Tag: Climate Change Downscaling Techniques

Abstract

Streamflow time series data typically exhibit nonlinear and nonstationary characteristics that complicate precise estimation. Recently, multifactorial machine learning (ML) models have been developed to enhance the performance of streamflow predictions. However, the lack of interpretability within these ML models raises concerns about their inner workings and reliability. This paper introduces an innovative hybrid architecture, the TCN-LSTM-Multihead-Attention model, which combines two layers of temporal convolutional networks (TCN) followed by one layer of long short-term memory (LSTM) units, integrated with a Multihead-Attention mechanism for predicting streamflow with streamflow causation–driven prediction samples (RCDP), employing local and global interpretability studies through Shapley values and partial dependency analysis. The find_peaks method was used to identify peak flow events in the test dataset, validating the model’s generality and uncovering the physical causative patterns of streamflow. The results show that (1) compared to the LSTM model with the same hyperparameter settings, the proposed TCN-LSTM-Multihead-Attention hybrid model increased the R² by 52.9%, 2.5%, 43.1%, and 10.7% respectively at four stations in the test set predictions using RCDP samples. Moreover, comparing the prediction results of the hybrid model under different samples in Hengshan station, the R² for RCDP increased by 5.06% and 1.22% compared to streamflow autoregressive prediction samples (RAP) and meteorological-soil volumetric water content coupled autoregressive prediction samples (MCSAP) respectively. (2) Historical streamflow data from the preceding 3 days predominantly influences predictions due to strong autocorrelation, with flow quantity (Q) typically emerging as the most significant feature alongside precipitation (P), surface soil moisture (SSM), and adjacent station flow data. (3) During periods of low and normal flow, historical data remains the most crucial factor; however, during flood periods, the roles of upstream inflow and precipitation become significantly more pronounced. This model facilitates the identification and quantification of various hydrodynamic impacts on flow predictions, including upstream flood propagation, precipitation, and soil moisture conditions. It also elucidates the model’s nonlinear relationships and threshold responses, thereby enhancing the interpretability and reliability of streamflow predictions.

Abstract

Subseasonal-to-seasonal (S2S) prediction is a global effort to forecast the state of the atmosphere and ocean with lead times between two weeks and a season. This study explores the feasibility of S2S prediction of the ocean using a variety of tools including statistical analysis, a statistical-dynamical mixed layer model, and a regional, high-resolution ocean circulation model based on physical principles. Ocean predictability on S2S timescales is analyzed by compositing winter sea surface temperature (SST) anomalies in the North Atlantic with respect to the state of the Madden–Julian Oscillation (MJO). It is found that statistically significant, large-scale SST changes, particularly along the eastern seaboard of North America, can be related to the MJO. This signal is shown to be driven by anomalous air–sea heat fluxes caused by atmospheric perturbations in response to the MJO. The high-resolution model of the Gulf of Maine and Scotian Shelf is used to downscale the mean ocean response to the MJO. The model is able to capture the observed relationship between the MJO and SST in the northwest Atlantic. It is also shown that the anomalous atmospheric circulation in response to the MJO leads to anomalous upwelling on the Scotian Shelf. Overall, this study demonstrates that it is feasible, and of value, to use regional ocean models for S2S prediction.

Abstract

Over the past few decades, increasing frequencies and intensities of extreme weather events have been noted worldwide, which are a result of climate change. There is an increasing need to understand surface air temperature variabilities, as these are required to devise adaptation and mitigation plans for the Egyptian Mediterranean coast (EMC). The current paper sheds light on current and future trends in the surface air temperature (T2m) by using modeling techniques (the regional climate model RegCM-SVN). Various atmospheric parameters—air temperature, geopotential height, relative humidity, and zonal/meridional wind components—are used to drive the RegCM-SVN model for dynamical downscaling simulation of the period from 2006 to 2060 under the RCP2.6 (2.6 W/m²) and RCP8.5 (> 8.5 W/m²) scenarios. The results showed that the RCP2.6 and RCP8.5 scenarios over the EMC lead to significant warming, ranging from + 0.35 to + 0.38 °C, suggesting a significant acceleration of T2m warming trends over the next decades. The RegCM-SVN model and ERA5 (the fifth-generation ECMWF reanalysis for the global climate and weather data) give rather similar results when simulating T2m and show a strong correlation of greater than 90% over the oceanic area and greater than 0.85% over the land area (the Sinai Peninsula and the Gulf of Aqaba) for the controlled period (2006–2020). The validation processes indicated that the RegCM-SVN model successfully simulated the increasing surface air temperature over the EMC, which is considered a major challenge posed by climate change. This challenge is expected to have a series of negative effects across all the sectors of human activities in Egypt (the tourism, agriculture, health, and marine sectors). Thus, understanding the dynamics of T2m over Egypt is a critical issue in coping with climate change. In the same context, determining the long-term variability of surface air temperature is essential for devising adaptation and mitigation plans over such hotspot marine areas.

