Multi-model Ensemble Approach Research Articles

A multi-model ensemble approach for reservoir dissolved oxygen forecasting based on feature screening and machine learning

Dissolved oxygen (DO) concentration in aquatic systems plays a vital role in water aquaculture. An innovative approach that combines feature selection and ensemble learning to predict DO in aquatic ecosystems was proposed. Feature selection was first performed using Maximum Information Coefficient (MIC). Five machine learning algorithms were then employed to construct five hybrid-MIC models, including K-Nearest Neighbors (KNN), Backpropagation (BP) Neural Network, Long Short-Term Memory (LSTM), Kernel Ridge Regression (KRR), and Support Vector Regression (SVR). Finally, an ensemble-RF prediction model was built using Random Forests(RF). The main findings are as follows: (1) The MIC technique can effectively identify the key factors influencing DO. (2) The MIC significantly improves model performance. (3) The hybrid-MIC model was further improved by the ensemble-RF model, the average R2 and NSE were both as high as 0.99, and the average MAE and RMSE were decreased by 72 % and 64 %, respectively.

A comprehensive analysis of future solar energy potential variations using a CMIP6 multi-model ensemble approach in Colombia

A comprehensive analysis of future solar energy potential variations using a CMIP6 multi-model ensemble approach in Colombia

Assessment of CMIP6 GCMs for selecting a suitable climate model for precipitation projections in Southern Thailand

The selection of General Circulation Models (GCMs) is critical due to computational limitations and underlying uncertainties. This study provides a comprehensive assessment of three bias correction (BC) methods, namely the delta change method (DT), quantile mapping (QM), and empirical quantile mapping (EQM). Utilizing precipitation data from 30 observation stations across Southern Thailand, the evaluation encompasses five CMIP6 GCM models (CAMS-CSM1-0, CanESM5, CNRM-CM6-1, CNRM-ESM2-1, IPSL-CM6A-LR). Evaluation metrics, root mean square error (RMSE), mean absolute error (MAE), Pearson's correlation (r), index of agreement (d), and mean bias error (MBE) are employed to assess BC methods. Evaluation measures suggest that the DT method outperforms EQM and QM, with higher accuracy and lower errors (DT: RMSE = 3.61, MAE = 2.32; EQM: RMSE = 3.82, MAE = 3.70). Taylor diagrams show that CNRM-ESM2-1 has the highest correlation across sites (r = 0.36), albeit with a wider dispersion, while CanESM5 has a more balanced performance. Significant annual precipitation increases are projected for different spans 2021-30, 2061-70, and 2091–2100, particularly under SSP585, which will influence flood risk, water management, and climate adaptation. Future projections with SSP585 continuously projecting a larger probability of increased precipitation. The DT method is recommended for the downscaling of CMIP6 GCMs for precipitation projections in Southern Thailand, recognizing its superior performance. The study's findings provide a foundation for informed decision-making and adaptation planning in southern Thailand, urging policymakers to prioritize climate resilience and adaptation strategies while using a multi-model ensemble approach for robust climate forecasts.

A novel approach to a multi-model ensemble for climate change models: Perspectives on the representation of natural variability and historical and future climate

This study developed a novel approach that integrated climate model selection and multi-model ensemble (MME) construction to effectively represent model uncertainties and, consequently, improve consistency in the evaluation of changes to extreme rainfall in different scenarios. Our focus was to combine 10 regional climate model (RCM) simulations, forced by two global climate models (GCM), especially for estimation of design rainfall in a changing climate. We hypothesized that the natural variability and statistically higher moment attributes in the extreme rainfall simulations from RCMs were not fully preserved. Therefore, the MME approach could be more effective in climate change studies, largely due to the use of multiple climate models. First, an experimental study was proposed to validate the efficacy of the proposed modeling framework approach adopting L-moments to quantify relative importance among climate models and their use for representing natural variability in the MME construction. The proposed approach was then applied to climate change scenarios collected from multiple RCMs for the Han-River watershed during both historical (1981–2005) and future (2006–2 100) periods. The results showed that the climate model selection informed by natural variability demonstrated better performance, representing nearly identical distribution to the observed annual maximum rainfall (AMR) in the Han River watershed. The range of the selected scenario was relatively narrower than that of all the scenarios, and the change rate was more consistent with the limited zero crossing, reflecting improvement in both model performance and consistency over historical and future periods, respectively. The change rate of the MME under RCP8.5 appeared to be an approximately 20% increase for the near (2011–2040) and far (2071–2 100) future, and the degree of increase in the rate for the mid-future (2041–2070) was slightly lower than that for the other periods, with increase of approximately 10%.

