Figure 16 - uploaded by Douglas Maraun
Content may be subject to copyright.
Source publication
Climate impact models often require unbiased point‐scale observations, but climate models typically provide biased simulations at the grid scale. While standard bias adjustment methods have shown to generally perform well at adjusting climate model biases, they cannot overcome the gap between grid‐box and point‐scale. To overcome this limitation, c...
Similar publications
While recent increases in heavy precipitation events in some midlatitude regions are consistent with climate model simulations, evidence of such increases in high latitudes is more tenuous, partly because of data limitaftions. The present study evaluates historical and future changes in extreme precipitation events in Alaska. Using the ERA5 reanaly...
Citations
... The most common approaches to the bias adjustment of climate models include a simple adjustment of the mean (Linear Scaling or Delta Change), a mapping of the two entire cumulative distribution functions (Quantile Mapping), or more advanced methods that also aim to preserve the trend projected in the climate model (such as CDFt or ISIMIP3BASD). The practice of 70 using bias adjustment methods to also downscale the climate model has been criticized in various publications (von Storch, 1999;Maraun, 2013;Switanek et al., 2022), therefore this paper focuses on bias adjustment of climate models purely for the purpose of reducing biases at constant resolution. ...
Statistical bias adjustment is commonly applied to climate models before using their results in impact studies. However, different methods, based on a distributional mapping between observational and model data, can change the simulated trends, as well as the spatiotemporal and inter-variable consistency of the model, and are prone to misuse if not evaluated thoroughly. Despite the importance of these fundamental issues, researchers who apply bias adjustment currently do not have the tools at hand to compare different methods or evaluate the results sufficiently to detect possible distortions. Because of this, widespread practice in statistical bias adjustment is not aligned with recommendations from the academic literature. To address the practical issues impeding this, we introduce ibicus, an open-source Python package for the implementation of eight different peer-reviewed and widely used bias adjustment methods in a common framework and their comprehensive evaluation. The evaluation framework introduced in ibicus allows the user to analyse changes to the marginal, spatiotemporal and inter-variable structure of user-defined climate indices and distributional properties, as well as any alteration of the climate change trend simulated in the model. Applying ibicus in a case study over the Mediterranean region using seven CMIP6 global circulation models, this study finds that the most appropriate bias adjustment method depends on the variable and impact studied and that even methods that aim to preserve the climate change trend can modify it. These findings highlight the importance of a use-case-specific choice of method and the need for a rigorous evaluation of results when applying statistical bias adjustment.
... Rainfall data is often stochastic in nature (Sivakumar, 2016;Takhellambam et al., 2022;Vosper et al., 2022), skewing observations made during periods of wet weather. It is therefore better to impute values using the period-specific median of observations made in periods of wet weather (Hadeed et al., 2020;Switanek et al., 2022). Rows of data containing any remaining missing data values were then removed, as they occurred due to a lack of available data to determine a mean or median to fill or replace values with for that period. ...
The expected performance of Green Stormwater Infrastructure (GSI) is typically quantified through numerical models based on hydrologic parameters and physics-based equations. With numerical models, the choice of a spatio-temporal discretization scheme for the computational domain is a strenuous task that requires extensive calibration and potentially lab-based parameters and experimentation. The performance of GSI has high temporal dynamics due to natural, anthropogenic, and climatic processes that are not well represented by the traditional physics-based hydrologic models, which are calibrated against only a few historical observations and have a user-defined and constrained set of computational outcomes. Deep learning-based predictive models, such as Long Short-Term Memory (LSTM) neural networks, offer an exciting opportunity to quantify GSI performance, accounting for its highly dynamic and constantly evolving nature by leveraging advancements in observational data. A LSTM regression can overcome some of the limitations associated with traditional hydrological models to aid the development of a fully data-informed GSI performance predictor. To demonstrate the LSTM and traditional model outcomes, both methods were applied to a rain garden in Villanova, PA, USA. Specifically, an LSTM model was used to predict the recession of ponded water depth in the rain garden using five years of observed data. A set of eight predictors (i.e., precipitation, air temperature, soil temperature, soil moisture content at a depth of 10 cm, 35 cm, 65 cm and 91 cm, water depth) and a target variable (i.e., recession rate) were considered for training/testing the LSTM model. A comparative study with the USEPA Storm Water Management Model (SWMM) was performed to observe the performance of a physics-based model and a LSTM model for continuous recession rate time series and specific storms. The LSTM model had a performance score, Root Mean Square Error (RMSE), of 0.081 for the continuous time series, outperforming the SWMM performance with a score of 2.173 when compared to observed data. In the case of storm-specific prediction, LSTM also outperformed SWMM simulation for four storms with lower RMSE values when compared to the observed data. The application of the LSTM model in predicting GSI performance is a crucial stride towards efficient real-time forecasting.
