Downscaling Research Articles

Assessing the impact of climate change by using Mann–Kendall, Pettitt and statistical downscaling model (case study: Tabriz station)

Assessing the impact of climate change by using Mann–Kendall, Pettitt and statistical downscaling model (case study: Tabriz station)

Study of the Future Evolution of the Urban Climate of Paris by Statistical–Dynamical Downscaling of the EURO-CORDEX Ensemble

Abstract High-resolution urban climate projections are needed for local decision-making on climate change adaptation. Regional climate models have resolutions that are too coarse to simulate the urban climate at such resolutions. A novel statistical–dynamical downscaling (SDD) approach is used here to downscale the EURO-CORDEX ensemble to a resolution of 1 km while adding the effect of the city of Paris (France) on air temperature. The downscaled atmospheric fields are then used to drive the Town Energy Balance urban canopy model to produce high-resolution temperature maps over the period 1970–2099, while maintaining the city’s land cover in its present state. The different steps of the SDD are evaluated for the summer season. The regional climate models simulate minimum (maximum) temperatures (TN/TX) that are too high (low). After correction and downscaling, the urban simulations inherit some of these biases but give satisfactory results for summer urban heat islands (UHIs), with average biases of −0.6 K at night and +0.3 K during the day. Changes in future summer temperatures are then studied for two greenhouse gas emission scenarios, RCP4.5 and RCP8.5. Outside the city, the simulations project average increases of 4.1 and 4.8 K for TN and TX for RCP8.5, respectively. In the city, warming is lower, resulting in a decrease in UHIs of −0.19 K at night (from 2.1 to 1.9 K) and −0.16 K during the day. The changes in UHIs are explained by higher rates of warming in rural temperatures due to lower summer precipitation and soil water content and are partially offset by increased ground heat storage in the city.

Climate projection of tropical cyclone lifetime in the western North Pacific basin

Abstract In this study, the potential changes of tropical cyclone (TC) lifetime in the western North Pacific basin are examined for different future climates. Using homogeneous 9 km-resolution dynamical downscaling with the Weather Research and Forecasting (WRF) model, we show that TC averaged lifetime displays insignificant change under both low and high greenhouse-gas concentration scenarios. However, more noticeable changes in the tails of TC lifetime statistics are captured in our downscaling simulations, with more frequent long-lived TCs (lifetime of 8-11 days) and less short-lived TCs (lifetime of 3-5 days). Unlike present-day simulations, it is found that the correlation between TC lifetime and the Niño index is relatively weak and insignificant in all future downscaling simulations, thus offering little explanation for these changes in TC lifetime statistics based on the El Niño-Southern Oscillation. More detailed analyses of TC track distribution in the western North Pacific basin reveal, nevertheless, a noticeable shift of TC track patterns towards the end of the 21st century. Such a change in TC track climatology results in an overall longer duration of TCs over the open ocean, which is consistent across future scenarios and periods examined in this study. This shift in the TC track pattern is ultimately linked to changes in the Western North Pacific Subtropical High, which retreats to the south during July and to the east during August-September. The results obtained in this study provide new insights into how large-scale circulations can affect TC lifetime in the western North Pacific basin in warmer climates.

Climatology of Orographic Precipitation Gradients Over High Mountain Asia Derived From Dynamical Downscaling

AbstractWithin High Mountain Asia (HMA), the annual melting of glaciers and snowpack provides vital freshwater to populations living downstream. Precipitation over HMA can directly affect the freshwater availability in this region by altering the mass balance of glaciers and snowpack. However, available reanalyses and downscaling simulations lack the resolution required to understand important glacier‐scale variations in precipitation. This study aimed to determine the current characteristics of orographic precipitation gradients (OPG) by curve‐fitting daily precipitation as a function of elevation from a 15‐year, 4‐km grid spaced Weather Research and Forecasting (WRF) model simulation focused on the Himalayan, Karakoram, and Hindu‐Kush mountain ranges. To facilitate precipitation curve‐fitting, the WRF model grid points were separated into regions of similar orientation, referred to as facets. Akaike Information Criterion‐corrected values and an F‐test p‐value identified the need for a curvature term to account for a varying OPG with elevation. Regions with similar seasonal variability were found using ‐means clustering of the monthly mean OPG coefficients. The central Himalayan slope's intra‐seasonal variability of OPG depended on synoptic scale conditions, in which cyclonically‐forced heavy‐precipitation events produced strong sublinear increases in precipitation with elevation. Initial testing of precipitation estimates using monthly coefficients showed promising results in downscaling daily WRF precipitation; the daily mean absolute error at each grid point had a lower magnitude than the daily mean precipitation total, on average. Results provide a physically‐based context for machine learning algorithms being developed to predict OPG and downscale precipitation output from global climate models over HMA.