Ensembles of multiple models for soil moisture retrieval from remote sensing data over agricultural areas: A deep learning-based framework

In agricultural areas, most surface soil moisture (SM) retrieval models are unstable in terms of their accuracy and performance during crop growth. As a result, there is no consensus on which model performs optimally during the agricultural season. This is because of the uncertainties associated with model physics, initial conditions, model inversion processes, input data, vegetation attenuation and soil characteristics. To better deal with these practical concerns, we propose a simple, but robust SM retrieval method of using combination of multiple models based on deep learning and multi-model ensemble approach, called the DL-MME method, which makes use of the ‘collective intelligence’ and ‘wisdom of crowds’ concepts. The advantages of this method are: (1) robustness to model selection, and (2) robustness to model calibration during the growing season. In addition, this method is less dependent on one type of data across various agricultural areas compared to the single model approach. Firstly, the coupled water cloud model (WCM) and soil backscattering models (Oh model or advanced integral equation model (AIEM)) with different vegetation descriptors were calibrated and validated during the growing season in sugarcane and winter wheat fields for Sentinel-1 backscattering coefficients (VV and VH). SM was also retrieved by employing the trapezoid model (OPTRAM) with different parameters from Sentinel-2 images. To optimize SM retrieval computations, we used the outputs from optical and SAR models, auxiliary features, and reliable in situ SM measurements as inputs to a deep learning convolutional neural network (DL-CNN). For sugarcane and wheat fields in the early stages of crop growth, WCM models retrieved more accurate time-series SM than optical models. OPTRAM soil moisture retrievals showed greater accuracy in the late crop growing season. Time-series SM retrieval accuracy using DL-MME was higher than for the optical and semi-empirical SAR models. According to the results of the in situ validation for wheat (sugarcane) fields, the minimum MAE by an optimal combination of models was around 0.01 (0.02) m3m−3 (RMSE = 0.036 (0.074) m3m−3; R = 0.87 (0.71)). The findings demonstrate that our method is reliable and feasible for SM retrieval. Additionally, our method provides a way to select an optimal model for retrieving time-series SM during the crop growing season.

Assessing Climate Change Impacts on Streamflow and Baseflow in the Karnali River Basin, Nepal: A CMIP6 Multi-Model Ensemble Approach Using SWAT and Web-Based Hydrograph Analysis Tool

Our study aims to understand how the hydrological cycle is affected by climate change in river basins. This study focused on the Karnali River Basin (KRB) to examine the impact of extreme weather events like floods and heat waves on water security and sustainable environmental management. Our research incorporates precipitation and temperature projections from ten Global Circulation Models (GCMs) under the Coupled Model Intercomparison Project Phase 6 (CMIP6). We applied thirteen statistical bias correction methods for precipitation and nine for temperatures to make future precipitation and temperature trend projections. The research study also utilized the Soil and Water Assessment Tool (SWAT) model at multi-sites to estimate future streamflow under the Shared Socioeconomic Pathway (SSP) scenarios of SSP245 and SSP585. Additionally, the Web-based Hydrograph Analysis Tool (WHAT) was used to distinguish between baseflow and streamflow. Our findings, based on the Multi-Model Ensemble (MME), indicate that precipitation will increase by 7.79–16.25% under SSP245 (9.43–27.47% under SSP585) and maximum temperatures will rise at rates of 0.018, 0.048, and 0.064 °C/yr under SSP245 (0.022, 0.066, and 0.119 °C/yr under SSP585). We also anticipate that minimum temperatures will increase at rates of 0.049, 0.08, and 0.97 °C/yr under SSP245 (0.057, 0.115, and 0.187 °C/yr under SSP585) for near, mid, and far future periods, respectively. Our research predicts an increase in river discharge in the KRB by 27.12% to 54.88% under SSP245 and 45.4% to 93.3% under SSP585 in different future periods. Our finding also showed that the expected minimum monthly baseflow in future periods will occur earlier than in the historical period. Our study emphasizes the need for sustainable and adaptive management strategies to address the effects of climate change on water security in the KRB. By providing detailed insights into future hydrological conditions, this research serves as a critical resource for policymakers and stakeholders, facilitating informed decision-making for the sustainable management of water resources in the face of climate change.