... The issue of climate models' uncertainties is especially concerning for impact studies at the watershed scale, for which unbiased precipitation is needed at small grids, requiring the implementation of RCMs. Many bias reduction methods have been applied in order to make climate models' outputs suitable for impact studies [Foughali et al., 2015, Dakhlaoui et al., 2022, Switanek et al., 2022. ...
The objective of the research is to explore the predictability of water resources directly with GCMs by analysing long-term effects of climate change on Northern Tunisia’s blue and green water. Hydrologic impacts rely on a rainfall-runoff lumped model using outputs of CMIP6 GCMs within the framework of the ssp2-45 scenario. Among the 30 CMIP6 models, the composite cnrm-esm2-1 and fgoals-g3 best restore observed runoff from 1995 to 2014 and give the best GCM. Hydrologic projections 2015–2100 show significant drops in rainfall (9%), runoff (21%), groundwater recharge (15%), as well as for green water (6%). The results show that the use of raw GCMs predictions on large basins is possible and provides precisions comparable to what is produced when using Regional Climate Models in medium size basins.
... This was made evident here for temperature, due to the highly non-linear thermal sensitivity of snow variables. If it seeks to be relevant, BC is not necessarily simple (Switanek et al., 2022;Robin et al., 2023). For temperature, BC is typically applied for the temperature of reference stations. ...
We assess the ability of two typical simulation chains to reproduce, over the last century (1902–2009) and from large-scale atmospheric information only, the temporal variations of river discharges, low flow sequences and flood events, observed at different locations of the Upper Rhône River (URR) catchment, an alpine river straddling France and Switzerland (10,900 km2). The two simulation chains are made up of a downscaling model, either statistical (SCAMP) or dynamical (MAR), and the glacio-hydrological model GSM-SOCONT. Both downscaling models, forced by atmospheric information from the ERA-20C atmospheric reanalysis, provide time series of daily scenarios of precipitation and temperature used as input to the hydrological model. With hydrological regimes ranging from highly glaciated ones in its upper part to mixed ones dominated by snow and rain downstream, the URR catchment is ideal to evaluate the different simulation chains in contrasting and demanding hydro-meteorological configurations where the interplay between weather variables, both in space and time, is determinant. Whatever the river sub-basin considered, the simulated discharges are in good agreement with the reference ones, provided that the weather scenarios are bias-corrected. The observed multi-scale variations of discharges (daily, seasonal and interannual) are well reproduced and the hydrological situations of low frequency (low flow sequences and flood events) are reasonably well reproduced. Bias correction is crucial for both precipitation and temperature and both downscaling models. For the dynamical one, a bias correction is also essential to get realistic daily temperature lapse rates. Uncorrected scenarios lead to irrelevant hydrological simulations, especially for the sub-catchments at high elevation, mainly due to irrelevant snowpack dynamic simulations. The simulations also highlight the difficulty to simulate precipitation dependency to elevation over mountainous areas.
... Future work should investigate the impact of this effect. Additionally, it would be of interest to extend the bias correction approaches presented here to consider the spatial dependence across grid cells to reduce the uncertainty in the bias estimates (Kim et al. 2021;Switanek et al. 2022). ...
