based on the LUCAS microdata database, consisting of georeferenced point samples collected during land cover surveys available online at (accessed on 25 September 2023).

2.3.2. Future LSA

The future LSA aims to assess land suitability for sustainable maize production in future time periods. Assuming consistent soil management over time, the focus is on evaluating the impact of future climate change on agricultural maize production. Various methods exist for projecting climate change impacts on agriculture, given the challenge of integrating local-scale agricultural models with coarse GCM projections. Most agricultural impact models require high-resolution environmental data, since agricultural and natural landscapes exhibit high spatial variability, with local climate (especially precipitation), soils and crop management, varying over small distances. To address this, downscaling procedures and bias correction of climate model output are commonly used to generate local projections of climate change and its impacts [31]. In this study, future climate assessment was based on previously downscaled and bias-corrected GCM projections, available from the CCAFS-Climate global database [31], that was the source of high-resolution (30 arcsec ≈ 1 km at the Equator) bias corrected climate projections. For its development, bias correction has been applied to 35 Coupled Model Intercomparison Project Phase 5 (CMIP5) models [35], and four RCPs. For this study, only a subset of the fully developed database was considered, both in terms of available GCM projections and RCP scenarios. The selection of the 10 GCMs appearing in Table 2, was based on previous results [36], regarding the evaluation of CMIP5 GCMs for multi-region downscaling purposes, based on the large-scale performance of the models in the European region, excluding those that are considered unrealistic, while eliminating models with poor performance (detailed statistics regarding projected climate variables over the study area from the chosen GCMs, are provided as supplementary material).
RCP scenarios are based on the concept of future radiative forcing for climate modeling, as a quantitative measure of the combined effect of GHGs, aerosols, and other factors that can influence climate to trap additional heat [45]. Based on the availability of projections from the selected GCMs, 2 out of the 4 available RCPs were selected for the future LSA: (1) RCP4.5, a scenario that stabilizes the anthropogenic component of radiative forcing, 4.5 W.m−2, by the year 2100, without ever exceeding this value (low-emissions scenario); and (2) RCP8.5, the scenario with the highest GHG emissions, 8.5 W.m−2, by the year 2100. The two scenarios differ substantially not only in the projected GHG emissions to 2100 but also in the scenario’s storyline that describes the driving forces behind these emissions. The RCP8.5 storyline describes a heterogeneous world with a steadily increasing global population, higher energy demands, and intensive agricultural production in the absence of policies to limit climate change impacts, whereas RCP4.5 follows a stabilization pathway, assuming that climate policies, such as the current global price on GHG emissions, are fully implemented to achieve limited emissions and associated radiative forcing. The selected GCM-RCP climate projections were available for four 30-year future periods, namely 2030s (2020–2049), 2050s (2040–2069), 2070s (2060–2089), and 2080s (2070–2099), for the climate variables of interest: monthly precipitation and monthly average and minimum temperature during the local growing season of maize (May–August).

Future LSA Computation

The assessment of the climate change impact on land suitability is considered within the context of consistent soil characteristics through time, with the climatic factor representing the primal motivating force of suitability changes. Consequently, the future LSA could be carried out in a manner equivalent to the current LSA, i.e., using AHP and MCE on multiple factors, by using the future climate projections generated by the GCMs as input data for climatic conditions. Inevitably, the inherent uncertainty of these projections is propagated into the processing chain of MCE for the LSA and manifests itself as uncertainty in the final suitability score. Thus, the LSA’s results were obtained individually from every single GCM projection, transforming in such a way all the available future climate projections into future land suitability projections. To emphasize the impact of climate change on suitability, the relative differences between current and future suitability scores were calculated for all LSA results on an analysis-cell basis. The latter were assembled into a multi-model ensemble (MME) of relative differences in suitability, providing, at the same time, an uncertainty assessment in terms of projected changes in suitability. The ensemble mean (EM) of the relative differences in suitability was computed on an analysis-cell basis for all RCP and future period combinations. These ensemble means represent the most probable change in maize suitability due to the impact of climate change. The variability around each ensemble mean, i.e., the ensemble standard deviation (ESD), can be interpreted as a measure of the uncertainty imposed on the suitability value by the future climate GCM projections.

Inter-Model Spread

In this study, the range of MME suitability projections with respect to a given emissions scenario is referred to as the inter-model spread, similarly to MMEs of future climate projections [46]. This spread is an effect of the differences between GCMs in the parameterization and structure of climate model components. In the case of MMEs assembled from future climate projections, examining the variability within the MME is the central approach to quantifying uncertainty, separating, at the same time, the individual contributions of model components to the overall uncertainty [13]. Still, an analysis of variance (ANOVA) approach, like the methodology suggested by Yip et al. (2011) [47] on MMEs of climate projections, could not be directly applied to MMEs of suitability projections due the fundamental difference of spatial dependence inherent in the latter. This is a direct result of the LSA process, which transformed the future climate projections into future suitability projections, using soil and landscape characteristics, thus reproducing their intrinsic spatial dependence in the derived suitability projections. Statistical inference under the appearance of spatial dependence can lead to biased results in regression analysis. Besides with geostatistics, where the modeling of spatial dependence is used for spatial prediction or interpolation, the residual spatial dependence can be modeled using regressors from a spatial relationship matrix (eigenvectors) that act as explanatory variables [48]. Eigenvector Spatial Filtering (ESF) methods [49] model spatial dependence by using eigenvectors based on Moran coefficients (MCs) and provide residual variables that represent spatially independent components of variation. Based on this approach, the total variability in change projection can be decomposed into the filtered variance components of the involved modeling terms, which represent spatially independent sources of variation. The statistical model is formulated as Equation (1):
Variability in change projection = RCP + GCM + FP + GCM:RCP + GCM:FP + RCP:FP + S + e
where RCP represents the variance component that is associated with the GHG emission scenario and GCM represents the variance component related to modeling the climate’s physical processes. The future period (FP) variation represents the long-term temporal trend over the projection period, while the interaction terms between the three modeling factors represent the non-additive pairwise combinations between them; e.g., the “GCM:RCP” term arises from how GCMs react differently to given emissions scenarios. The spatially uncorrelated error term (e) represents the unexplained variability in suitability projections, expressing the internal variability of the climate system due to its randomly varying nature, which is inherent in GCM projections. The involved modeling terms “RCP”, “GCM”, and “FP” and their interactions in Equation (1) refer to the filtered suitability projections rather than the initial projections. Therefore, the difference between the initial and the filtered projections that represents the spatially dependent component of variation (S) is finally introduced into Equation (1) as an additional independent variable. The regression modeling was carried out separately for each agricultural region due to the computational intensity of the ESF process that estimates the spatially dependent component (S), which is O(n3), with n being the size of the dataset. Computational approximations suitable for big-data processing were required to perform a dimensionality reduction to the modeling of spatial dependence. Following the memory-free approach on random-effects ESF modeling [50], suitable for very large datasets, implemented in R package spmoran [51], an efficient elimination of residual spatial dependence was realized by using a spatial process that is interpretable in terms of the MC (Moran I statistic, [52]).

Future LSA under SICS

The future LSA with SICS builds on the assessment of climate change impacts on land suitability for maize by additionally adopting the scenario of the SICS application within the time span of the CCAFS future climate projections, i.e., 2020–2099. The application of the organic matter-specific SICS could be considered a mitigation measure against the impact of future climate change on land suitability for maize. This potential is investigated by applying the Crop Residue Management (CRM) scenario from the available SICS for a period of 80 years and examining the resulting differences in suitability. To take the SICS scenario into consideration, based on the previous results in the same study area [33,53,54] and published reviews on the effect of CRM in long-term experiments [55,56], the impact of applying 80 years (2020–2099) of CRM practices was calculated as an increase in SOC at a rate of 200 kg/ha per year. This perspective was incorporated into the LSA procedures by calculating new values for SOC and CEC. A factor of 1.72 was used to convert soil organic matter values to SOC values, while the value of CEC value for soil organic matter was assumed to be 200 cmol/kg. The LSA was performed on the last 30-year period of the CGM projections, i.e., the 2080s (2070–2099), by repeating the future-LSA process, this time with the improved SOC and CEC values, and thus generating new projected changes in suitability. To determine the impact of the SICS on the most probable changes in suitability score due to climate change, two cases of results were generated: SICS application (CRM) or not (n.a.) to be directly comparable on an analysis-cell basis.

