Projected Changes in Agricultural Drought Severity and Spatial Patterns across Türkiye under RCP4.5 and RCP8.5 Scenarios

Journal of Biological and Environmental Sciences e-ISSN 1308-2019 https://dergipark.org.tr/en/pub/jbes http://www.uludag.edu.tr/jbes December/2025, 19(57), 35-47. Received: 20 June 2025 RESEARCH ARTICLE Accepted: 29 August 2025 https://doi.org/10.5281/zenodo.17568859 *Corresponding Author: 1 Ali Kaan YETİK, Niğde Ömer Halisdemir University, Faculty of Agricultural Sciences and Technologies, Department of Biosystems Engineering, Niğde, Türkiye, [email protected], ORCID:0000-0003-13728407 2 Burak Nazmi CANDOĞAN, Bursa Uludağ University, Faculty of Agriculture, Department of Biosystems Engineering, Bursa, Türkiye, [email protected] , ORCID: 0000-0001-9898-5685 Atıf/Citation: Yetik, A. K. ve Candoğan, B., N. 2025. Projected Changes in Agricultural Drought Severity and Spatial Patterns across Türkiye under RCP4.5 and RCP8.5 Scenarios. Journal of Biological and Environmental Sciences, 19(57), 35-47. 35 Projected Changes in Agricultural Drought Severity and Spatial Patterns across Türkiye under RCP4.5 and RCP8.5 Scenarios A Ali Kaan YETİK1*, Burak Nazmi CANDOĞAN2 Abstract: This study evaluates the future impacts of climate change on drought severity across Türkiye using high-resolution climate projections from three Global Climate Models (GFDL-ESM2M, HadGEM2-ES, MPIESM-MR) under RCP4.5 and RCP8.5 scenarios. Drought conditions were assessed using the 12-month Reconnaissance Drought Index (RDI-12), which integrates both precipitation and evapotranspiration (ETo), for historical (1971–2000), near-future (2023–2052), and distant-future (2069–2098) periods. The Mann-Kendall test was used to detect trends, while Sen’s slope estimator and spatial visualization techniques were applied to assess the magnitude and spatial distribution of changes. Results indicate a significant increase in drought frequency and severity, particularly under the RCP4.5 and RCP8.5 scenarios. Multi-model averages indicate a projected increase of up to 21% in extreme drought frequency (RDI ≤ –2) in the near future under RCP4.5. Central Anatolia, Eastern Mediterranean, and parts of Eastern Anatolia were identified as regions highly vulnerable to intensified drought. Spatial analyses revealed considerable regional heterogeneity, showing varying drought trends depending on the climate model and emission scenario. Overall, the study emphasizes the critical need for region-specific drought mitigation strategies and robust adaptation policies to safeguard agricultural productivity and water resource sustainability under anticipated climate change scenarios. Keywords: GCMs, GIS, RCP4.5, RCP8.5, spatial analysis, Türkiye A The study does not require approval from an ethics committee. The article has been prepared according to research and publication ethics. Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 36 Introduction Climate change poses a significant threat to global water resources, with far-reaching implications for agricultural sustainability, particularly in regions with vulnerable climatic and hydrological regimes. As global temperatures continue to rise, shifts in precipitation patterns, increased evapotranspiration, and altered hydrological cycles are expected to intensify the frequency and severity of drought events (IPCC, 2021). Among the various manifestations of climate-induced stressors, drought stands out as one of the most complex and least understood phenomena due to its gradual onset, multidimensional impacts, and spatial-temporal variability (Mishra and Singh, 2010; Mehdizadeh et al., 2020). Drought not only disrupts ecological balances but also jeopardizes food security, water availability, and socioeconomic stability, particularly in semi-arid and arid regions (Lesk et al., 2016). The need for robust drought monitoring systems has consequently led to the development of numerous drought indices. Among these, the Reconnaissance Drought Index (RDI), which integrates both precipitation and reference crop evapotranspiration (ETo), has gained prominence due to its ability to capture climatic water balance variations more comprehensively than precipitation-only indices such as the Standardized Precipitation Index (SPI) (Tsakiris et al., 2007; Mahmood et al., 2023). Türkiye is particularly susceptible to the impacts of climate change due to its diverse topography and transitional climatic characteristics between arid and humid zones. Several studies in Türkiye have applied various drought indices, particularly