Exploring Spatio-Temporal Dynamics of Future Extreme Precipitation, Runoff, and Flood Risk in the Hanjiang River Basin, China

The hydrological cycle is altered by climate change and human activities, amplifying extreme precipitation and heightening the flood risk regionally and globally. It is imperative to explore the future possible alterations in flood risk at the regional scale. Focusing on the Hanjiang river basin (HRB), this study develops a framework for establishing a scientific assessment of spatio-temporal dynamics of future flood risks under multiple future scenarios. In this framework, a GCMs statistical downscaling method based on machine learning is used to project future precipitation, the PLUS model is used to project future land use, the digitwining watershed model (DWM) is used to project future runoff, and the entropy weight method is used to calculate risk. Six extreme precipitation indices are calculated to project the spatio-temporal patterns of future precipitation extremes in the HRB. The results of this study show that the intensity (Rx1day, Rx5day, PRCPTOT, SDII), frequency (R20m), and duration (CWD) of future precipitation extremes will be consistently increasing over the HRB during the 21st century. The high values of extreme precipitation indices in the HRB are primarily located in the southeast and southwest. The future annual average runoff in the upper HRB during the near-term (2023–2042) and mid-term (2043–2062) is projected to decrease in comparison to the baseline period (1995–2014), with the exception of that during the mid-term under the SSP5-8.5 scenario. The high flood risk center in the future will be distributed in the southwestern region of the upper HRB. The proportions of areas with high and medium–high flood risk in the upper HRB will increase significantly. Under the SSP5-8.5 scenario, the area percentage with high flood risk during the future mid-term will reach 24.02%. The findings of this study will facilitate local governments in formulating effective strategic plans for future flood control management.

Dynamical downscaling and data assimilation for a cold-air outbreak in the European Alps during the Year Without a Summer of 1816

Abstract. The “Year Without a Summer” in 1816 was characterized by extraordinarily cold and wet periods in central Europe, and it was associated with severe crop failures, famine, and socio-economic disruptions. From a modern perspective, and beyond its tragic consequences, the summer of 1816 represents a rare opportunity to analyze the adverse weather (and its impacts) after a major volcanic eruption. However, given the distant past, obtaining the high-resolution data needed for such studies is a challenge. In our approach, we use dynamical downscaling, in combination with 3D variational data assimilation of early instrumental observations, for assessing a cold-air outbreak in early June 1816. We find that the cold spell is well represented in the coarse-resolution 20th Century Reanalysis product which is used for initializing the regional Weather Research and Forecasting Model. Our downscaling simulations (including a 19th century land use scheme) reproduce and explain meteorological processes well at regional to local scales, such as a foehn wind situation over the Alps with much lower temperatures on its northern side. Simulated weather variables, such as cloud cover or rainy days, are simulated in good agreement with (eye) observations and (independent) measurements, with small differences between the simulations with and without data assimilation. However, validations with partly independent station data show that simulations with assimilated pressure and temperature measurements are closer to the observations, e.g., regarding temperatures during the coldest night, for which snowfall as low as the Swiss Plateau was reported, followed by a rapid pressure increase thereafter. General improvements from data assimilation are also evident in simple quantitative analyses of temperature and pressure. In turn, data assimilation requires careful selection, preprocessing, and bias-adjustment of the underlying observations. Our findings underline the great value of digitizing efforts of early instrumental data and provide novel opportunities to learn from extreme weather and climate events as far back as 200 years or more.