3. Results

3.1. Current LSA

For the current LSA case, we adopted the AHP approach to the relative importance of input factor groups and parameters [33] of LSA for maize in the same region. The resulting weighting scheme for individual factors and factor groups was adopted throughout the entire processing chain of LSA computations included in this study. The weighted suitability scores were computed for soil, climate, and topography factor groups in the three main agricultural regions. The climate factor group received the lowest suitability scores across the region, while the topography group scored the highest suitability values. Soil and climatic factors were identified by AHP as being equally important than topography and provided weights that were used directly in a WLC of the individual factor-group scores. This resulted in an assessment of the overall suitability for maize in the FAO land evaluation context, constrained by the geographical extent of the main agricultural regions (Figure 2a). The distribution of the FAO suitability classes (Figure 2b) shows that the S3 class is predominant, with over 87% (8062 km2) of the total area of the three main agricultural regions in Flanders. In contrast, the S2 suitability class (where S2 refers to higher suitability than S3) covers only 13% of this area (1210 km2). These are concentrated in areas that are imperfectly drained, often due to a clay-enriched subsoil.
The validation was performed using LUCAS microdata, including 165 georeferenced points with maize cover across the entire Flanders area; they were mostly identified as S3 class (139/165) locations, with the rest (26/165) identified as S2 (Figure 2c).

3.2. Limiting Factors Assessment

The source of the lowest suitability score on an analysis-cell basis is presented in Figure 3 for the three main agricultural regions in Flanders. It shows the limiting factors for maize suitability, considering all factors involved. Figure 4, on the other hand, focuses specifically on the soil factors, highlighting the limitations within the geographical extent of the three main agricultural regions. The primary limiting factor for maize in Flanders is climate, mainly precipitation, while the soil pH is the major reason for the lowest score within the soil factor group.
Precipitation dominates the limiting factors, whether it is the precipitation of the growing cycle or the precipitation of each month, especially August (Figure 3). Texture limits productivity in the north–central and northeastern parts (Figure 4a). SOC limits LS in most of the southern part, followed by CEC (Figure 4a); lower CEC values in these areas probably reflect the weighted averaging of CEC layers of up to 60 cm. The distribution of soil-limiting factors according to agricultural region is presented in Figure 4b. The Sandy and Campine regions are dominated by pH mainly and texture limitations, while the Sandy Loam region is dominated by pH, SOC, and CEC limitations.

3.3. Future LSA under Climate Change

An LSA was carried out on all the available climate projections (10 GCMs, 2 RCPs, and 4 future periods), generating 80 distributions of future suitability projections, accordingly. The relative differences in suitability were combined with the three main agricultural regions for maize cultivation in Flanders, i.e., Sandy, Sandy Loam, and Campine. The box-plot distributions in Figure 5 present the percentage of change in suitability from all LSA runs due to climate change impacts during four future periods, up to 2099, under two RCPs scenarios, and by the geographical extent of the selected agricultural regions. The group medians of the relative differences, from each of the 10 GCMs for each case of RCP scenario and future period are also displayed. The lines connecting the group medians between the two RCPs for each future period form a linear slope that represents the decline in suitability score (% score reduction) from RCP4.5 to RCP8.5 (transition towards a more adverse scenario, i.e., when shifting from a low-emissions to a high-emissions scenario). Across the entire time span of future suitability projections, especially for the RCP8.5 scenario, a trend for reduction in projected suitability is evident from the 2030s to the 2080s. Based on the graphical representation of the percentage of change in suitability score in Figure 5, it is apparent that not all the models perform equally; MOHC.HADGEM2 consistently has the highest percentage score reduction, while the CESM1.CAM5 consistently has the lowest percentage score reduction and sometimes increases. As anticipated, models from the same research center tend to have a similar variation in the projected suitability, a characteristic typical of the dependence between GCMS from the same family of models (GFDLs and MOHCs). Moreover, upon careful examination of the boxplots, differences in models for the future LSA can be spotted between the three main agricultural regions and across the decades. Although all models show almost the same constant performance in the 2030s, a change is expected in the 2050s: the Sandy region is anticipated to experience the most significant (negative) impact, whereas the suitability of the Sandy Loam and Campine regions appears to be comparatively less affected. Notably, during the periods of 2070s and 2080s, all the regions considered are projected to experience the largest (negative) changes in suitability scores.
The distribution of the full ensemble of projected suitability changes is displayed in Figure 6, computed as the EM of all the available suitability projections, regardless of RCP, future period, or GCM. The projected suitability changes are decomposed into three different modeling components: (a) RCP, (b) future period, and (c) GCM. The contribution of each modeling component to the entire ensemble of suitability projections becomes less ambiguous; the shift in projected suitability from RCP4.5 to RCP8.5 is evident (Figure 6a), and the long-term decreasing trend in suitability projections from the 2030s to the 2080s is visualized (Figure 6b). The substantial deviation between the projected suitability changes from different GCMs is apparent in Figure 6c.
The tabular representation of the entire MME projections exceeded 1.19 × 107 records. The computationally demanding process of ESF on suitability projections was carried out separately for each region by specifying the corresponding kernel function for modeling spatial dependence and the number of involved Moran eigenvectors. The modeling choices are presented in Table 3, along with the recorded computational times. The resulting scaled MC (Moran I/max (Moran I)) is a scale indicator of the detected spatial dependence, taking large positive values in the presence of large-scale spatial dependence, and small values on the small scale exhibited spatial dependence. Based on the results of the scaled MC in Table 3, strong medium-scale spatial dependence was detected in all three regions of interest. This scale of dependence lies in between the small scale of spatial dependence inherent in soil and topography data and the large scale of climatic data dependence and is the synthetic result of the LSA process.
The ANOVA results, performed separately for each region, are presented in Table 4. The GCM model explained a proportion of approximately 20–26% of the variation in projected suitability changes. The future period also contributed to the total variance, with proportions of approximately 15–20%. The total variance in suitability projections was least affected by RCPs (approximately 7–11%) compared with the abovementioned factors. From the interaction terms, GCM:future period made the largest contribution to the total variance (approximately 5–10%), followed by RCP:future period and GCM:RCP terms, which had a contribution of 3–6% approximately. All of these contributions represent the independent non-spatial components of variation, whereas the spatially dependent component of variation (S) is expressed by the filtered-out variance from the ESF. The greatest contribution of spatial dependence occurs in the Sandy Loam region, where the extended inter-model spread is related to the increased variability of soil characteristics, whose natural outcome is the appearance of spatial autocorrelation in the soil attributes used as input variables in the LSA process. The residual or unexplained variance (e) accounted for approximately 14–22% of the total variance in the entire MME of suitability projections, maximizing in the Sandy region. The regression models explained a proportion of approximately 78–86% of the total variability in suitability projections.
The response of suitability to climate change for each RCP scenario and future period involved is derived from the multi-model ensemble means (EMs), and the uncertainty components are extracted from the dispersion between individual responses. The EMs of relative changes in suitability represent the most probable change in land suitability for each case of RCP scenario and future period involved. Combined with the agricultural regions of interest in Flanders, the box-plot distributions in Figure 7 (left) represent the EM of projected suitability changes due to climate change impact during four future periods, up to 2099, under two RCPs scenarios, and by the extent of the agricultural regions of interest. Statistical inference was possible after filtering out the spatial dependence from the EMs, which reached a total of 1.18672 × 106, using similar ESF procedures with the full MME projections case (scaled MC = 0.552). The differences in the filtered means of these distributions between the three regions for each case of RCP and future period were statistically significant for all the pairwise comparisons (p-value ≤ 0.0001), with a small-to-moderate effect size. There is a gradual decrease in suitability scores over time for both RCP scenarios. However, under RCP4.5, the decrease is noticeable until the 2050s, and then it becomes very gentle. On the contrary, under RCP8.5, the decrease is continuous until the 2080s. Across the three regions of interest, the Sandy region shows the highest decreases consistently over the decades and RCP scenarios, closely followed by Campine.
Accordingly, the box-plot distributions in Figure 7 (right) display the uncertainty associated with the mean of the most probable changes in land suitability, i.e., the ensemble standard deviation (ESD), for each case of RCP scenario and future period involved, and by the extent of the agricultural regions of interest. Similarly, statistical inference was possible after filtering out the spatial dependence (scaled MC = 0.635). The differences in the filtered means of these distributions between the three regions for each RCP case and future period were also statistically significant for all the pairwise comparisons (p-value ≤ 0.0001) with a small-to-moderate effect size. The ensemble standard deviation is expected to double after the 2030s for both RCPs. Across the regions of interest, the Sandy region has the highest standard deviation, except for Campine in the 2080s, under RCP8.5.
The spatial representation of the EMs of projected suitability changes due to climate change impact is displayed in Figure 8. The highest decrease in the suitability score is observed in the southeast and central parts of Flanders. The lowest values (<−8%) cover most of the area in the 2070 and 2080s under RCP8.5. On the contrary, an increase of suitability is observed at the eastern and western parts of the study area, while most of the study area is under minimal change (>−4%) in the 2030s. Statistics for the last future period, the 2080s, under the high-emissions scenario RCP8.5 (Table 5), indicate an average reduction in projected suitability of approximately 7% by 2099, with minor differences between the three agricultural regions of interest. The global maximum reduction in suitability across all three agricultural regions is approximately 10%.
The spatial representation of the associated uncertainty is displayed in Figure 9. Very low uncertainties are observed in the 2030s for both RCPs, while the highest are observed in the western and northeast areas for most of the examined scenarios.