Multi-model ensemble of frost risks across East Asia (1850–2100)

Frost events can cause the deterioration of a wide range of heritage materials, including stone, brick and earth. In a warming world, the frequency and location of frost events is likely to change, affecting the conservation strategies required at heritage sites. We use a multi-model ensemble approach to investigate three types of frost events in East Asia: freeze–thaw cycles; deep frost days and wet frosts. The study uses nine CMIP6 models for the period 1850 to 2100, with future projections run under the SPS585 scenario. Additional analysis is undertaken for five specific 2° ✕ 2° areas located across East Asia. The three frost event parameters are spatially and temporally distinct. A decrease in all three frost parameters is found in Japan, South Korea and East China, with some areas projected to have no frost events by the end of the twenty-first century. However, Northwest China is distinctive as wet frosts are projected to increase over the twenty-first century, while on the Tibetan plateau of Southwest China, freeze–thaw cycles are projected to increase. This suggests that except in some localised regions, heritage managers can focus on risks other than frost weathering in developing plans to address climate change.Graphical

Forecasting some climate parameters of Türkiye using the SSP3-7.0 scenario for the years 2040–2059

This study employs the Coupled Model Inter-comparison Projects (CMIPs) and the Sixth phase of CMIPs (CMIP6) to unravel the multifaceted impacts of global climate change on climate of Türkiye. The CMIP6 data, fundamental to the Intergovernmental Panel on Climate Change (IPCC) Assessment Reports, forms the basis for projecting future climate scenarios, specifically under the medium-high reference scenario SSP3-7. Utilizing a suite of global climate models, including the innovative Multi-Model Ensemble (MME) approach, this study combines predictions to enhance the precision climate projections of Türkiye. Historical data spanning from 1951 to 2020 were subjected to rigorous statistical analysis, including descriptive statistics and regression analysis. The findings reveal an unequivocal upward trajectory in Türkiye’s annual mean temperature, with an accelerated pace in recent decades. Despite a lack of a significant long-term trend in annual precipitation from 1951 to 2020, the rate of change in precipitation is accelerating, indicating potential future challenges. Projections for 2040-2059 under the SSP3-7.0 scenario indicate a non-uniform increase in mean temperature across Türkiye, with the southern and western regions facing the most significant impact. This warming trend poses imminent threats to agriculture, altering crop yields and increasing the risk of heat stress for livestock. Additionally, the projected decrease in precipitation, alongside a surge in hot days and tropical nights, underscores the urgency for adaptive measures. As Türkiye navigates the complex terrain of climate change, this study provides valuable insights, emphasizing the significance of robust climate modeling for informed decision-making. The results underscore the imminent challenges Türkiye faces and emphasize the critical importance of proactive climate action on both national and global fronts.

New insights into culvert scour depth calculation by soft computing models using multi-model ensemble approach and uncertainty analysis

ABSTRACT The accurate prediction of scour depth in culverts is crucial to ensure public safety and due to the uncertainties in classical empirical/deterministic equations, it remains challenging. This study presents new ensemble multi-models to predict scour depth and quantify model bias and uncertainty. In this study, culvert scour was predicted by gene expression programming (GEP) and least square support vector machine (LSSVM), firstly based on dimensionless parameters. In the multi-model ensemble strategy, four ensemble approaches, namely SM, WM, LSSVM, and BMA were applied. The LSSVM showed overall best skill for the multi-model approach, that reduced the error by 61.3% and increased the coefficient of determination by 52.1%. The results showed the proposed approach can satisfy the requirements of simplicity, applicability, and accuracy at the same time. For application purposes, we provided a simple code to obtain scour depth with great accuracy using LSSVM. BMA and LSSVM were used to quantify the scour culvert uncertainty. The results of the uncertainty analysis showed that the LSSVM predicts scour depth quite well and generates high-quality prediction intervals. The results revealed that a multi-model ensemble strategy improved accuracy, certainty and reliability for culvert scour compared to empirical equations.