Major distribution shifts are projected for key rangeland grasses under a high-emission scenario in East Africa at the end of the 21st century

Grassland landscapes are important ecosystems in East Africa, providing habitat and grazing grounds for wildlife and livestock and supporting pastoralism, an essential part of the agricultural sector. Since future grassland availability directly affects the future mobility needs of pastoralists and wildlife, we aim to model changes in the distribution of key grassland species under climate change. Here we combine a global and regional climate model with a machine learning-based species distribution model to understand the impact of regional climate change on different key grass species. The application of a dynamical downscaling step allows us to capture the fine-scale effects of the region’s complex climate, its variability and future changes. We show that the co-occurrence of the analysed grass species is reduced in large parts of eastern Africa, and particularly in the Turkana region, under the high-emission RCP8.5 scenario for the last 30 years of the 21st century. Our results suggest that future climate change will alter the natural resource base, with potentially negative impacts on pastoralism and wildlife in East Africa.

Climate Change Effects on Spatiotemporal Distribution of Precipitation over West Central India: A Statistical Downscaling Approach

The long-term shifts in temperatures and weather patterns are referred to as climate change. Climate change not only leads to long-term shifts in average temperatures but also changes in the spatial and temporal distribution of rainfall over continents. This study aims to predict likely changes in the rainfall pattern induced by climate change over the West Central (WC) region of India. The approach uses a statistical downscaling technique that converts coarse-scale outputs of the global climate model (GCM) to high-resolution future precipitation projections giving refined distribution. The decision support tool that uses a robust statistical downscaling technique viz. statistical downscaling model (SDSM) is used for assessing local climate change impacts. The research includes the study of likely regional climate variability, by integrating historical observational data and empirical relationships between large-scale climate variables and local weather patterns. Historical data and the SDSM, version 4.2, are employed to forecast future rainfall trends. Rainfall data from the India Meteorological Department and the National Centre for Environmental Prediction (NCEP) from 1961 to 2001 are used along with outputs from general circulation models (GCMs), viz. Hadley Centre coupled model, version 3 (HadCM3), and coupled global climate model, version 3 (CGCM3), for the period 1961–2099. Rainfall scenarios are presented for three future time periods (2011–2040, 2041–2070, and 2071–2099). The study indicates a significant increase in the mean annual precipitation across the West Central India region, particularly in the 2050s and 2080s. Mean annual rainfall is projected to rise by 10–19.4% under HadCM3 A2 and B2 scenarios. The HadCM3 indicates the month of September as the month of the highest precipitation in later time periods, whereas it is the month of August, according to CGCM3 simulations. When comparing the results of the two models, HadCM3 gives better results, as indicated by better R2 value in validation. Thus, the analysis gives climate change-induced likely changes in the spatiotemporal distribution of precipitation over the West Central India region. The insight given by the work will be useful for decision making in many sectors like agriculture, water management, disaster risk reduction, and infrastructure planning.

Kilometer-Scale Precipitation Forecasting Utilizing Convolutional Neural Networks: A Case Study of Jiangsu’s Coastal Regions

High-resolution precipitation forecasts play a pivotal role in formulating comprehensive disaster prevention and mitigation plans. As spatial resolution enhances, striking a balance between computation, storage, and simulation accuracy becomes imperative to ensure optimal cost-effectiveness. Convolutional neural networks (CNNs), a cornerstone of deep learning, are examined in this study for their downscaling capabilities in precipitation simulation. During a precipitation event on 23 June 2022, in Jiangsu Province, China, distinct rain belts emerged in both southern and northern Jiangsu, precisely captured by a numerical model (the Weather Research and Forecasting, WRF) with a 3 km spatial resolution. Specifically, precipitation was prevalent in northern Jiangsu from 00:00 to 11:00 Beijing Time (BJT), transitioning to southern Jiangsu from 12:00 to 23:00 BJT on the same day. Upon dynamic downscaling, the model reproduced precipitation in these periods with an average error of 12.35 mm at 3 km and 12.48 mm at 1 km spatial resolutions. Employing CNN technology for statistical downscaling to a 1 km spatial resolution, samples from the initial period were utilized for training, while those from the subsequent period served for validation. Following dynamic downscaling, CNNs with four, five, six, and seven layers exhibited average errors of 8.86 mm, 8.93 mm, 9.71 mm, and 9.70 mm, respectively, accompanied by correlation coefficients of 0.550, 0.570, 0.574, and 0.578, respectively. This analysis indicates that for this precipitation event, a shallower CNN depth yields a lower average error and correlation coefficient, whereas a deeper architecture enhances the correlation coefficient. By employing deep network architectures, CNNs are capable of capturing nonlinear patterns and subtle local features from complex meteorological data, thereby providing more accurate predictions during the downscaling process. Leveraging faster computation and reduced storage requirements, machine learning has demonstrated immense potential in high-resolution forecasting research. There is significant scope for advancing technologies that integrate numerical models with machine learning to achieve higher-resolution numerical forecasts.