3.4. Future LSA under SICS

LSA was carried out on all the available climate projections (10 GCMs and 2 RCPs) for the last future period of 2080s, using the updated soil attributes from SICS application, generating this way 20 corresponding distributions of future suitability projections. The ensemble means of the relative differences were computed for each case of RCP scenario and combined with the three agricultural regions of interest. The box-plot distributions in Figure 10, represent the EMs of the projected suitability changes due to climate change impact, for each case of RCP scenario and region involved, concerning two cases of distributions: CRM application up until 2099, or not (n.a.). Statistical inference was possible after filtering out the spatial dependence from the EMs using similar ESF procedures with the full MME projections. The filtered means of these distributions were significantly different (p-value ≤ 0.0001), with a moderate-to-large effect size, in all three agricultural regions of interest for both RCP scenarios.
The spatial representation of the EMs of projected suitability differences due to the impact of climate change, mitigated by the application of CRM for 80 years, is displayed in Figure 11. A slight decrease in the suitability score is observed in the central north parts of Flanders, while, in the south, west, and east borders, there is an increase. Between the two RCPs, the largest reductions in magnitude and extent are observed in RPC8.5.
The actual improvement in the projected suitability change due to CRM application for 80 years (2020–2099) is essentially a quantitative measure of the SICS’s potential to mitigate the impact of climate change on land suitability for maize. This measure of the CRM mitigation potential can be derived by computing the differences between the projected changes under the SICS depicted in Figure 11 and the corresponding projections displayed in Figure 8 for the last future period of the 2080s that represent a quantitative measure of climate change impact without mitigation, both expressed as relative changes in suitability. Likewise, the ensemble ESD in Figure 9 represents the degree to which individual suitability projections differ from the EM, also expressed in terms of relative suitability changes. These three quantitative criteria, averaged over the selected agricultural regions, are presented in Table 6.

4. Discussion

The results from the current LSA for maize suggest that suitability for maize cultivation is marginal for most of the agricultural land in the investigated regions. In Flanders, both grain and silage maize are considered essential crops for animal farming. Due to the large number of livestock in the area, maize production is extensive. Climate, and mainly precipitation, was recognized as the main limiting factor, with extreme wet or dry periods negatively affecting yields [57,58]. The impact of climate change on land suitability for maize was quantified in terms of relative score reduction, using MMEs of projected suitability differences, as generated by several GCM climate projections transformed into suitability projections through the LSA process. The significant potential of Crop Residue Management (CRM) SICS to mitigate the impact of future climate change on land suitability for maize was evident in all three agricultural regions and under both RCPs.
The relative differences in the suitability projections under SICS revealed disparities between the three agricultural regions; under both RCP scenarios, the Sandy Loam region benefits the most from the 80-year CRM-SICS application up to 2099, while the Sandy region benefits the least. This difference between the two regions is explained by the distribution of soil-limiting factors across the entire Flanders area (Figure 4). Overall, pH, SOC, and CEC appear as soil-limiting factors in the Sandy Loam region, favored by the application of organic matter-specific SICSs. The vulnerability assessment of agroecosystems to drought and heavy rainfall also pointed to the importance of farming practices and, in particular, soil characteristics such as soil organic carbon to alleviate extreme weather impacts [59]. Based on the comparison between uncertainty and mitigation gain (Table 5), the Sandy Loam region is most probable to benefit from CRM application for 80 years, considering that the mitigation gain exceeds the associated uncertainty from GCM ensemble modeling, unlike the Campine and Sandy regions.

4.1. Data-Related Implications

LSA relies on spatial data on soil, climate, and landscape characteristics (Table 1), which have inherent uncertainties, associated with either model prediction errors or inaccurate representations of the physical environment. SoilGrids [60] is the result of global predictions for standard soil properties based on ML methods, using several remote sensing-based soil covariates, mostly MODIS land products, and an extensive database of soil profiles for training purposes. The distribution of soil profiles greatly affects the accuracy of the derived predictions, as reported by a recent study [61] in Central FR, concerning the accuracy of SoilGrids at the local level. The study area of Flanders has a very high density coverage of soil profiles (228 profiles per 1000 km2) [62], with more than 7000 profiles covering the whole of BE. Using the uncertainty layers that accompany SoilGrids predictions [60], prediction uncertainties were derived for specific soil properties and averaged over the study area. The averaged errors and their standard deviations were 0.56 for PH (sd = 0.1), 2.23 cmol/kg for CEC (sd = 0.32), 5.58 g/kg for SOC (sd = 2.3), and 4.12% for CRFVOL (sd = 0.9). The magnitude of these errors, together with the very high density of soil profiles in the area, suggests that SoilGrids provided a reliable representation of the soil characteristics for LSA purposes and that no significant biases were introduced into the process. Concerning climate data, monthly averages for the variables of interest were extracted for the 20-year period of 2001–2020 from the available time series. TerraClimate uses climatologically assisted interpolation on different global gridded climate datasets to provide higher spatial-resolution (1/24°) climatic variables with lower temporal resolution (monthly), showing a significant improvement in overall mean absolute error compared to the original coarser-resolution gridded datasets [30]. ERA5-Land combines gridded climatic data and observations, using physical modeling to describe the evolution of the water and energy cycles over land during long time periods, covering both past and present climatic conditions, with enhanced spatial resolution (1/10°). Validation against independent in situ observations and global reference datasets has shown the added value of ERA5-Land in describing the hydrological cycle, while highlighting the existence of significant precipitation biases, especially in the tropical regions [29]. To minimize the introduction of bias into the LSA process, climate data on monthly precipitation and minimum temperature were extracted from the TerraClimate time series, while ERA5-Land was used only for the extraction of monthly average temperature.

4.2. LSA Implications

According to recent review of published LSA studies [2], the adopted approach of the AHP-MCE application on spatial environmental factors induces several limitations related to data consistency and quality, bias from analytical procedures, and uncertainty involved in the selection of criteria. Limitations are either inherent to the use of AHP methods, implying contradictions with expert judgments, or they are related to data quality issues. Constraints related to the reliability of the LSA results usually arise from the involvement of expert opinion in the LSA process, frequently used during input parameter selection and the weighting importance of individual factors. Expert opinion on the relative importance of input factors on land suitability was processed through AHP, using pairwise comparison of alternative outcomes, and transformed into individual or grouped factor weights that were directly integrated into the LSA process. In the AHP application, the pairwise comparison of factors is based on subjective expert opinions that can impact the final judgement with personal preferences and inaccuracies in the assignment of weights [2]. Although the AHP technique is an efficient way to determine the weights of multiple factors systematically in a logical manner, by simplifying the rating of preferences among decision criteria, the downside remains in the subjectivity of expert opinion that may overlook possible interrelations between the involved factors [63]. Usually, during the involvement of stakeholders in participatory planning, the experts fulfil the essential role of technical advisors in the complex modeling process of LSA [64].