Systematic review of the uncertainty of coral reef futures under climate change

Climate change impact syntheses, such as those by the Intergovernmental Panel on Climate Change, consistently assert that limiting global warming to 1.5 °C is unlikely to safeguard most of the world’s coral reefs. This prognosis is primarily based on a small subset of available models that apply similar ‘excess heat’ threshold methodologies. Our systematic review of 79 articles projecting coral reef responses to climate change revealed five main methods. ‘Excess heat’ models constituted one third (32%) of all studies but attracted a disproportionate share (68%) of citations in the field. Most methods relied on deterministic cause-and-effect rules rather than probabilistic relationships, impeding the field’s ability to estimate uncertainty. To synthesize the available projections, we aimed to identify models with comparable outputs. However, divergent choices in model outputs and scenarios limited the analysis to a fraction of available studies. We found substantial discrepancies in the projected impacts, indicating that the subset of articles serving as a basis for climate change syntheses may project more severe consequences than other studies and methodologies. Drawing on insights from other fields, we propose methods to incorporate uncertainty into deterministic modeling approaches and propose a multi-model ensemble approach to generating probabilistic projections for coral reef futures.

Regional ensemble of CMIP6 global climate models for Sakha (Yakutia) Republic, Northern Eurasia

Future climate projections based on multi-model ensemble approach are seen as more reliable, but not all models are equally performant at reproducing climate features at a regional scale. An optimal regional GCM ensemble was developed for Sakha (Yakutia) Republic based on error statistics and spatial correlation metrics. Historical Coupled Model Intercomparison Project, version 6 (CMIP6) simulations from 48 global climate models (GCMs) were used to evaluate model quality compared to mean annual air temperature (MAAT) reanalysis data for 1961–1990, 1971–2000 and 1981–2010 reference periods, and the MAAT change between 1961-1990 and 1981–2010, ΔT81-61. The best-performing reanalysis, GHCN-CAMS, was validated using observational data. This five-member ensemble includes CESM2-WACCM, CMCC-ESM2, CNRM-CM6-1-HR, INM-CM5-0, MPI-ESM1-2-HR models, weighted by Pearson's coefficient of spatial correlation between observed and modeled ΔT81-61 fields. Model weighting based on spatial correlation metrics improved the performance of the developed multi-model regional ensemble, which can be used in projecting future climate under different climate change scenarios.

Seasonal and Monthly Climate Variability in South Korea’s River Basins: Insights from a Multi-Model Ensemble Approach

This study conducts a comprehensive analysis of the impacts of climate change on South Korea’s climate and hydrology, utilizing a Multi-Model Ensemble (MME) approach with thirteen Climate Model Intercomparison Project Phase 5 (CMIP5) models under two Representative Concentration Pathways, RCP4.5 and RCP8.5. We observed an average temperature increase of up to 3.5 °C under RCP8.5 and around 2.0 °C under RCP4.5. Precipitation patterns showed an overall increase, particularly during the summer months, with increases up to 20% under RCP8.5 and 15% under RCP4.5, characterized by more intense and frequent rainfall events. Evapotranspiration rates are projected to rise by approximately 5–10% under RCP8.5 and 3–7% under RCP4.5. Runoff is expected to increase significantly, particularly in the summer and autumn months, with increases up to 25% under RCP8.5 and 18% under RCP4.5. This research focuses on employing the Precipitation Runoff Modeling System (PRMS) to project future streamflow across South Korea, with an emphasis on both monthly and seasonal scales to understand the varying impacts of climate change on different river basins. These climatic changes have profound implications for agriculture, urban water management, and ecosystem sustainability, stressing the need for dynamic and region-specific adaptation measures. This study emphasizes the critical role of localized factors, such as topography, land use, and basin-specific characteristics, in influencing the hydrological cycle under changing climatic conditions.

Uncertainties in deforestation emission baseline methodologies and implications for carbon markets

Carbon credits generated through jurisdictional-scale avoided deforestation projects require accurate estimates of deforestation emission baselines, but there are serious challenges to their robustness. We assessed the variability, accuracy, and uncertainty of baselining methods by applying sensitivity and variable importance analysis on a range of typically-used methods and parameters for 2,794 jurisdictions worldwide. The median jurisdiction’s deforestation emission baseline varied by 171% (90% range: 87%-440%) of its mean, with a median forecast error of 0.778 times (90% range: 0.548-3.56) the actual deforestation rate. Moreover, variable importance analysis emphasised the strong influence of the deforestation projection approach. For the median jurisdiction, 68.0% of possible methods (90% range: 61.1%-85.6%) exceeded 15% uncertainty. Tropical and polar biomes exhibited larger uncertainties in carbon estimations. The use of sensitivity analyses, multi-model, and multi-source ensemble approaches could reduce variabilities and biases. These findings provide a roadmap for improving baseline estimations to enhance carbon market integrity and trust.