Improving modeling capacities requires a better understanding of both the physical relationship between the variables and climate models with a higher degree of skill than is currently achieved by Global Climate Models (GCMs). Although Regional Climate Models (RCMs) are commonly used to resolve finer scales, their application is restricted by the inherent systematic biases within the GCM datasets that can be propagated into the RCM simulation through the model input boundaries. Hence, it is advisable to remove the systematic biases in the GCM simulations prior to downscaling, forming improved input boundary conditions for the RCMs. Various mathematical approaches have been formulated to correct such biases. Most of the techniques, however, correct each variable independently leading to physical inconsistencies across the variables in dynamically linked fields. Here, we investigate bias corrections ranging from simple to more complex techniques to correct biases of RCM input boundary conditions. The results show that substantial improvements in model performance are achieved after applying bias correction to the boundaries of RCM. This work identifies that the effectiveness of increasingly sophisticated techniques is able to improve the simulated rainfall characteristics. An RCM with multivariate bias correction, which corrects temporal persistence and inter-variable relationships, better represents extreme events relative to univariate bias correction techniques, which do not account for the physical relationship between the variables.
... In this study we feed a semi-distributed implementation of a processbased, conceptual hydrological model with 28 bias-adjusted and downscaled climate change projections on a daily basis until 2100 (Switanek et al., 2022) (Fig. 2). The hydrological model was previously calibrated and exhaustively tested for the six study regions in the Austrian Alps over a period of 40+ years (Prenner et al., , 2019. ...
... Recent research has identified severe artefacts when attempting to use bias adjustment as a simple statistical downscaling (Maraun, 2013;Maraun et al., 2017). We therefore apply a novel approach that separates the adjustment from the downscaling: The regional EURO-CORDEX projections were first bias adjusted against gridded reference observations and then further downscaled to the weather monitoring stations by a spatial stochastic downscaling model using daily observational measurement data from the respective stations (Switanek et al., 2022). The bias adjustment is conducted with SDM (Switanek et al., 2017). ...
... The model performs excellent for a range of diagnostics representing marginal, temporal and spatial aspects for moderate and extreme precipitation. Methodological details about the stochastic downscaling method as well as a comprehensive evaluation can be found in Switanek et al. (2022). That study demonstrated good performance at simulating mean and intense precipitation, dry-and wet-day probabilities, spatial correlations and spatial dependence of extreme rainfall for the catchments addressed in the study at hand. ...
Debris-flow activity is strongly controlled by hydro-meteorological trigger conditions, which are expected to change in a future climate. In this study we connect a regional hydro-meteorological susceptibility model for debris flows with climate projections until 2100 to assess changes of the frequency of critical trigger conditions for different trigger types (long-lasting rainfall, short-duration storm, snow-melt, rain-on-snow) in six regions in the Austrian Alps. We find limited annual changes of the number of days critical for debris-flow initiation when averaged over all regions, but distinct changes when separating between hydro-meteorological trigger types and study region. Changes become more evident at the monthly/seasonal scale, with a general trend of critical debris-flow trigger conditions earlier in the year. The outcomes of this study serve as a basis for the development of adaption strategies for future risk management.
... RCP8.5 represents a pathway with increasing greenhouse gas emissions and no mitigation measures. The simulations provide the temperature and precipitation data on a daily basis at the station scale corresponding to the location of precipitation and temperature stations ( Fig. 1; Switanek et al., 2021). ...
Hydrological regimes of alpine catchments are expected to be strongly affected by climate change, mostly due to their dependence on snow and ice dynamics. While seasonal changes have been studied extensively, studies on changes in the timing and magnitude of annual extremes remain rare. This study investigates the effects of climate change on runoff patterns in six contrasting Alpine catchments in Austria using a process-based, semi-distributed hydrological model and projections from 14 regional and global climate model combinations for two representative concentration pathways, namely RCP4.5 and RCP8.5. The study catchments represent a spectrum of different hydrological regimes, from pluvial–nival to nivo-glacial, as well as distinct topographies and land forms, characterizing different elevation zones across the eastern Alps to provide a comprehensive picture of future runoff changes. The climate projections are used to model river runoff in 2071–2100, which are then compared to the 1981–2010 reference period for all study catchments. Changes in the timing and magnitude of annual maximum and minimum flows, as well as in monthly runoff and snowmelt, are quantified and analyzed. Our results indicate a substantial shift to earlier occurrences in annual maximum flows by 9 to 31 d and an extension of the potential flood season by 1 to 3 months for high-elevation catchments. For low-elevation catchments, changes in the timing of annual maximum flows are less pronounced. Magnitudes of annual maximum flows are likely to increase by 2 %–18 % under RCP4.5, while no clear changes are projected for four catchments under RCP8.5. The latter is caused by a pronounced increase in evaporation and decrease in snowmelt contributions, which offset increases in precipitation. In the future, minimum annual runoff will occur 13–31 d earlier in the winter months for high-elevation catchments, whereas for low-elevation catchments a shift from winter to autumn by about 15–100 d is projected, with generally larger changes for RCP8.5. While all catchments show an increase in mean magnitude of minimum flows by 7–30% under RCP4.5, this is only the case for four catchments under RCP8.5. Our results suggest a relationship between the elevation of catchments and changes in the timing of annual maximum and minimum flows. For the magnitude of the extreme flows, a relationship is found between catchment elevation and annual minimum flows, whereas this relationship is lacking between elevation and annual maximum flow.
... scenarios RCP4.5 and RCP8.5 (Switanek et al., 2017). The simulations provide temperature and precipitation data on a daily basis at the station scale corresponding to the location of precipitation and temperature stations ( Fig.1) (Switanek et al., 2021). ...
Hydrological regimes of alpine catchments are expected to be strongly affected by climate change mostly due to their dependence on snow and ice dynamics. While seasonal changes have been studied extensively, studies on changes in the timing and magnitude of annual extremes remain rare. This study investigates the effects of climate change on runoff patterns in six contrasting alpine catchments in Austria using a process-based semi-distributed hydrological model and projections from 14 regional climate and global climate model combinations for RCP 4.5 and RCP 8.5. The study catchments represent a spectrum of different hydrological regimes, from pluvial-nival to nivo-glacial, as well as distinct topographies and land forms, characterizing different elevation zones across the Eastern Alps to provide a comprehensive picture of future runoff changes. The climate projections are used to model river runoff in 2071–2100, which are then compared to the 1981–2010 reference period for all study catchments. Changes in timing and magnitude of annual maximum and minimum flows as well as in monthly runoff and snow melt are quantified and analyzed. Our results indicate a substantial shift to earlier occurrences in annual maximum flows by 9 to 31 days and an extension of the potential flood season by one to three months for high-elevation catchments. For low-elevation catchments, changes in timing of annual maximum flows are less pronounced. Magnitudes of annual maximum flows are likely to increase by 2–18 % under RCP 4.5, while no clear changes are projected for four catchments under RCP 8.5. The latter is caused by a pronounced increase in evaporation and decrease in snow melt contributions which offset increases in precipitation. Minimum annual runoff occur 13–31 days earlier in the winter months for high-elevation catchments, whereas for low-elevation catchments a shift from winter to autumn by about 15–100 days is projected. While all catchments show an increase in mean magnitude of minimum flows by 7–30 % under RCP 4.5, this is only the case for four catchments under RCP 8.5. Our results suggest a relationship between the elevation of catchments and changes in timing of annual maximum and minimum flows. For the magnitude of the extreme flows, a relationship is found between catchment elevation and annual minimum flows, whereas this relationship is lacking between elevation and annual maximum flow.
Societally relevant weather impacts typically result from compound events, which are rare combinations of weather and climate drivers. Focussing on four event types arising from different combinations of climate variables across space and time, here we illustrate that robust analyses of compound events — such as frequency and uncertainty analysis under present-day and future conditions, event attribution to climate change, and exploration of low-probability-high-impact events — require data with very large sample size. In particular, the required sample is much larger than that needed for analyses of univariate extremes. We demonstrate that Single Model Initial-condition Large Ensemble (SMILE) simulations from multiple climate models, which provide hundreds to thousands of years of weather conditions, are crucial for advancing our assessments of compound events and constructing robust model projections. Combining SMILEs with an improved physical understanding of compound events will ultimately provide practitioners and stakeholders with the best available information on climate risks.