4.3. Future-Climate Modeling Implications

As reported by the Intergovernmental Panel on Climate Change (IPCC) in a previous report (AR5), the global mean surface temperature is likely to increase by between 2.6 and 4 K under the RCP8.5 warming scenario by the end of the 21st century relative to the end of the 20th century [65]. The significant spread of the global temperature change projection reflects the uncertainty from the involved climate models that ultimately cascades into future land suitability projections. Based on a relative study [66] that reports on the temperature regime and its projected changes under CMIP5 modeling, these models, on average, show a cold bias in winter, especially in northern Europe. The GCMs also overestimate summer temperatures in Central Europe, associated with a larger diurnal range than observed. Previous studies have argued that the projected climatic changes, particularly the expected rise in mean temperature, could be proven advantageous for agricultural production in Central and Northern Europe [67]. Still, reports from IPCC emphasize the concurrent influence of other climate change impacts, such as the increase in the magnitude and frequency of extreme weather events [24]. Although findings of temperature increases seem to be favorable for maize in the Flanders area, the overall suitability is likely to decrease due to the decrease in the projected precipitation in the same area. The very likely increase in drought stress necessitates the use of soil and water conservation practices, drought-tolerant crop varieties, or irrigation in the more distant future [68].
Concerning the modeling terms and their impact on the projected changes in suitability, the ANOVA results (Table 4) indicate that changes in suitability are most influenced by the choice of the specific GCMs from the entire collection of CMIP5 climate models. On the contrary, the projected changes are least influenced by the choice of RCP (GHG emission scenario). The relatively low-to-medium residual error variance in the results indicates either that (a) the response of the projected changes in suitability for a given scenario and future period varies across GSMs only to a certain degree, suggesting a partial dependence between the involved models; or (b) the internal variability of the climate system expressed in residual variability is sufficient to counterbalance, up to a certain point, the influence of GCM on suitability projections.
The dependence between different GCMs projections is partly a result of the selection process during the initialization of LSA. Three models were selected from the NOAA Geophysical Fluid Dynamics Laboratory (GFDL models) and two from the UK Met Office Hadley Centre (MOHC-HadGEM2 models) (Table 2). Dependence between GCMs from the same research facility is not unexpected, since these models share hypotheses and codes in the physical modeling of climate processes. Moreover, MMEs are rarely assembled from independent climate models; different modeling centers may contribute alongside several models in the CMIP ensemble, sharing, at the same time, identical modeling components. The main argument is that the EMs represent more confident projections than single model-based projections and that model independence improves the accuracy of the ensemble means [46]. From this perspective, convergence of future suitability projections, leading to reduced dispersion between models, is not a direct indicator of the robustness of the involved models, as it may come because of common biases or other undesired dependences between models. Similarly, the divergence of future suitability projections, leading to an increased inter-model spread, is an indication of actual model independence, which is a preferable feature for more reliable statistical EMs.

4.4. SICS-Related Implications

The prospect of mitigating climate impacts on land suitability for maize through the application of CRM is part of the long-term benefits of SICS for enhancing soil quality and carbon storage. The successful adoption of SICS over long periods of time is a complicated process that is controlled not only by the efficiency and economy of agricultural management practices (CRM) but is also influenced by several other factors, including social characteristics, cultural features, and policy-related issues [4]. Once a positive environmental and soil quality impact is attained, recording at the same time a positive balance between production costs and profits, then the dual objectives of SICS, i.e., farm profitability and environmental sustainability, are reached, making the application of the SICS sustainable.
In addition, the modeling approach for soil carbon storage is based on the results of long-term field experiments (LTEs) that are commonly used in the assessment of SOC stocks in agroecosystems. In LTEs, the effect of agricultural practices and cropping systems on SOC stock changes is estimated using effect size indices that are relative to a reference treatment. They are expressed usually as stock change rates (SCRs, kg C ha−1 year−1) and are considered important for evaluating policy measures and providing guidelines for relevant research activities. The SCR value of 200 kg C ha−1 year−1 that was adopted in this study for the whole period of CRM application is an approximation in order to simulate the effect of CRM on SOC stock changes. The results from LTEs indicate that, even though the response of SOC stocks in different management practices is more immediate during the initial phase of their application, positive SCRs are still sustained for several more decades [55]. Moreover, the effects of CRM practices on SOC stocks, derived from LTEs, only relate to studies conducted at the field scale of experimental plots. Upscaling the results of LTEs from the field to the regional scale is not always a straightforward process. A regional impact of CRM practices, or influence at a regional scale, is possible if SOC losses due to soil erosion, residue removal, or losses occurring in deep layers are considered and compensated for [55,69]. Appropriate methods are measures that stimulate photosynthetic rates per unit area and time, targeting net primary productivity, as well as the allocation of carbon to belowground plant tissues through appropriate agricultural management, leading to the enrichment of SOC stocks. Similarly, the combined effect of several management practices applied together, such as CRM, conservation tillage, and the application of organic amendments, promotes SOC stock changes [69].

5. Conclusions

A Land Suitability Analysis (LSA), according to the FAO framework, was used to evaluate the current suitability for maize in Flanders, Belgium, and the effectiveness of Soil-Improving Cropping Systems (SICS) as a long-term mitigation measure against climate change impacts on land suitability for maize. Two different scenarios of future climate change were studied, RCP4.5 and RCP8.5, which represent two different future climates based on the projected GHG emissions under different socioeconomic storylines, i.e., low- and high-emissions scenario. The impact of climate change on suitability was assessed through a future LSA, using MME of projected suitability changes, developed from GCMs future climate projections. Combined with a scenario of future soil conditions based on the CRM-SICS application for a period of 80 years, up to 2099, the findings of this study allow the following conclusions to be drawn:
  • The marginal suitability for maize (mostly S3 class) resulting from the current Land Suitability Analysis (LSA) showed that climate, and specifically precipitation, is the major limiting factor for suitability.
  • The potential of improved Crop Residue Management (CRM-SICS) to mitigate the impact of future climate change on suitability was investigated by applying the CRM-SICS scenario to the future LSA process and was found to be significant under both RCPs adopted (low or high emissions scenarios) and in all the three main agricultural regions.
  • Among the main agricultural regions, the Sandy Loam region shows the strongest response to the long-term application of CRM-SICS for a period of 80 years, up to 2099.
The proposed methodology presents an effective framework for applying an LSA involving future climate projections and SICS to promote sustainable agricultural practices that ensure the efficient management of land resources. These findings offer crucial insights into adopting agricultural practices for better land management. Improved CRM and reduced tillage practices should be promoted, considering their valuable role in moderating soil degradation and increasing soil productivity. Soil carbon storage makes a significant contribution to these functions and can be proved essential in mitigating and adapting to global climate change. The Flanders area is characterized by intense agricultural land use, where the presence of cropping systems and the abundancy of animal manure, combined with a rigorous environmental policy concerning nutrient management, has resulted in soil organic matter decline, leading to increased soil erosion and compaction. Therefore, based on the results of the current study, soil quality and soil-improving practices are strongly recommended to farmers to address possible issues due to climate change in the forthcoming years. Enhancing soil health through inter alia increasing the Zeffective organic carbon content of arable land through the application of SICS is considered one of the most promising approaches to climate change adaptation in the area.

Supplementary Materials

The following supporting information can be downloaded at:, LSA critical values for maize, monthly mean temperatures over the study area for the growing season (May–August), Climate projections from 10 GCMs, under 2 RCPs & for four 30-year future periods: 2030s (2020–2049), 2050s (2040–2069), 2070s (2060–2089) and 2080s (2070–2099), for the climate variables of interest over the study area.

Author Contributions

Conceptualization, N.K., A.G. and T.K.A.; methodology, N.K., G.B., A.G. and T.K.A.; software, N.K. and V.P.; validation, A.G. and E.K.; all authors participated in writing—review and editing; visualization, N.K.; supervision, T.K.A. All authors have read and agreed to the published version of the manuscript.


This work was funded by EU-H2020 project ‘Resilient farming by adaptive microclimate management’ (STARGATE—818187).

Data Availability Statement

Data will be available upon request.

Conflicts of Interest

The authors declare no conflict of interest.