A novel committee-based framework for modeling groundwater level fluctuations: A combination of mathematical and machine learning models using the weighted multi-model ensemble mean algorithm

water level (GWL) fluctuations simulation can be divided into three general categories: Analytical solution, conceptual or physical-based models, and data-based or data-driven models. The goal of this research is to develop the MODFLOW conceptual model and the GMDH data-driven model as single models (SMs) for GWL modeling, and then to combine these models to develop a novel mathematical-machine learning model based on a weighted multi-model ensemble mean approach (MMWEM) for a more accurate GWL simulation. For this purpose, the Miqan aquifer, located in northeastern Iran, was selected as a case study. Hydrological and hydrogeological data of the Miqan wetland for ten years (September 2013 to August 2023) with a monthly time step were used to compile models. First, the MODFLOW model was developed for one month (steady-state), then calibrated for nine years (unsteady-state), and finally tested for one year. In the second step, the GMDH model was developed. The grasshopper optimization algorithm (GOA) and the singular value decomposition (SVD) algorithms were used to determine the Ivakhnenko polynomial coefficients of the GMDH (SVD-GMDH and GOA-GMDH). 80% of the data was used to develop SVD-GMDH and GOA-GMDH, and other data was used to test models. Consequently, the sequential forward floating selection (SFFS) method was used to select the best input variables for the SVD-GMDH and GOA-GMDH. The coefficient of determination (R2), root mean squared error (RMSE), and Nash-Sutcliffe Efficiency coefficient (NSE) were used to evaluate the performance of the models. Based on the results of this study, using each of the SMs, GWL can be simulated with acceptable accuracy. However, a closer examination of the results of SMs showed that the MODFLOW has the worst performance among SMs, while the GOA-GMDH has the best function for simulating GWL. In the third step, SMs was used to develop the MMWEM. The GWL of 13 of the 34 piezometers studied in this study was better simulated with the MODFLOW and GOA-GMDH than the SVD-GMDH. Therefore, these models were used to develop the MMWEM for these piezometers. For GWL simulation in 21 piezometers, the SVD-GMDH and GOA-GMDH perform better than the MODFLOW, so these models were used to develop the MMWEM in these piezometers. Finally, the results of the SMs Moreover, the MMWEM were compared with each other. Based on the results of this study, the MMWEM performs better than SMs. In many studies, only the performance of each numerical model and machine learning to simulate groundwater level (GWL) fluctuations is evaluated. Then, their results are compared with each other. However, according to this study, the combination of the results of these models is a big step in improving the accuracy of the groundwater level simulation process. In addition to simulating the groundwater level, this method can simulate other quantitative and qualitative parameters of water resources.

Safety assessment of tunnel construction based on counterintuitivity detection using multi-profile multi-model ensemble learning

Motivated to accurately predict the building tilt rate (BTR) in tunnel construction, a new multi-profile multi-model ensemble learning approach is proposed in this study by distinguishing between counterintuitive data and intuitive data, and handling each type appropriately. The proposed approach consists of three phases, namely the definition, identification, and handling of counterintuitive data, which are the major theoretical contributions of this study. Firstly, counterintuitive data is defined based on the input–output causal relation of safety-related data, introducing a novel concept in this research. Secondly, counterintuitive data is initially identified using a multi-profile ensemble learning approach and subsequently validated through multi-model ensemble learning. Finally, the identified and confirmed counterintuitive data is handled by assigning reduced weights. To validate the practicality of the approach, a case study on predicting the building tilt rate (BTR) in tunnel construction is conducted. The results of the case study demonstrate that the proposed multi-profile multi-model approach yields more accurate predictions of BTR compared to direct prediction and several other machine learning approaches. Furthermore, the proposed approach aids in identifying counterintuitive data in the testing dataset. Additionally, the effectiveness and superiority of the proposed approach are verified by comparing the prediction results when traditional abnormal data or random data is identified and treated as counterintuitive data.