Abstract

The aim of this research is to assess the impact of future climate change on hydrological parameters (e.g., precipitation and temperature) in Ethiopia’s upper Genale River basin. Future climate scenarios for the 2021–2050 and 2051–2080 periods were developed from four different GCM–RCM combinations of CORDEX-Africa projections using the Representative Concentration Pathways (RCP 4.5 and RCP 8.5). These climate models were bias corrected and used as input to the Soil and Water Assessment Tool (SWAT) model. During the 2030s (2021–2050) and 2060s (2051–2080), under the two RCPs, the projected precipitation in the annual and seasonal periods tends to decrease while temperatures increase. The simulated result revealed a significant change in hydrological components (e.g., During the 2060s), under the RCP4.5 scenario, CNRM-CM5 climate model runoff, ground water flow, and total water yield increased by 24.47%, 27.98%, and 28.56%, respectively. On the contrary, during the 2060s under the MIROC5 climate model, runoff, ground water flow, and total water yield reduced by 20.84%, 34.34%, and 25.8%, respectively. The annual hydrological components of the study area under MPI-ESM-LR, EC-EARTH, and MIROC5 showed a decrease in total water yield, surface runoff, ground waterflow, and lateral flow. However, due to a rise in temperature, evapotranspiration showed an increase up to 8.1% under all climate models (MPI-ESM-LR, EC-EARTH, CNRM-CM5, and MIROC5). The reduction in rainfall, which coincides with rising temperatures, is expected to reduce annual water yield, surface runoff, ground waterflow, and lateral flow by up to 39.8%, 39.3%, 50%, and 40.1%, respectively, across MPI-ESM-LR, EC-EARTH, and MIROC5 scenarios for the entire study basin in future projections. Our study helps to give a better insight into understanding climate change in watershed and can benefit the planning of water resources by strengthening adaptation strategies against the impacts of future climate change.

Abstract

Global Circulation Models (GCMs) and Regional Climate Models (RCMs) are fundamental tools for investigating climate change and its impact. This study focuses on evaluation the ability of two Regional Climate Models (RCMs), REMO2015 and RegCM4-7, to simulate minimum and maximum temperature as well as monthly precipitation in Northwestern Iraq. The evaluation is based on comparisons with observed data from four key stations in the region. Both the regional models, REMO2015 and RegCM4-7, are driven by three GCMs, so the number of models evaluated is six GCM-RCM combination models. The performance of the six GCM-RCM model outputs was evaluated against observed data using a suite of statistical criteria: Bias Percent (PBIAS), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Correlation Coefficient (R), and Cumulative Distribution Function (CDF). Additionally, the Innovative Test (IT) was employed to analyze trends in both annual and monthly precipitation data. Also, Bias correction was employed the “Linear scaling: method to correct temperature and “Distribution Mapping” method for monthly precipitation. In general, the six models exhibited better performance in simulating minimum and maximum temperature than monthly precipitation. An interesting finding is that all six models exhibited similar behavior in simulating maximum temperature across all stations. Statistical criteria values revealed that the NCC-RegCM4-7 model produces the best results for minimum temperature simulation. Meanwhile for in maximum temperature, both MOHC-REMO2015 and NCC-REMO2015 emerged as the top performers. Additionally, MIROC-RegCM4-7 demonstrated the greatest skill in capturing t trends in the observed data, closely matching the trends measured at the studied stations. The application of the linear scaling method effectively addressed the biases present in the raw temperature data from the six GCM-RCM combinations. This correlation resulted in reduced MAE and RMSE values for both minimum and maximum temperatures. Similarly, the DM method successfully removed biases in the monthly precipitation simulation from all six models.

Abstract

Extensive research has been diligently conducted on wind energy technologies in response to pressing global environmental challenges and the growing demand for energy. Accurate wind speed predictions are crucial for the effective integration of large wind power systems. This study presents a novel and hybrid framework called ICEEMDAN-Informer-GWO, which combines three components to enhance the accuracy of wind speed predictions. The improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) component improves the decomposition of wind speed data, the Informer model provides computationally efficient wind speed predictions, and the grey wolf optimisation (GWO) algorithm optimises the parameters of the Informer model to achieve superior performance. Three different sets of wind speed prediction (WSP) models and wind farm data from Block Island, Gulf Coast, and Garden City are used to thoroughly assess the proposed hybrid framework. This evaluation focusses on WSP for three specific time horizons: 5 minutes, 30 minutes, and 1 hour ahead. The results obtained from the three conducted experiments conclusively demonstrate that the proposed hybrid framework exhibits superior performance, leading to statistically significant improvements across all three time horizons.