  1. Mugiyo, H.; Chimonyo, V.G.P.; Sibanda, M.; Kunz, R.; Masemola, C.R.; Modi, A.T.; Mabhaudhi, T. Evaluation of Land Suitability Methods with Reference to Neglected and Underutilised Crop Species: A Scoping Review. Land 2021, 10, 125. [Google Scholar] [CrossRef]
  2. Akpoti, K.; Kabo-bah, A.T.; Zwart, S.J. Agricultural Land Suitability Analysis: State-of-the-Art and Outlooks for Integration of Climate Change Analysis. Agric. Syst. 2019, 173, 172–208. [Google Scholar] [CrossRef]
  3. Aertsens, J.; De Nocker, L.; Gobin, A. Valuing the Carbon Sequestration Potential for European Agriculture. Land Use Policy 2013, 31, 584–594. [Google Scholar] [CrossRef]
  4. Hessel, R.; Wyseure, G.; Panagea, I.S.; Alaoui, A.; Reed, M.S.; van Delden, H.; Muro, M.; Mills, J.; Oenema, O.; Areal, F.; et al. Soil-Improving Cropping Systems for Sustainable and Profitable Farming in Europe. Land 2022, 11, 780. [Google Scholar] [CrossRef]
  5. Magdoff, F.; Van Es, H. Building Soils for Better Crops: Ecological Management for Healthy Soils; Sustainable Agriculture Research & Education (Program), National Institute of Food and Agriculture (U.S.): Beltsville, MD, USA, 2021; ISBN 9781888626193. [Google Scholar]
  6. Wang, X.; He, C.; Liu, B.; Zhao, X.; Liu, Y.; Wang, Q.; Zhang, H. Effects of Residue Returning on Soil Organic Carbon Storage and Sequestration Rate in China’s Croplands: A Meta-Analysis. Agronomy 2020, 10, 691. [Google Scholar] [CrossRef]
  7. Malczewski, J. GIS-Based Land-Use Suitability Analysis: A Critical Overview. Prog. Plan. 2004, 62, 3–65. [Google Scholar] [CrossRef]
  8. FAO. Land Evaluation towards a Revised Framework Land Evaluation towards a Revised Framework Land Evaluation towards a Revised Framework. 2007. Available online: (accessed on 11 December 2023).
  9. Greene, R.; Devillers, R.; Luther, J.E.; Eddy, B.G. GIS-Based Multiple-Criteria Decision Analysis. Geogr. Compass 2011, 5, 412–432. [Google Scholar] [CrossRef]
  10. Bagherzadeh, A.; Ghadiri, E.; Souhani Darban, A.R.; Gholizadeh, A. Land Suitability Modeling by Parametric-Based Neural Networks and Fuzzy Methods for Soybean Production in a Semi-Arid Region. Model. Earth Syst. Environ. 2016, 2, 104. [Google Scholar] [CrossRef]
  11. Olsen, K.; Svenning, J.C.; Balslev, H. Climate Change Is Driving Shifts in Dragonfly Species Richness across Europe via Differential Dynamics of Taxonomic and Biogeographic Groups. Diversity 2022, 14, 1066. [Google Scholar] [CrossRef]
  12. Liambila, R.N.; Kibret, K. Climate Change Impact on Land Suitability for Rainfed Crop Production in Lake Haramaya Watershed, Eastern Ethiopia. J. Earth Sci. Clim. Chang. 2016, 7, 343. [Google Scholar] [CrossRef]
  13. Wallach, D.; Mearns, L.O.; Ruane, A.C.; Rötter, R.P.; Asseng, S. Lessons from Climate Modeling on the Design and Use of Ensembles for Crop Modeling. Clim. Chang. 2016, 139, 551–564. [Google Scholar] [CrossRef] [PubMed]
  14. Zhuang, X.W.; Li, Y.P.; Huang, G.H.; Liu, J. Assessment of Climate Change Impacts on Watershed in Cold-Arid Region: An Integrated Multi-GCM-Based Stochastic Weather Generator and Stepwise Cluster Analysis Method. Clim. Dyn. 2016, 47, 191–209. [Google Scholar] [CrossRef]
  15. Moss, R.H.; Edmonds, J.A.; Hibbard, K.A.; Manning, M.R.; Rose, S.K.; Van Vuuren, D.P.; Carter, T.R.; Emori, S.; Kainuma, M.; Kram, T.; et al. The next Generation of Scenarios for Climate Change Research and Assessment. Nature 2010, 463, 747–756. [Google Scholar] [CrossRef] [PubMed]
  16. Wang, B.; Zheng, L.; Liu, D.L.; Ji, F.; Clark, A.; Yu, Q. Using Multi-Model Ensembles of CMIP5 Global Climate Models to Reproduce Observed Monthly Rainfall and Temperature with Machine Learning Methods in Australia. Int. J. Climatol. 2018, 38, 4891–4902. [Google Scholar] [CrossRef]
  17. Bonfante, A.; Monaco, E.; Alfieri, S.M.; De Lorenzi, F.; Manna, P.; Basile, A.; Bouma, J. Climate Change Effects on the Suitability of an Agricultural Area to Maize Cultivation: Application of a New Hybrid Land Evaluation System. Adv. Agron. 2015, 133, 33–69. [Google Scholar] [CrossRef]
  18. Iizumi, T.; Ramankutty, N. How Do Weather and Climate Influence Cropping Area and Intensity? Glob. Food Secur. 2015, 4, 46–50. [Google Scholar] [CrossRef]
  19. Ramirez-Villegas, J.; Jarvis, A.; Läderach, P. Empirical Approaches for Assessing Impacts of Climate Change on Agriculture: The EcoCrop Model and a Case Study with Grain Sorghum. Agric. For. Meteorol. 2013, 170, 67–78. [Google Scholar] [CrossRef]
  20. Bassu, S.; Brisson, N.; Durand, J.L.; Boote, K.; Lizaso, J.; Jones, J.W.; Rosenzweig, C.; Ruane, A.C.; Adam, M.; Baron, C.; et al. How Do Various Maize Crop Models Vary in Their Responses to Climate Change Factors? Glob. Chang. Biol. 2014, 20, 2301–2320. [Google Scholar] [CrossRef]
  21. Kriticos, D.J.; Murphy, H.T.; Jovanovic, T.; Taylor, J.; Herr, A.; Raison, J.; O’Connell, D. Balancing Bioenergy and Biosecurity Policies: Estimating Current and Future Climate Suitability Patterns for a Bioenergy Crop. GCB Bioenergy 2014, 6, 587–598. [Google Scholar] [CrossRef]
  22. Ramirez-Cabral, N.Y.Z.; Kumar, L.; Shabani, F. Global Alterations in Areas of Suitability for Maize Production from Climate Change and Using a Mechanistic Species Distribution Model (CLIMEX). Sci. Rep. 2017, 7, 5910. [Google Scholar] [CrossRef]
  23. Nguyen, T.T.; Verdoodt, A.; Van, Y.T.; Delbecque, N.; Tran, T.C.; Van Ranst, E. Design of a GIS and Multi-Criteria Based Land Evaluation Procedure for Sustainable Land-Use Planning at the Regional Level. Agric. Ecosyst. Environ. 2015, 200, 1–11. [Google Scholar] [CrossRef]
  24. de Frutos Cachorro, J.; Gobin, A.; Buysse, J. Farm-Level Adaptation to Climate Change: The Case of the Loam Region in Belgium. Agric. Syst. 2018, 165, 164–176. [Google Scholar] [CrossRef]
  25. Geopunt Vlaanderen. Available online: (accessed on 17 October 2023).
  26. Gobin, A. Impact of Heat and Drought Stress on Arable Crop Production in Belgium. Nat. Hazards Earth Syst. Sci. 2012, 12, 1911–1922. [Google Scholar] [CrossRef]
  27. Dondeyne, S.; Vanierschot, L.; Langohr, R.; Van Ranst, E.; Deckers, J. The Soil Map of the Flemish Region Converted to the 3rd Edition of the World Reference Base for Soil Resources. 2014. Available online: (accessed on 11 December 2023).
  28. Hengl, T.; De Jesus, J.M.; Heuvelink, G.B.M.; Gonzalez, M.R.; Kilibarda, M.; Blagotić, A.; Shangguan, W.; Wright, M.N.; Geng, X.; Bauer-Marschallinger, B.; et al. SoilGrids250m: Global Gridded Soil Information Based on Machine Learning. PLoS ONE 2017, 12, e0169748. [Google Scholar] [CrossRef] [PubMed]