the RDI, to evaluate meteorological drought under historical climate conditions. Anlı (2014) used the RDI method to analyze drought trends in Southeastern Anatolia and reported increasing drought tendencies based on both parametric and non-parametric trend tests. Yüce and Eşit (2020) assessed drought events in the Ceyhan Basin using ten different indices, including RDI, SPI, and SPEI, and identified strong correlations among them across different time scales. Geyikli et al. (2022) compared SPI and RDI results in the Yeşilırmak Basin and concluded that while both indices yielded similar values during wet periods, RDI more frequently detected severe drought events. Katipoğlu et al. (2022) evaluated drought trends in the Euphrates Basin using multiple indices, including RDI, and applied Mann-Kendall and Modified Mann-Kendall tests, identifying increasing drought frequency across the region. Kesgin et al. (2024) calculated SPI, SPEI, and RDI at multiple time scales using data from 11 meteorological stations along the Mediterranean coast of Türkiye and showed regional drought variability between 1972 and 2020. Şimşek et al. (2024) examined meteorological drought across the Black Sea Region using SPI and RDI, reporting high correlation between the indices, with RDI identifying the most extreme drought risk values. Unlike previous studies, this research provides a comprehensive assessment of future drought severity and spatial variability across Türkiye by integrating three distinct Global Climate Models (GCMs) and two emission scenarios (RCP4.5 and RCP8.5). To the best of our knowledge, no prior study has investigated the combined effects of multiple GCMs and emission scenarios on drought dynamics at the national scale using the RDI methodology. This study aims to investigate the spatial and temporal evolution of drought across Türkiye under future climate conditions by applying RDI to high-resolution outputs of three Global Climate Models (GFDL-ESM2M, HadGEM2-ES, and MPI-ESM-MR) under RCP4.5 and RCP8.5 scenarios. By identifying future drought hotspots and evaluating shifts in drought severity, the study seeks to provide a scientific basis for climate-resilient agricultural planning and sustainable water resource management at the national scale. Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 37 Materials and Methods Study Site The study was conducted across the entire territory of Türkiye, a country characterized by diverse topography and climatic zones ranging from Mediterranean to continental. To ensure national-scale representativeness, meteorological data from 81 provinces were utilized. The geographical variability of Türkiye plays a crucial role in spatial drought dynamics, making it a suitable case for investigating the regional impacts of climate change on drought severity and trends. Dataset Description and Future Climate Scenarios This study evaluates drought severity across Türkiye using model-based climatic data derived from future climate projections. All datasets were generated and provided by the Turkish State Meteorological Service (TSMS). Specifically, projections from three Global Climate Models (GFDL-ESM2M, HadGEM2-ES, and MPI-ESM-MR) were dynamically downscaled to 20 km spatial resolution using the RegCM4.3.4 regional climate model (Akçakaya et al., 2015). Downscaled outputs were extracted for 81 locations, each representing a provincial center, to ensure national-scale spatial coverage. The reference period was defined as 1971-2000, while future projections were analyzed for the near future (2023-2052) and distant future (2069-2098). These projections were categorized into five scenario groups: RF (reference period), N-45 (near future under RCP4.5), N-85 (near future under RCP8.5), D-45 (distant future under RCP4.5), and D-85 (distant future under RCP8.5). Drought Index Calculation To assess agro-meteorological drought, the RDI was used. RDI considers both precipitation and ETo, offering a more comprehensive evaluation of drought severity compared to precipitation-only indices. Reference crop evapotranspiration was calculated using the FAO-56 Penman-Monteith method via the ETo Calculator software developed by FAO (Allen et al., 1998). This method was selected for its ability to incorporate a range of climatic variables (temperature, relative humidity, wind speed, and solar radiation), providing a physically based and widely accepted standard for ETo estimation in climate change studies. The standardized RDI values were calculated annually (12-month timescale) for each station using the DrinC software. As a first step, the climatic water balance (a0) was computed for each station and year using precipitation and ETo values (Equation 1). a0 𝑖𝑗 =∑ 𝑃𝑖𝑗 𝑘 𝑗=1 ∑𝐸𝑇0𝑖𝑗 𝑘 𝑗=1 (1) Where Pij and EToij denote the precipitation and reference crop evapotranspiration (in mm) for the jth month of year i, and k=12 for annual aggregation. To standardize the a0 values and enable temporal comparison of drought severity, a z-score transformation was applied, as described in Equation (2). 