Abstract

In the current study, three optimistic (SSP1-2.6), medium (SSP2-4.5), and pessimistic (SSP5-8.5) scenarios were used to examine changes in precipitation based on the sixth phase of Coupled Model Intercomparison Project (CMIP6) in the Gorganrood watershed over two time periods: the near future (2021–2060) and the far future (2061–2100). To do this, the rainfall of 27 meteorological stations was studied. Using the RClimdex software in the R software, precipitation extreme indicators (11 indicators) were determined for different scenarios and periods, and Mann-Kendall (MK) and Sen’s estimator tests were then used to detect the trend. The results showed that in the near future under SSP1-2.6, the indicators of consecutive dry days (CDD) and consecutive wet days (CWD) have a significant downward and upward trend, respectively. While in the SSP5-8.5, the indicators of maximum five-day rainfall (RX1day), CDD, number of very wet days (R95p) and total wet day precipitation (PRCPTOT) have a significant downward trend in some stations. Similarly, in the far future, in the SSP5-8.5, the trend of rainfall indicators is insignificant compared to the near future, but still a significant decreasing trend can be seen in R95p, R99p, and PRCPTOT. Z score index (ZI) values ​​in both future periods showed that drought peaks occurred in the optimistic scenario and drought peaks occurred in the pessimistic scenario, and almost normal conditions prevailed in the intermediate scenario. The results can be effective in policies to deal with global warming and climate change.

Graphical Abstract

Abstract

The coarse spatial resolution of the Gravity Recovery and Climate Experiment (GRACE) dataset has limited its application in local water resource management and accounting. Despite efforts to improve GRACE spatial resolution, achieving high resolution downscaled grids that correspond to local hydrological behaviour and patterns is still limited. To overcome this issue, we propose a novel statistical downscaling approach to improve the spatial resolution of GRACE-terrestrial water storage changes (ΔTWS) using precipitation, evapotranspiration (ET), and runoff data from the Australian Water Outlook. These water budget components drive changes in the GRACE water column in much of the global land area. Here, the GRACE dataset is downscaled from the original resolution of 1.0° × 1.0° to 0.05° × 0.05° over a large hydro-geologic basin in northern Australia (the Cambrian Limestone Aquifer—CLA), capturing sub- grid heterogeneity in ΔTWS of the region. The downscaled results are validated using data from 12 in-situ groundwater monitoring stations and water budget estimates of the CLA’s land water storage changes from April 2002 to June 2017. The change in water storage over time (ds/dt) estimated from the water budget model was weakly correlated (r = 0.34) with the downscaled GRACE ΔTWS. The weak relationship was attributed to the possible uncertainties inherent in the ET datasets used in the water budget, particularly during the summer months. Our proposed methodology provides an opportunity to improve freshwater reporting using GRACE and enhances the feasibility of downscaling efforts for other hydrological data to strengthen local-scale applications.

Abstract

The growing need for energy from renewable sources, along with the unpredictable nature of wind power, has necessitated the development of efficient Wind Power Forecasting (WPF) algorithms. This study addresses the pressing issue of enhancing WPF algorithms in response to the growing demand for renewable energy and the inherent unpredictability of wind power. Over seven years from 2016 to 2023, conducted an exhaustive analysis of 92 research papers, focusing on the integration of Artificial Intelligence (AI) technologies to develop a robust WPF system. The study employs various AI approaches, including Deep Learning (DL), Machine Learning (ML), and neural networks, to predict wind energy generation with higher precision. Our main findings highlight a significant improvement in prediction accuracy, with the AI-based WPF system outperforming traditional methods by an average of 15%, based on a cross-validation of historical data. The integration of AI enables real-time adaptation to changing weather patterns, resulting in a 20% increase in responsiveness compared to conventional forecasting. Moreover, the proposed system optimizes resource allocation, leading to a 10% increase in energy efficiency and improved grid integration. Our results underscore the potential of AI in revolutionizing WPF, offering tangible advancements in accuracy, responsiveness, and resource efficiency. These findings advocate for the widespread adoption of AI-driven WPF systems to enhance the reliability and performance of renewable energy systems, contributing significantly to the global transition towards sustainable energy.

Abstract

Wind dynamics are extremely complex and have critical impacts on the level of damage from natural hazards, such as storms and wildfires. In the wake of climate change, wind dynamics are becoming more complex, making the prediction of future wind characteristics a more challenging task. Nevertheless, having long-term projections of some wind characteristics, such as daily wind speed, is crucial for effective monitoring of climate change, and for efficient disaster risk management. Furthermore, accurate projections of wind speed result in optimized generation of wind-based electric power. General circulation models (GCMs) provide long-term simulations (often till year 2100 or more) of multiple climate variables. However, simulations from a GCM are at a grid with coarse spatial resolution, rendering them ineffective to resolve and analyze climate change at the local regional level. Spatial downscaling techniques are often used to map such global large-scale simulations to a local small-scale region. In this paper, we present a novel deep learning framework for spatial downscaling, specifically for forecasting the daily average wind speed at a local station level using GCM simulations. Our framework, named wind convolutional neural networks with transformers, or WCT for short, consists of multi-head convolutional neural networks, followed by stacked transformers, and an uncertainty quantification component based on Bayesian inference. Experimental results show the suitability of WCT when applied on four wind stations in New Jersey and Pennsylvania, USA. Moreover, we use the trained WCT on future GCM simulations to produce local-scale daily wind speed projections up to the year 2100.