  29. Muñoz-Sabater, J.; Dutra, E.; Agustí-Panareda, A.; Albergel, C.; Arduini, G.; Balsamo, G.; Boussetta, S.; Choulga, M.; Harrigan, S.; Hersbach, H.; et al. ERA5-Land: A State-of-the-Art Global Reanalysis Dataset for Land Applications. Earth Syst. Sci. Data 2021, 13, 4349–4383. [Google Scholar] [CrossRef]
  30. Abatzoglou, J.T.; Dobrowski, S.Z.; Parks, S.A.; Hegewisch, K.C. TerraClimate, a High-Resolution Global Dataset of Monthly Climate and Climatic Water Balance from 1958–2015. Sci. Data 2018, 5, 170191. [Google Scholar] [CrossRef]
  31. Navarro-Racines, C.; Tarapues, J.; Thornton, P.; Jarvis, A.; Ramirez-Villegas, J. High-Resolution and Bias-Corrected CMIP5 Projections for Climate Change Impact Assessments. Sci. Data 2020, 7, 7. [Google Scholar] [CrossRef]
  32. Zhang, L.; Liu, G.; Yang, Y.; Guo, X.; Jin, S.; Xie, R.; Ming, B.; Xue, J.; Wang, K.; Li, S.; et al. Root Characteristics for Maize with the Highest Grain Yield Potential of 22.5 Mg Ha−1 in China. Agriculture 2023, 13, 765. [Google Scholar] [CrossRef]
  33. Bilas, G.; Karapetsas, N.; Gobin, A.; Mesdanitis, K.; Toth, G.; Hermann, T.; Wang, Y.; Luo, L.; Koutsos, T.M.; Moshou, D.; et al. Land Suitability Analysis as a Tool for Evaluating Soil-Improving Cropping Systems. Land 2022, 11, 2200. [Google Scholar] [CrossRef]
  34. Asaad, A.-A.B.; Salvacion, A.R.; Yen, B.T. ALUES: R Package for Agricultural Land Use Evaluation System. J. Open Source Softw. 2022, 7, 4228. [Google Scholar] [CrossRef]
  35. Taylor, K.E.; Stouffer, R.J.; Meehl, G.A. An Overview of CMIP5 and the Experiment Design. Bull. Am. Meteorol. Soc. 2012, 93, 485–498. [Google Scholar] [CrossRef]
  36. McSweeney, C.F.; Jones, R.G.; Lee, R.W.; Rowell, D.P. Selecting CMIP5 GCMs for Downscaling over Multiple Regions. Clim. Dyn. 2015, 44, 3237–3260. [Google Scholar] [CrossRef]
  37. Xin, X.; Zhang, L.; Zhang, J.; Wu, T.; Fang, Y. Climate Change Projections over East Asia with BBC_CSM1.1 Climate Model under RCP Scenarios. J. Meteorol. Soc. Jpn. 2013, 91, 413–429. [Google Scholar] [CrossRef]
  38. Bi, D.; Dix, M.; Marsland, S.J.; O’Farrell, S.; Rashid, H.A.; Uotila, P.; Hirst, A.C.; Kowalczyk, E.; Golebiewski, M.; Sullivan, A.; et al. The ACCESS Coupled Model: Description, Control Climate and Evaluation. Aust. Meteorol. Oceanogr. J. 2013, 63, 41–64. [Google Scholar] [CrossRef]
  39. Hurrell, J.W.; Holland, M.M.; Gent, P.R.; Ghan, S.; Kay, J.E.; Kushner, P.J.; Lamarque, J.F.; Large, W.G.; Lawrence, D.; Lindsay, K.; et al. The Community Earth System Model: A Framework for Collaborative Research. Bull. Am. Meteorol. Soc. 2013, 94, 1339–1360. [Google Scholar] [CrossRef]
  40. Donner, L.J.; Wyman, B.L.; Hemler, R.S.; Horowitz, L.W.; Ming, Y.; Zhao, M.; Golaz, J.C.; Ginoux, P.; Lin, S.J.; Schwarzkopf, M.D.; et al. The Dynamical Core, Physical Parameterizations, and Basic Simulation Characteristics of the Atmospheric Component AM3 of the GFDL Global Coupled Model CM3. J. Clim. 2011, 24, 3484–3519. [Google Scholar] [CrossRef]
  41. Dunne, J.P.; John, J.G.; Shevliakova, E.; Stouffer, R.J.; Krasting, J.P.; Malyshev, S.L.; Milly, P.C.D.; Sentman, L.T.; Adcroft, A.J.; Cooke, W.; et al. GFDL’s ESM2 Global Coupled Climate-Carbon Earth System Models. Part II: Carbon System Formulation and Baseline Simulation Characteristics*. J. Clim. 2013, 26, 2247–2267. [Google Scholar] [CrossRef]
  42. Martin, G.M.; Bellouin, N.; Collins, W.J.; Culverwell, I.D.; Halloran, P.R.; Hardiman, S.C.; Hinton, T.J.; Jones, C.D.; McDonald, R.E.; McLaren, A.J.; et al. The HadGEM2 Family of Met Office Unified Model Climate Configurations. Geosci. Model Dev. 2011, 4, 723–757. [Google Scholar] [CrossRef]
  43. Reick, C.H.; Raddatz, T.; Brovkin, V.; Gayler, V. Representation of Natural and Anthropogenic Land Cover Change in MPI-ESM. J. Adv. Model. Earth Syst. 2013, 5, 459–482. [Google Scholar] [CrossRef]
  44. Gent, P.R.; Danabasoglu, G.; Donner, L.J.; Holland, M.M.; Hunke, E.C.; Jayne, S.R.; Lawrence, D.M.; Neale, R.B.; Rasch, P.J.; Vertenstein, M.; et al. The Community Climate System Model Version 4. J. Clim. 2011, 24, 4973–4991. [Google Scholar] [CrossRef]
  45. van Vuuren, D.P.; Edmonds, J.; Kainuma, M.; Riahi, K.; Thomson, A.; Hibbard, K.; Hurtt, G.C.; Kram, T.; Krey, V.; Lamarque, J.F.; et al. The Representative Concentration Pathways: An Overview. Clim. Chang. 2011, 109, 5–31. [Google Scholar] [CrossRef]
  46. Jebeile, J.; Barberousse, A. Model Spread and Progress in Climate Modelling. Eur. J. Philos. Sci. 2021, 11, 66. [Google Scholar] [CrossRef]
  47. Yip, S.; Ferro, C.A.T.; Stephenson, D.B.; Hawkins, E. A Simple, Coherent Framework for Partitioning Uncertainty in Climate Predictions. J. Clim. 2011, 24, 4634–4643. [Google Scholar] [CrossRef]
  48. Murakami, D.; Griffith, D.A. Random Effects Specifications in Eigenvector Spatial Filtering: A Simulation Study. J. Geogr. Syst. 2015, 17, 311–331. [Google Scholar] [CrossRef]
  49. Chun, Y.; Griffith, D.A. A Quality Assessment of Eigenvector Spatial Filtering Based Parameter Estimates for the Normal Probability Model. Spat. Stat. 2014, 10, 1–11. [Google Scholar] [CrossRef]
  50. Murakami, D.; Griffith, D.A. A Memory-Free Spatial Additive Mixed Modeling for Big Spatial Data. Jpn. J. Stat. Data Sci. 2020, 3, 215–241. [Google Scholar] [CrossRef]
  51. Murakami, D. Spmoran: An R Package for Moran’s Eigenvector-Based Spatial Regression Analysis (Version 2017/06). 2017. Available online: (accessed on 11 December 2023).
  52. Anselin, L.; Rey, S. Properties of Tests for Spatial Dependence in Linear Regression Models. Geogr. Anal. 1991, 23, 112–131. [Google Scholar] [CrossRef]
  53. Van De Vreken, P.; Gobin, A.; Baken, S.; Van Holm, L.; Verhasselt, A.; Smolders, E.; Merckx, R. Crop Residue Management and Oxalate-Extractable Iron and Aluminium Explain Long-Term Soil Organic Carbon Sequestration and Dynamics. Eur. J. Soil Sci. 2016, 67, 332–340. [Google Scholar] [CrossRef]
  54. Nadeu, E.; Gobin, A.; Fiener, P.; van Wesemael, B.; van Oost, K. Modelling the Impact of Agricultural Management on Soil Carbon Stocks at the Regional Scale: The Role of Lateral Fluxes. Glob. Chang. Biol. 2015, 21, 3181–3192. [Google Scholar] [CrossRef]
  55. Bolinder, M.A.; Crotty, F.; Elsen, A.; Frac, M.; Kismányoky, T.; Lipiec, J.; Tits, M.; Tóth, Z.; Kätterer, T. The Effect of Crop Residues, Cover Crops, Manures and Nitrogen Fertilization on Soil Organic Carbon Changes in Agroecosystems: A Synthesis of Reviews. Mitig. Adapt. Strateg. Glob. Chang. 2020, 25, 929–952. [Google Scholar] [CrossRef]