𝑅𝐷𝐼 =𝑦𝑘(𝑖)−  𝑦𝑘  𝜎𝑦𝑘 (2) Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 38 In this equation, yk(i) refers to the natural logarithm of a0 for year i, yk is the mean, and 𝜎 is the standard deviation of the yk series. This standardization enables the classification of drought severity into defined categories. Drought classification was based on the standardized RDI thresholds proposed by Nalbantis and Tsakiris (2009). These categories allow for consistent interpretation of drought severity across regions and time periods. The classification thresholds are presented in Table 1. Table 1. Classification of drought severity based on RDI values RDI Value Classification RDI≥ 0 No drought -1.0≤RDI<0 Mild drought -1.5≤RDI<-1.0 Moderate drought -2.0≤RDI<-1.5 Severe drought RDI<-2.0 Extreme drought Trend Analysis and Spatial Visualization To identify temporal trends in drought severity, the non-parametric Mann-Kendall test was applied, and trend magnitudes were estimated using Sen’s slope estimator (Gilbert 1987). Spatial interpolation of the RDI values was performed using the Inverse Distance Weighted (IDW) method in ArcGIS 10.2 software. The interpolation was based on outputs from 81 points representing each province, ensuring even administrative coverage across Türkiye. A power value of 2 and 12 neighboring points were used in the interpolation process, following standard practice. IDW was chosen for its computational simplicity and effectiveness in capturing local spatial variability, especially when high-resolution gridded data are unavailable (Rahman et al., 2024). Although geostatistical methods such as Kriging could better represent spatial autocorrelation in heterogeneous terrains, IDW was considered suitable for this study due to its transparent structure and practical applicability with regularly spaced input points. Nonetheless, the limitations associated with IDW, particularly in complex topographies, are acknowledged as a constraint on spatial representativeness. Results and Discussion Frequency and Severity of Drought Events The analysis of drought frequency using the RDI across Türkiye under different climate scenarios and models (GFDL, HadGEM, and MPI under RCP4.5 and RCP8.5) reveals significant regional differences and future drought patterns. The RDI values, categorized into varying drought severity classes, indicate distinct temporal and spatial distributions of drought intensity. These variations across Türkiye are visually represented in Figures 1-3, which show the full distribution of annual RDI values for all 81 provinces over each 30-year period and climate scenario, based on the outputs of the GFDL, HadGEM, and MPI models, respectively. Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 39 Figure 1. Distribution of annual Reconnaissance Drought Index (RDI) values across 81 provinces of Türkiye under different time-scenario combinations based on the GFDL model: Reference (RF), Nearfuture RCP4.5 (N-45), Near-future RCP8.5 (N-85), Distant-future RCP4.5 (D-45), and Distant-future RCP8.5 (D-85) For the GFDL model, the reference period (1971-2000) showed 69 extreme drought occurrences (RDI ≤ -2). Under the near-future RCP4.5 scenario (N-45), extreme droughts slightly decreased to 62 occurrences, while a further decrease to 56 was noted under the N-85 scenario (Figure 1). However, by the distant future (2069-2098), extreme drought events increased again, particularly under the D-85 scenario, reaching 67 occurrences. This fluctuation may reflect the non-linear interaction between climate variables under different scenarios, though further investigation is needed to identify the underlying causes (Amognehegn et al., 2023). Figure 2. Distribution of annual Reconnaissance Drought Index (RDI) values across 81 provinces of Türkiye under different time-scenario combinations based on the HadGEM model: Reference (RF), Near-future RCP4.5 (N-45), Near-future RCP8.5 (N-85), Distant-future RCP4.5 (D-45), and Distantfuture RCP8.5 (D-85) Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 40 