Correction: Dynamical downscaling improves upon gridded precipitation products in the Sierra Nevada, California

Correction: Dynamical downscaling improves upon gridded precipitation products in the Sierra Nevada, California

RASCAL v1.0: an open-source tool for climatological time series reconstruction and extension

Abstract. The reduction of in situ observations over the last few decades poses a potential risk of losing important information in regions where local effects dominate the climatology. Reanalyses face challenges in representing climatologies with highly localized effects, especially in regions with complex orography. Empirical downscaling methods offer a cost-effective and easier-to-implement alternative to dynamic downscaling methods and can partially overcome the aforementioned limitations of reanalyses by taking into account the local effects through statistical relationships. This article introduces RASCAL (Reconstruction by AnalogS of ClimatologicAL time series), an open-source Python tool designed to extend time series and fill gaps in observational climate data, especially in regions with limited long-term data and significant local effects, such as mountainous areas. Employing an object-oriented programming style, RASCAL's methodology effectively links large-scale circulation patterns with local atmospheric features using the analog method in combination with principal component analysis (PCA). The package contains routines for preprocessing observations and reanalysis data, generating reconstructions using various methods, and evaluating the reconstruction's performance in reproducing the time series of observations, statistical properties, and relevant climatic indices. Its high modularity and flexibility allow fast and reproducible downscaling. The evaluations carried out in central Spain, in mountainous and urbanized areas, demonstrate that RASCAL performs better than the ERA20C and ERA20CM reanalysis, as expected, in terms of R2, standard deviation, and bias. When analyzing reconstructions against observations, RASCAL generates series with statistical properties, such as seasonality and daily distributions, that closely resemble observations. This confirms the potential of this method for conducting robust climate research. The adaptability of RASCAL to diverse scientific objectives is also highlighted. However, as with any other method based on empirical training, this method requires the availability of sufficiently long-term data series. Furthermore, it is susceptible to disruption caused by changes in land use or urbanization processes that might compromise the homogeneity of the training data. Despite these limitations, RASCAL's positive outcomes offer opportunities for comprehensive climate variability analyses and potential applications in downscaling short-term forecasts, seasonal predictions, and climate change scenarios. The Python code and the Jupyter Notebook for the reconstruction validation are publicly available as an open project.

Antarctic sea ice surface temperature bias in atmospheric reanalyses induced by the combined effects of sea ice and clouds

Sea-ice surface temperature from atmospheric reanalysis has been used as an indicator of ice melt and climate change. However, its performance in atmospheric reanalyses is not fully understood in Antarctica. Here, we quantified biases in six widely-used reanalyses using satellite observations, and found strong and persistent warm biases in most reanalyses examined. Further analysis of the biases revealed two main culprits: incorrect cloud properties, and inappropriate sea-ice representation in the reanalysis products. We found that overestimated cloud simulation can contribute more than 4 K warm bias, with ERA5 exhibiting the largest warm bias. Even in reanalysis with smaller biases, this accuracy is achieved through a compensatory relationship between relatively lower cloud fraction bias and overestimated sea ice insulation effect. A dynamic downscaling simulation shows that differences in sea-ice representation can contribute a 2.3 K warm bias. The representation of ice concentration is the primary driver of the spatial distribution of biases by modulating the coupling between sea ice and clouds, as well as surface heat conduction. The lack of a snow layer in all reanalyses examined also has an impact on biases.