  56. Peng, Y.; Rieke, E.L.; Chahal, I.; Norris, C.E.; Janovicek, K.; Mitchell, J.P.; Roozeboom, K.L.; Hayden, Z.D.; Strock, J.S.; Machado, S.; et al. Maximizing Soil Organic Carbon Stocks under Cover Cropping: Insights from Long-Term Agricultural Experiments in North America. Agric. Ecosyst. Environ. 2023, 356, 108599. [Google Scholar] [CrossRef]
  57. Zamani, S.; Gobin, A.; Van de Vyver, H.; Gerlo, J. Atmospheric Drought in Belgium—Statistical Analysis of Precipitation Deficit. Int. J. Climatol. 2016, 36, 3056–3071. [Google Scholar] [CrossRef]
  58. Gobin, A.; Van de Vyver, H. Spatio-Temporal Variability of Dry and Wet Spells and Their Influence on Crop Yields. Agric. For. Meteorol. 2021, 308–309, 108565. [Google Scholar] [CrossRef]
  59. Vanwindekens, F.M.; Gobin, A.; Curnel, Y.; Planchon, V. New Approach for Mapping the Vulnerability of Agroecosystems Based on Expert Knowledge. Math. Geosci. 2018, 50, 679–696. [Google Scholar] [CrossRef]
  60. Poggio, L.; De Sousa, L.M.; Batjes, N.H.; Heuvelink, G.B.M.; Kempen, B.; Ribeiro, E.; Rossiter, D. SoilGrids 2.0: Producing Soil Information for the Globe with Quantified Spatial Uncertainty. SOIL 2021, 7, 217–240. [Google Scholar] [CrossRef]
  61. Richer-de-Forges, A.C.; Arrouays, D.; Poggio, L.; Chen, S.; Lacoste, M.; Minasny, B.; Libohova, Z.; Roudier, P.; Mulder, V.L.; Nédélec, H.; et al. Hand-Feel Soil Texture Observations to Evaluate the Accuracy of Digital Soil Maps for Local Prediction of Soil Particle Size Distribution: A Case Study in Central France. Pedosphere 2023, 33, 731–743. [Google Scholar] [CrossRef]
  62. Batjes, N.H.; Ribeiro, E.; Van Oostrum, A. Standardised Soil Profile Data to Support Global Mapping and Modelling (WoSIS Snapshot 2019). Earth Syst. Sci. Data 2020, 12, 299–320. [Google Scholar] [CrossRef]
  63. Zhang, J.; Su, Y.; Wu, J.; Liang, H. GIS Based Land Suitability Assessment for Tobacco Production Using AHP and Fuzzy Set in Shandong Province of China. Comput. Electron. Agric. 2015, 114, 202–211. [Google Scholar] [CrossRef]
  64. Brandt, P.; Kvakić, M.; Butterbach-Bahl, K.; Rufino, M.C. How to Target Climate-Smart Agriculture? Concept and Application of the Consensus-Driven Decision Support Framework “TargetCSA”. Agric. Syst. 2017, 151, 234–245. [Google Scholar] [CrossRef]
  65. Hu, X.; Fan, H.; Cai, M.; Sejas, S.A.; Taylor, P.; Yang, S. A Less Cloudy Picture of the Inter-Model Spread in Future Global Warming Projections. Nat. Commun. 2020, 11, 4472. [Google Scholar] [CrossRef]
  66. Cattiaux, J.; Douville, H.; Peings, Y. European Temperatures in CMIP5: Origins of Present-Day Biases and Future Uncertainties. Clim. Dyn. 2013, 41, 2889–2907. [Google Scholar] [CrossRef]
  67. Trnka, M.; Olesen, J.E.; Kersebaum, K.C.; Skjelvåg, A.O.; Eitzinger, J.; Seguin, B.; Peltonen-Sainio, P.; Rötter, R.; Iglesias, A.; Orlandini, S.; et al. Agroclimatic Conditions in Europe under Climate Change. Glob. Chang. Biol. 2011, 17, 2298–2318. [Google Scholar] [CrossRef]
  68. Zhao, J.; Bindi, M.; Eitzinger, J.; Ferrise, R.; Gaile, Z.; Gobin, A.; Holzkämper, A.; Kersebaum, K.C.; Kozyra, J.; Kriaučiūnienė, Z.; et al. Priority for Climate Adaptation Measures in European Crop Production Systems. Eur. J. Agron. 2022, 138, 126516. [Google Scholar] [CrossRef]
  69. Aguilera, E.; Lassaletta, L.; Gattinger, A.; Gimeno, B.S. Managing Soil Carbon for Climate Change Mitigation and Adaptation in Mediterranean Cropping Systems: A Meta-Analysis. Agric. Ecosyst. Environ. 2013, 168, 25–36. [Google Scholar] [CrossRef]
Figure 1. Flanders (BE) agricultural regions 2023. Source: [25].
Figure 1. Flanders (BE) agricultural regions 2023. Source: [25].
Land 13 00063 g001
Figure 2. Overall suitability assessment: FAO classification with distribution and validation in the three agricultural regions: (a) FAO suitability classes in the region; (b) relative frequencies (areas in Km2) for each of the FAO classes identified; and (c) validation for both FAO classes identified based on 165 georeferenced points with maize cover across the Flanders region using LUCAS microdata.
Figure 2. Overall suitability assessment: FAO classification with distribution and validation in the three agricultural regions: (a) FAO suitability classes in the region; (b) relative frequencies (areas in Km2) for each of the FAO classes identified; and (c) validation for both FAO classes identified based on 165 georeferenced points with maize cover across the Flanders region using LUCAS microdata.
Land 13 00063 g002
Figure 3. Limiting factors for land suitability in the three regions of interest in Flanders.
Figure 3. Limiting factors for land suitability in the three regions of interest in Flanders.
Land 13 00063 g003
Figure 4. Soil-limiting factors for land suitability in Flanders (a), with distribution by agricultural region (b).
Figure 4. Soil-limiting factors for land suitability in Flanders (a), with distribution by agricultural region (b).
Land 13 00063 g004
Figure 5. Box plot distributions of future suitability projections by region of interest (plot by ggpubr).
Figure 5. Box plot distributions of future suitability projections by region of interest (plot by ggpubr).
Land 13 00063 g005
Figure 6. Distributions of projected suitability changes. The density estimate (Gaussian kernel) of the full ensemble’s distribution is identical in every panel (black line) and is decomposed into individual distributions from modeling components, according to RCP (a), future period (b), and GCM (c).
Figure 6. Distributions of projected suitability changes. The density estimate (Gaussian kernel) of the full ensemble’s distribution is identical in every panel (black line) and is decomposed into individual distributions from modeling components, according to RCP (a), future period (b), and GCM (c).
Land 13 00063 g006
Figure 7. Box-plot distributions of EM (left) and ESD (right) of projected suitability differences by agricultural region of interest.
Figure 7. Box-plot distributions of EM (left) and ESD (right) of projected suitability differences by agricultural region of interest.
Land 13 00063 g007
Figure 8. Spatial representation of the EMs of the most probable changes in land suitability due to the impact of climate change.
Figure 8. Spatial representation of the EMs of the most probable changes in land suitability due to the impact of climate change.
Land 13 00063 g008
Figure 9. Spatial representation of the uncertainty associated with the EMs of the projected suitability changes due to the impact of climate change.