For the HadGEM model, the frequency of extreme drought conditions notably increased from 44 events in the reference period to 67 under the N-45 scenario (Figure 2). A further increase to 71 occurrences was projected under D-85 conditions. These findings indicate that severe drought events are expected to become more frequent, particularly under high-emission scenarios, corroborating the increased vulnerability and potential intensification noted by Yüce and Eşit (2020) and Geyikli et al. (2022) for Türkiye. Figure 3. Distribution of annual Reconnaissance Drought Index (RDI) values across 81 provinces of Türkiye under different time-scenario combinations based on the MPI model: Reference (RF), Nearfuture RCP4.5 (N-45), Near-future RCP8.5 (N-85), Distant-future RCP4.5 (D-45), and Distant-future RCP8.5 (D-85) The MPI model depicted a more complex drought pattern. The reference period exhibited 54 extreme drought occurrences (Figure 3). In contrast, the N-45 scenario showed a substantial increase in extreme drought frequency to 73 occurrences, followed by a decrease to 45 under the N-85 scenario (RCP8.5). Similar observations were reported by Wubneh et al. (2024), who identified increased drought frequencies under intermediate emission scenarios due to pronounced regional climatic feedback. Trend Detection Using Mann-Kendall and Sen’s Slope Tests The temporal trends in annual RDI values were assessed using the non-parametric Mann-Kendall test, while the rate of change was estimated using Sen’s slope method. Figure 4 presents the spatial distribution of these results for all five scenarios (RF, N-45, N-85, D-45, D-85) based on the GFDL climate model. Downward and upward triangles indicate decreasing and increasing RDI trends, respectively. Shaded colors represent Sen’s slope values, and significance levels are marked at 1% (p≤0.01), 5% (p≤0.05), and 10% (p≤0.10). While the inclusion of one meteorological station per province ensures a balanced administrative representation across Türkiye, it does not necessarily guarantee full geographical representativeness, especially in regions characterized by complex topography and pronounced microclimatic gradients, such as the Taurus and Pontic mountain ranges. In sparsely monitored or mountainous areas, localized climatic variability may not be adequately captured by centrally located provincial stations. This spatial limitation may influence the precision of interpolated drought patterns, particularly in terrains with high elevation variability. Therefore, the results should be interpreted with this caveat in mind, and future studies would benefit from denser station networks or the incorporation of elevation-dependent climate interpolation techniques. Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 41 In the N-45 scenario (Figure 4a), a widespread increasing drought trend is observed across Central Anatolia and the eastern Mediterranean region. Statistically significant negative trends at the 1% and 5% levels were detected in provinces such as Kayseri, Adana, Karaman, and Gaziantep. The slope values in these areas indicate considerable drought intensification. Under the N-85 scenario (Figure 4b), negative trends became even more pronounced, especially in Eastern Anatolia. Provinces like Van and Malatya exhibited statistically significant negative Z-values at the 1% level, reflecting an alarming increase in drought severity under high-emission conditions. In the D-45 scenario (Figure 4c), negative trends continued to dominate Central Anatolia, with some regions in Marmara and the Western Black Sea showing slight positive slopes. However, most of these increases were statistically insignificant. For the D-85 scenario (Figure 4d), drought trends remained negative across much of the country. Although significant patterns were not widespread, strong trends were concentrated in Central Anatolia (particularly in Çankırı, Kırıkkale, and Karaman) where Q values indicated sharp declines in RDI. These findings align with previous regional drought studies (Anlı, 2014; Katipoğlu et al., 2022; Yılmaz, 2023), stating central vulnerability of Anatolia to intensified aridity. During the reference period (RF) (Figure 4e), the most severe drought trends were concentrated in the Marmara and Western Black Sea regions. These areas experienced strong negative slope values, indicating persistent long-term drying. Slightly positive trends observed in Van and Hatay were not statistically significant. Overall, the GFDL model projections indicate a consistent intensification of