Blue and green water accounting for climate change adaptation in a water scarce river basin

Understanding how climate change influences the portioning of rainfall into blue water (BW) and green water (GW) is vital for establishing adaptive water management strategies in water-scarce regions. This study presents a conceptual framework to investigate climate change impacts on blue and green water accounting by integrating a hydrological model with a multi-criteria decision-making approach to prioritize watersheds for climate change adaptation. A Soil and Water Assessment Tool (SWAT) is adopted to investigate the hydrologic variability of BW and GW for baseline (1995–2019) and future (2020–2100) periods in a semi-arid Dharoi catchment of the Sabarmati River basin, India. The future climate data, derived from five general circulation models, are ensembled using a multi-model ensemble mean approach under two RCP (representative concentration pathways) scenarios. Several non-parametric tests are employed to ascertain statistically significant changes in the hydrologic variables between future and baseline periods. Further, Analytic Hierarchy Process (AHP) is applied considering seven attributes related to climate change, land use, and demographic parameters for watershed prioritization. Higher weights are assigned to the climate change attributes, followed by land use and demographic attributes. The SWAT results indicate a decrease in rainfall, BW, and GWS (green water storage) and an increase in temperature and GWF (green water flow) while traversing from the beginning of 2040 till the end of the 21st century under both RCPs. Thus, 2040 is established as a ‘tipping point’ for climate-induced changes in the Dharoi catchment, wherein a distinct reversal in the hydroclimatic regime is evident before and after 2040. Further, significant warming is observed by the end of 21st century under RCP4.5 and RCP8.5, contributing to a reduction in rainfall (up to 24.8%) and a decline in BW (up to 78.8%) across two-thirds of the watersheds in the Dharoi catchment. The AHP analysis identifies three out of ten watersheds as vulnerable and assigned ‘high priority’ for implementing adaptation measures to counter climate change-induced water scarcity.

Supermodeling: Improving Predictions with an Ensemble of Interacting Models

Abstract The modeling of weather and climate has been a success story. The skill of forecasts continues to improve and model biases continue to decrease. Combining the output of multiple models has further improved forecast skill and reduced biases. But are we exploiting the full capacity of state-of-the-art models in making forecasts and projections? Supermodeling is a recent step forward in the multimodel ensemble approach. Instead of combining model output after the simulations are completed, in a supermodel individual models exchange state information as they run, influencing each other’s behavior. By learning the optimal parameters that determine how models influence each other based on past observations, model errors are reduced at an early stage before they propagate into larger scales and affect other regions and variables. The models synchronize on a common solution that through learning remains closer to the observed evolution. Effectively a new dynamical system has been created, a supermodel, that optimally combines the strengths of the constituent models. The supermodel approach has the potential to rapidly improve current state-of-the-art weather forecasts and climate predictions. In this paper we introduce supermodeling, demonstrate its potential in examples of various complexity, and discuss learning strategies. We conclude with a discussion of remaining challenges for a successful application of supermodeling in the context of state-of-the-art models. The supermodeling approach is not limited to the modeling of weather and climate, but can be applied to improve the prediction capabilities of any complex system, for which a set of different models exists.

Using CNN framework to improve multi-GCM ensemble predictions of monthly precipitation at local areas: An application over China and comparison with other methods

Multi-model ensemble (MME) from various global climate models (GCMs) is widely used to quantify hydrological impacts of climatic change from local to global scales. Convolutional neural network (CNN) is a powerful tool for efficiently extracting hidden spatial information from huge datasets, and has been increasingly implemented in geoscientific studies. In this study, a CNN framework is developed to perform MME of CMIP6 monthly precipitation over China particularly at local scale, compared to 32 involved GCMs, quantile mapping (QM), and other prevalent MME approaches, including arithmetic mean ensemble (ENS), multiple linear regression (MLR), support vector machine (SVM), random forest (RF), and K-nearest neighbors (KNN), against in-situ measurements during 1970–2014. Overall, EC_Earth3_Veg outperforms other GCMs and ENS, with most monthly taylor skill scores (TSSs) between 0.8 and 0.9. Its performance is further improved through QM bias-correction, which however still presents slight gaps compared to other MME methods except SVM. CNN performs best with the highest comprehensive rating metric (MR) value of 0.96, corresponding to the highest mean TSS of 0.98, followed by KNN, RF, and MLR. It is also optimal for reproducing geographic distributions and seasonal cycle of precipitation and reducing biases at local scale, attributed to its unique advantage in resolving spatial dependences. Despite being slightly inferior to the best GCM, it outperforms other MME methods in capturing observed interannual variations and probability density functions, with weak sensitivity to varying ensemble sizes. Overall, CNN model is highly recommended in multi-GCM ensemble predictions for its superior and stable performance.