TC‐GEN: Data‐Driven Tropical Cyclone Downscaling Using Machine Learning‐Based High‐Resolution Weather Model

AbstractSynthetic downscaling of tropical cyclones (TCs) is critically important to estimate the long‐term hazard of rare high‐impact storm events. Existing downscaling approaches rely on statistical or statistical‐deterministic models that are capable of generating large samples of synthetic storms with characteristics similar to observed storms. However, these models do not capture the complex two‐way interactions between a storm and its environment. In addition, these approaches either necessitate a separate TC size model to simulate storm size or involve post‐processing to capture the asymmetries in the simulated surface wind. In this study, we present an innovative data‐driven approach for TC synthetic downscaling. Using a machine learning‐based high‐resolution global weather model (ML‐GWM), our approach can simulate the full life cycle of a storm with asymmetric surface wind that accounts for the two‐way interactions between the storm and its environment. This approach consists of multiple components: a data‐driven model for generating synthetic TC seeds, a blending method that seamlessly integrates storm seeds into the surrounding while maintaining the seed structure, and a model based on a recurrent neural network to correct for biases in storm intensity. Compared to observations and synthetic storms simulated using existing statistical‐deterministic and statistical downscaling approaches, our method shows the ability to effectively capture many aspects of TC statistics, including track density, landfall frequency, landfall intensity, and outermost wind extent. Leveraging the computational efficiency of ML‐GWM, our approach shows substantial potential for TC regional hazard and risk assessment.

Assessment of Maximum Temperature for Future Time Series over Aurangabad, Maharashtra State, India

Climatic downscaling is an important tool used by many researchers in the field of climate to downscale climatic data available for global area to study the impacts of climate on smaller area. In climatic downscaling, there are mainly two types: 1) Statistical Downscaling and 2) Dynamical Downscaling. In the present study, statistical downscaling has been performed using SDSM (Statistical Downscaling Model) to find out future values of maximum temperatures (Tmax) over Aurangabad region (Latitude: 19.8911° N, Longitude: 75.1545° E). For this downscaling, CanESM2 (Canadian Earth System Model) CMIP5 (Coupled Model Intercomparison Project Phase 5) GCM (General Circulation Model) were selected and downscaling was performed under three different RCPs (Representative Concentration Pathways): RCP 2.6, RCP 4.5 and RCP 8.5. Downscaling model is calibrated and validated successfully over the baseline period 1961-2005 and results are presented graphically as well as statistically. Future downscaling values of maximum temperatures are presented with the help of three future time series: 2020s (2011-2040), 2050s (2041-2070) and 2080s (2071-2099) and it is observed that there is increase in the values of Tmax for the future time series with respect to the baseline period. Further the heat maps are developed for increasing monthly mean daily Tmax values under three RCPs with respect to the baseline period over the duration 2006 to 2013 and compared with increased observed monthly mean daily Tmax values over the same duration. This comparison has been done to assess the pattern of increasing monthly mean daily Tmax of downscaling results over Aurangabad region. All three RCPs are showing increasing values of monthly mean daily Tmax values for the selected three future time series with respect to the baseline period. Decade wise study of downscaling results is also showing increasing values of mean monthly daily Tmax with respect to the baseline period. Tmax values are predicted to be increased under RCP 2.6 by 1.10 0C, under RCP 4.5 by 2.13 0C and under RCP 8.5 by 4.20 0C at the end of 2099 which follows the prediction given by IPCC for global level temperature rise up to 2100.

Evaluating land use and climate change impacts on Ravi river flows using GIS and hydrological modeling approach