drought trends, especially under high-emission scenarios. Central and Southeastern Anatolia are projected to be the most vulnerable regions. Figure 4. Spatial distribution of Mann-Kendall Z-values and Sen’s slope (Q) under the GFDL model scenarios (A: Near-future RCP4.5, B: Near-future RCP8.5, C: Distant-future RCP4.5, D: Distant-future RCP8.5, E: Reference period) Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 42 The spatial distribution of Mann-Kendall Z-values and Sen’s slope (Q) under the HadGEM model is presented in Figure 5. In the N-45 scenario (Figure 5a), the HadGEM model projected more widespread drought intensification compared to GFDL. Statistically significant negative trends were identified across the entire Marmara Region and most provinces along the Black Sea coast, except for Rize, Trabzon, and Artvin. The Konya Basin also exhibited strong drought signals. Trend intensity decreased toward the eastern provinces. Under the N85 scenario (Figure 5b), the overall spatial extent of drought intensification weakened. Statistically significant trends were mainly limited to Van and Malatya, while western Türkiye showed largely insignificant changes. The D-45 scenario (Figure 5c) revealed a clear spatial divide. Provinces east of a hypothetical Sakarya-Antalya axis showed increasing drought trends, while the northwest (Thrace and northern Aegean) displayed slightly positive or non-significant trends. In D-85 (Figure 5d), no province showed statistically significant intensification at the 1% level. Ardahan exhibited a significant decrease in drought severity at the 5% level. Interestingly, Adana was the only province where a statistically significant reduction in drought (i.e., positive RDI trend) was detected despite the high-emission scenario. During the reference period (RF) (Figure 5e), drought trends were generally weak in both western and eastern Türkiye. However, local intensifications were observed in parts of Central Anatolia and the Middle Black Sea. These spatial patterns are consistent with previous findings by Geyikli et al. (2022), who emphasized the sensitivity of RDI to not only precipitation but also ETo, showing that variations in evapotranspiration may contribute to spatial drought differences; however, the exact mechanisms require more targeted physical analysis. Figure 5. Spatial distribution of Mann-Kendall Z-values and Sen’s slope (Q) under the HadGEM model scenarios (A: Near-future RCP4.5, B: Near-future RCP8.5, C: Distant-future RCP4.5, D: Distant-future RCP8.5, E: Reference period) Journal of Biological and Environmental Sciences Yetik & Candoğan 12/2025, 19(57) 43 Türkiye under five MPI scenarios. In the N-45 scenario (Figure 6a) drought intensity increases over most northern provinces, while parts of the south display weak positive Q values. No province reaches the %1 significance level; only Kırklareli shows a minor (10 %) upward trend, implying locally improving conditions. Under N-85 (Figure 6b) MPI exhibits the largest inter-scenario contrast: drought intensification spreads across almost the entire country except south-eastern Anatolia. The smallest (most negative) Q values occur along the Aegean coast, increasing east-ward, with modest positive trends (up to 0.04) in the south-east. For the D-45 scenario (Figure 6c) pronounced negative trends emerge in eastern provinces, whereas parts of south-western Aegean show slight, non-significant drought relief. The Mediterranean, Marmara, and Black Sea regions remain dominated by negative Q values. The high-emission D-85 scenario (Figure 6d) reinforces these patterns: significant RDI declines (i.e. stronger drought) appear in the Aegean-Marmara belt and the Eastern Black Sea. Conversely, Van, Hatay, and Şanlıurfa record weak but positive Z-scores, indicating isolated reductions in drought severity. During the reference period (RF) (Figure 6e) trends are largely negative in western and central Türkiye, becoming spatially heterogeneous toward the east. The highest positive Q values cluster around the Kars-Iğdır corridor, while the steepest drying trends occur on the eastern flank of Marmara. Collectively, MPI results suggest a potential sensitivity to emission pathways, although further multi-model comparison would be needed to confirm this behavior across different regions and timeframes. Figure 6. Spatial distribution of Mann-Kendall Z-values and Sen’s slope (Q) under the MPI model scenarios (A: Near-future RCP4.5, B: Near-future RCP8.5, C: Distant-future RCP4.5, D: Distant-future RCP8.5, E: Reference period)