Extreme Heat Events over Southeast Europe Based on NEX-GDDP Ensemble: Present Climate Evaluation and Future Projections

Southeast Europe is considered one of the most vulnerable regions in the context of climate change, and projected future summer warming is expected to exceed global rates significantly. Despite the importance of this problem, there have been few studies that utilized Coupled Model Intercomparison Project Phase 5 (CMIP5) Global Circulation Models (GCMs) and the multi-model ensemble approach to examine extreme heat events on a regional scale. Additionally, the NEX-GDDP dataset, successfully applied in other parts of the world to assess extreme heat, has not yet been utilized for Southeast Europe. This study aims to fill that gap, presenting the time evolution and spatial distribution of extreme heat events in Southeast Europe for the historical period 1950–2005 and for the expected future climate up to the end of the 21st century based on the NEX-GDDP dataset. In order to leverage the strengths of the multi-model ensemble approach, a set of purpose-tailored indicators, such as the annual number of hot days, the maximum number of consecutive hot days, and hot spell duration at different thresholds, is computed by the daily maximum temperature data from all datasets, produced by NEX-GDDP (21 for the historical period and 42 for the future period). The E-OBS dataset is used as a reference for evaluating the NEX-GDDP’s capability to simulate the features of the observed historical extreme heat events. The results show that the multi-model ensemble can satisfactorily capture the occurrence of extreme heat events in the historical period, and therefore it is reasonable to assume that the NEX-GDDP dataset has the potential to reproduce such extremes in the projected future. The study provides clear evidence that the persistence and spatial extent of extreme heat will increase significantly. Some indicators that were not relevant for the historical period due to the high-temperature threshold will become helpful in assessing extreme heat in Southeast Europe in the latter part of the century. Thus, under the RCP8.5 scenario, the area-averaged duration of hot spells at 32 °C and 34 °C will increase from near zero in 1976–2005 to 60 and 45 days, respectively, by the end of the century. The indicators used in the study may be helpful for decision-makers to implement climate change mitigation strategies and actions adequately. The findings are consistent with general tendencies in maximum temperatures considered in our previous works but also with the outcomes of recent studies dedicated to the future climate of the region.

Evaluation of CMIP6 Models and Multi-Model Ensemble for Extreme Precipitation over Arid Central Asia

Simulated historical extreme precipitation is evaluated for Coupled Model Intercomparison Project Phase 6 (CMIP6) models using precipitation indices defined by the Expert Team on Climate Change Detection and Indices (ETCCDI). The indices of 33 Global Circulation Models (GCMs) are evaluated against corresponding indices with observations from the Global Climate Center Precipitation Dataset (GPCC V2020) over five sub-regions across Arid Central Asia (ACA), using the Taylor diagram, interannual variability skill score (IVS) and comprehensive rating index (MR). Moreover, we compare four multi-model ensemble approaches: arithmetic average multi-model ensemble (AMME), median multi-model ensemble (MME), pattern performance-based multi-model ensemble (MM-PERF) and independence weighted mean (IWM). The results show that CMIP6 models have a certain ability to simulate the spatial distribution of extreme precipitation in ACA and the best ability to simulate simple daily intensity (SDII), but it is difficult to capture the spatial bias of consecutive wet days (CWD). Almost all models represent different degrees of wet bias in the southern Xinjiang (SX). Most GCMs are generally able to capture extreme precipitation trends, but to reproduce the performance of interannual variability for heavy precipitation days (R10mm), SDII and CWD need to be improved. The four multi-model ensemble methods can reduce the internal system bias and variability within individual models and outperform individual models in capturing the spatial and temporal variability of extreme precipitation. However, significant uncertainties remain in the simulation of extreme precipitation indices in SX and Tianshan Mountain (TM). Comparatively, IWM simulations of extreme precipitation in the ACA and its sub-regions are more reliable. The results of this study can provide a reference for the application of GCMs in ACA and sub-regions and can also reduce the uncertainty and increase the reliability of future climate change projections through the optimal multi-model ensemble method.