The study investigates the interplay of land use dynamics and climate change on the hydrological regime of the Ravi River using a comprehensive approach integrating Geographic Information System (GIS), remote sensing, and hydrological modeling at the catchment scale. Employing the Soil and Water Assessment Tool (SWAT) model, simulations were conducted to evaluate the hydrological response of the Ravi River to both current conditions and projected future scenarios of land use and climate change. This study differs from previous ones by simulating future land use and climate scenarios, offering a solid framework for understanding their impact on river flow dynamics. Model calibration and validation were performed for distinct periods (1999–2002 and 2003–2005), yielding satisfactory performance indicators (NSE, R2, PBIAS = 0.85, 0.83, and 10.01 in calibration and 0.87, 0.89, and 7.2 in validation). Through supervised classification techniques on Landsat imagery and TerrSet modeling, current and future land use maps were generated, revealing a notable increase in built-up areas from 1990 to 2020 and projections indicating further expansion by 31.7% from 2020 to 2100. Climate change projections under different socioeconomic pathways (SSP2 and SSP5) were derived for precipitation and temperature, with statistical downscaling applied using the CMhyd model. Results suggest substantial increases in precipitation (10.9 − 14.9%) and temperature (12.2 − 15.9%) across the SSP scenarios by the end of the century. Two scenarios, considering future climate conditions with current and future land use patterns, were analyzed to understand their combined impact on hydrological responses. In both scenarios, inflows to the Ravi River are projected to rise significantly (19.4 − 28.4%) from 2016 to 2100, indicating a considerable alteration in seasonal flow patterns. Additionally, historical data indicate a concerning trend of annual groundwater depth decline (0.8 m/year) from 1996 to 2020, attributed to land use and climate changes. The findings underscore the urgency for planners and managers to incorporate climate and land cover considerations into their strategies, given the potential implications for water resource management and environmental sustainability.

An ensemble approach to downscale climatological variables via screening of clustered ANNs

ABSTRACT The statistical downscaling of global circulation models presents a significant challenge in selecting appropriate input variables from a vast pool of predictors. To address the issue, we developed ensemble approach based on the Combining Multiple Clusters via Similarity Graph (COMUSA), which integrates k-means and self-organizing maps (SOMs) methods with the mutual information (MI)-random sampling approach. This innovative feature extraction technique demonstrated a 21% improvement in the classification efficacy of large-scale climatic variables. When comparing feature extraction methods, the combination of MI-random sampling and ensemble clustering yielded more accurate results than SOM clustering alone. The most efficient artificial neural networks (ANNs)-based downscaling model was employed to project near- and mid-future precipitation and temperature (2025–2035, 2035–2045), revealing varied outcomes under different scenarios (SSP3-7.0 and SSP5-8.5). Under SSP3-7.0 and SSP5-8.5, annual mean precipitation values are projected to decrease by 2–3 and by 4–5%. Also, projected annual mean temperature values indicate an increase in 21-27 and 29-35% under SSP3-7.0 and SSP5-8.5 scenarios. Integrating COMUSA ensemble clustering with MI-random sampling enhances the estimation accuracy of the ANN downscaling model, contributing to accurate projections of future precipitation and temperature values.

Assessing Climate Extremes in Dynamical Downscaling Simulations Driven by a Novel Bias‐Corrected CMIP6 Data

AbstractDynamical downscaling is a widely‐used approach for generating regional projections of climate extremes at a finer scale. However, the systematic bias of the global climate model (GCM) generally degrades the reliability of projections. Recently, novel bias‐corrected CMIP6 data was generated using a mean‐variance‐trend (MVT) bias correction method for dynamical downscaling simulation. To validate the effectiveness of this data in the dynamical downscaling simulation of climate extremes, we carry out three Weather Research and Forecasting (WRF) simulations over Asia‐western North Pacific with a 25 km grid spacing from 1980 to 2014. The dynamical downscaling simulations driven by the raw GCM data set (hereafter WRF_GCM) and the bias‐corrected GCM (hereafter WRF_GCMbc) were assessed against the simulation driven by the European Center for Medium‐Range Weather Forecasts Reanalysis 5 data set. The results indicate that the MVT bias correction significantly improves the climatological mean and inter‐annual variability of downscaled climate extreme indices. In terms of the climatological mean, the WRF_GCMbc shows a 25%–82% decrease in root mean square errors (RMSEs) against the WRF_GCM. As for the inter‐annual variability, the MVT bias correction can improve the downscaling simulation of almost all precipitation extreme indices and 70% of the temperature extreme indices, characterized by the RMSEs' reduction of 1%–58%. The improvements of the climate extremes in terms of the climatological mean in the WRF_GCMbc primarily stem from the improved large‐scale circulation and ocean evaporation in the WRF, which in turn improves the downscaled precipitation and 2 m temperature through the advection, radiation, and surface energy exchange process.

Statistical downscaling model for future projection of daily IDF relationship by Markov chain and kernel density estimator

ABSTRACT Changes in climate might have a significant impact on rainfall characteristics, including extreme rainfall. This study aims to project the future daily rainfall, preserving most of the rainfall characteristics, including extreme rainfall incorporating climate changes. This paper presents two hybrid semi-parametric statistical downscaling models for future projection of IDF curves. The precipitation flux from seven scenarios of ten GCMs and observed daily rainfall data are considered as predictors and predictand variables, respectively. At site, daily rainfall occurrence is modeled using a two-state first-order Markov chain. Rainfall amounts on each wet day are modelled using a univariate nonparametric kernel density estimator. Two types of amount generation models are presented in this study. The bounded model (KDE-SP) is developed, considering the support for the kernel distribution as positive. In the unbounded model (KDE-Ext), the wet days are reclassified as extreme and non-extreme rainy days. A significant increasing trend can be observed in the future projected intensity–duration–frequency relationships. The maximum increment using empirical distribution is observed as 93.21 and 80.93% on a 5-year return period in the far future for the SSP5-8.5 scenario, using KDE-Ext and KDE-SP models, respectively. Although both methods show similar results, the KDE-Ext model performs better in simulating extreme rainfall.

Application of deep learning in summer climate prediction over northwestern China based on CWRF model

This study analyzes the performance of the Climate–Weather Research and Forecasting (CWRF) model in predicting the summer temperature and precipitation in northwestern China (NWC) for the 1991–2021 period. It also examines the improvements in prediction resulting from the implementation of convolutional neural network (CNN) and long short-term memory (LSTM) downscaling methods. The results indicate that the CWRF model demonstrates reasonable ability in capturing the characteristics of climatological temperature and precipitation in NWC. Both the climatological temperature and precipitation predictions consistently demonstrate a systematic underestimation, revealing evident biases in regions characterized by complex terrain. In terms of interannual variation, the temperature prediction outperforms the precipitation prediction, whereas there is no significant difference in the temperature predictions for lead Months 1–3. However, uncertainties increase as the lead time is extended in precipitation prediction. Therefore, the combination of dynamical and statistical downscaling is employed to the summer temperature and precipitation prediction over NWC. It is shown both the CNN and LSTM downscaling methods can improve the prediction ability of the CWRF model for summer climatological temperature and precipitation. The LSTM method significantly reduces the root mean square error (RMSE) of precipitation and temperature predictions, indicating an improvement in predicting the spatial structure. At the interannual scale, the CNN method is less dependent on the lead time of prediction than the LSTM method is, and the interannual correlation coefficient of precipitation and temperature is greater than 0.1 compared with that of the raw CWRF model. These results provide valuable insights into understanding the prediction capabilities of the CWRF model in NWC and highlight the necessity of applying downscaling methods to the CWRF model to increase its prediction ability.

An assessment of the impact of boundary conditions in dynamical downscaling techniques for fetch-limited waves

ABSTRACT The handling of boundary conditions (BCs) is a key issue for the development of dynamical downscaling techniques in which nearshore wave models require the definition of the sea conditions at the offshore boundaries. This study addresses the implications of using simplified BCs represented by bulk or partitioned wave parameters with respect to using the complete information from directional wave spectra. As a large-scale model, we used WAVEWATCH III (WW3) running over the Mediterranean Sea providing the wave BCs for the small-scale SWAN model. One year of measured directional wave spectra from three deepwater buoys allowed the assessment of the WW3 model accuracy. The nested SWAN model simulations propagated the wave field over three nearshore areas. The accuracy of SWAN simulations was assessed through the comparison of computed results with wave measurements from nearshore buoys. Being the wave conditions dominated by unimodal seas, modest differences of error metrics are found between the datasets forced by different types of BCs. Conversely, focusing the attention on multimodal seas, an accurate definition of wave BCs achieved by the use of the directional wave spectra or, in alternative, by the reconstruction of wave spectra from partitioned wave parameters becomes crucial for nearshore wave predictions.