Figure 9. Spatial representation of the uncertainty associated with the EMs of the projected suitability changes due to the impact of climate change.
Land 13 00063 g009
Figure 10. Box-plot distributions of EMs of relative differences between applied and not applied CRM cases, by agricultural region.
Figure 10. Box-plot distributions of EMs of relative differences between applied and not applied CRM cases, by agricultural region.
Land 13 00063 g010
Figure 11. Spatial representation of the projected suitability changes under two RCP scenarios, mitigated by 80 years of CRM application.
Figure 11. Spatial representation of the projected suitability changes under two RCP scenarios, mitigated by 80 years of CRM application.
Land 13 00063 g011
Table 1. LSA input data statistics per agricultural region.
Table 1. LSA input data statistics per agricultural region.
CampineSandySandy Loam
Raster DataMinMaxMeansdMinMaxMeansdMinMaxMeansd
BD 1.15 1.54 1.43 0.05 1.12 1.55 1.41 0.048 1.23 1.63 1.47 0.047
CEC 5.51 28.8 12.7 2.44 9.84 35.1 19.1 4.87 9.74 32.7 14.5 2.86
CRFVOL1.8 18.6 6.71 1.88 3.47 20.8 6.55 1.67 4.04 23 8.88 3.3
CLAY2.57 32.6 15 6.07 7.29 37.9 21.1 4.31 10.8 37.7 22.9 2.9
PH3.88 7.97 5.27 0.272 3.91 8.27 5.92 0.414 3.95 8.83 6.36 0.516
SAND11.2 93.7 57.3 18.8 10.5 82.1 45.4 10.6 8.75 71.4 27.9 9.07
SILT2.95 70.2 27.7 13.8 9.93 69.1 33.4 7.83 16.5 73.8 49.1 8.56
SOC6.33 77.8 16.8 6.13 5.2 116 19.1 8.09 3.68 68.5 12.5 4.72
MAY.PREC55 70 62.2 2.39 53 66 60.6 2.26 49 70 61.5 3.84
JUN.PREC70 82 75.9 2.08 61 82 72.9 3.48 56 80 71.5 4.71
JUL.PREC63 81 72.6 3.09 62 74 68.1 1.52 56 78 67.5 4.24
AUG.PREC54 73 62.3 3.5 53 71 60.1 2.76 51 75 59.8 5.68
PREC.GC249 301 273 9.43 234 284 262 6.39 214 298 260 16.7
TEMP.AVG.GC15.9 17 16.2 0.179 15.2 16.9 16.1 0.4 14.9 17 16 0.464
TEMP.MIN.GC11.5 13.4 12.3 0.267 11.5 13.2 12.4 0.402 11.3 13.3 12.3 0.453
SLP0 15.6 0.502 0.757 0 11 0.482 0.726 0 13 1.64 1.53
BD: bulk density of the fine earth fraction (g/cm3). CEC: cation exchange capacity (cmol/kg). CRFVOL: % volumetric fraction of coarse fragments (>2 mm). CLAY: % fraction of clay particles (<0.002 mm). PH: pH(H2O), SAND: % fraction of sand particles (>0.05 mm). SILT: % fraction of silt particles (≥0.002 mm and ≤0.05 mm). SOC: organic carbon content in fine earth fraction (g/kg). MAY.PREC: May precipitation (mm). JUN.PREC: June precipitation (mm). JUL.PREC: July precipitation (mm). AUG.PREC: August precipitation (mm). PREC.GC: growing season May–August precipitation (mm). TEMP.AVG.GC: growing season May–Aug average monthly temperature (°C). TEMP.MIN.GC: growing season May–Aug average monthly minimum temperature (°C). SLP: topographic slope (%).
Table 2. Selected GCMs from the CCAFS-Climate global database.
Table 2. Selected GCMs from the CCAFS-Climate global database.
Model (Reference)InstituteRCP
2.6 4.5 6.0 8.5
BCC-CSM1.1(m) [37]Beijing Climate Center and China Meteorological Administration
CSIRO-ACCESS1.0 [38]Commonwealth Scientific and Industrial Research Organization (CSIRO) and Bureau of Meteorology (BOM), Australia
CESM1-CAM5 [39]National Science Foundation, U.S. Department of Energy, and National Center for Atmospheric Research (NCAR)
GFDL-CM3 [40]NOAA Geophysical Fluid Dynamics Laboratory
MOHC-HadGEM2-CC [42]UK Met Office Hadley Centre
MOHC-HadGEM2-ES [42]
MPI-ESM-LR [43]Max Planck Institute for Meteorology
NCAR-CCSM4 [44]US National Centre for Atmospheric Research
Table 3. ESF filtering of total variation in projected suitability changes by agricultural region of interest.
Table 3. ESF filtering of total variation in projected suitability changes by agricultural region of interest.
Region n (Cells) Total Response Variance 1 Spatial Dependence Kernel Moran Eigenvectors
Computational Time
(Minutes) 2
Scaled MC
Max (Moran.I)]
Campine 3.90 × 1069.355Exponential500430.6836812
Sandy 3.514 × 10610.76Gaussian500320.6570547
Sandy Loam 4.453 × 10611.42Spherical1000890.6463084
1 % change in suitability score. 2 Xeon E-2224 CPU—30GB RAM Workstation.
Table 4. Independent variance components of total variation in projected suitability changes by agricultural region of interest.
Table 4. Independent variance components of total variation in projected suitability changes by agricultural region of interest.
Region GCM RCP Future Period GCM:RCP GCM:Future Period RCP:Future Period Se Proportion (%) of Variance Explained
Campine 26.586.7719.704.959.795.7512.2214.2585.75
Sandy 27.0310.9116.193.735.723.5311.1221.7878.22
Sandy Loam 19.777.3114.792.955.303.8026.6819.3980.61
Table 5. MME statistics by agricultural region of interest (EM and ESD).
Table 5. MME statistics by agricultural region of interest (EM and ESD).
Future PeriodRCPAgricultural Regionn (Cells)MinMaxMedianQuant.1Quant.3IQRMean
Ensemble mean
2080s RCP8.5 Sandy Loam 55,664 −9.776 −1.483 −6.290 −7.233 −5.061 2.172 −6.012
2080s RCP8.5 Sandy 43,931 −10.06−3.357 −7.449 −8.071 −6.242 1.829 −7.152
2080s RCP8.5 Campine 48,745 −9.610 −3.083 −6.431 −7.379 −5.617 1.762 −6.406
Ensemble standard deviation
2080s RCP8.5 Sandy Loam 55,664 1.515 6.194 2.524 2.264 2.878 0.615 2.675
2080s RCP8.5 Sandy 43,931 1.389 5.607 2.121 1.883 2.663 0.780 2.265
2080s RCP8.5 Campine 48,745 1.761 5.102 3.061 2.366 3.338 0.972 2.897
Table 6. Relative changes in suitability (%) by the year 2099, averaged by agricultural regions.
Table 6. Relative changes in suitability (%) by the year 2099, averaged by agricultural regions.
RegionClimate Change ImpactUncertainty from GCMs Ensemble ModelingCRM Mitigation Gain
(80-Year Application)
Campine −6.4062.897 2.564
Sandy −7.1522.265 0.694
Sandy Loam −6.0122.675 3.421
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Karapetsas, N.; Gobin, A.; Bilas, G.; Koutsos, T.M.; Pavlidis, V.; Katragkou, E.; Alexandridis, T.K. Analysis of Land Suitability for Maize Production under Climate Change and Its Mitigation Potential through Crop Residue Management. Land 2024, 13, 63.

AMA Style

Karapetsas N, Gobin A, Bilas G, Koutsos TM, Pavlidis V, Katragkou E, Alexandridis TK. Analysis of Land Suitability for Maize Production under Climate Change and Its Mitigation Potential through Crop Residue Management. Land. 2024; 13(1):63.

Chicago/Turabian Style

Karapetsas, Nikolaos, Anne Gobin, George Bilas, Thomas M. Koutsos, Vasileios Pavlidis, Eleni Katragkou, and Thomas K. Alexandridis. 2024. "Analysis of Land Suitability for Maize Production under Climate Change and Its Mitigation Potential through Crop Residue Management" Land 13, no. 1: 63.

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop