Abstract:Abstract. In climate change impact research, the assessment of future river runoff as well as the catchment-scale water balance is impeded by different sources of modeling uncertainty. Some research has already been done in order to quantify the uncertainty of climate projections originating from the climate models and the downscaling techniques, as well as from the internal variability evaluated from climate model member ensembles. Yet, the use of hydrological models adds another layer of uncertainty. Within … Show more
“…This means that some of their statistical properties, such as mean, variance, distribution or even temporal, spatial or inter-variable dependence structures may not be representative of what is observed in the reference dataset. Consequently, before employing climate simulations to feed an impact model, it is often mandatory to "bias correct" (or to "adjust") them in order to correct some of their statistical properties (e.g., Christensen et al, 2008;Muerth et al, 2013).…”
Abstract. Climate simulations often suffer from statistical biases with respect to observations or reanalyses. It is therefore common to correct (or adjust) those simulations before using them as inputs into impact models. However, most bias correction (BC) methods are univariate and so do not account for the statistical dependences linking the different locations and/or physical variables of interest. In addition, they are often deterministic, and stochasticity is frequently needed to investigate climate uncertainty and to add constrained randomness to climate simulations that do not possess a realistic variability. This study presents a multivariate method of rank resampling for distributions and dependences (R 2 D 2 ) bias correction allowing one to adjust not only the univariate distributions but also their inter-variable and inter-site dependence structures. Moreover, the proposed R 2 D 2 method provides some stochasticity since it can generate as many multivariate corrected outputs as the number of statistical dimensions (i.e., number of grid cell × number of climate variables) of the simulations to be corrected. It is based on an assumption of stability in time of the dependence structure -making it possible to deal with a high number of statistical dimensions -that lets the climate model drive the temporal properties and their changes in time. R 2 D 2 is applied on temperature and precipitation reanalysis time series with respect to high-resolution reference data over the southeast of France (1506 grid cell). Bivariate, 1506-dimensional and 3012-dimensional versions of R 2 D 2 are tested over a historical period and compared to a univariate BC. How the different BC methods behave in a climate change context is also illustrated with an application to regional climate simulations over the 2071-2100 period. The results indicate that the 1d-BC basically reproduces the climate model multivariate properties, 2d-R 2 D 2 is only satisfying in the inter-variable context, 1506d-R 2 D 2 strongly improves inter-site properties and 3012d-R 2 D 2 is able to account for both. Applications of the proposed R 2 D 2 method to various climate datasets are relevant for many impact studies. The perspectives of improvements are numerous, such as introducing stochasticity in the dependence itself, questioning its stability assumption, and accounting for temporal properties adjustment while including more physics in the adjustment procedures.
“…This means that some of their statistical properties, such as mean, variance, distribution or even temporal, spatial or inter-variable dependence structures may not be representative of what is observed in the reference dataset. Consequently, before employing climate simulations to feed an impact model, it is often mandatory to "bias correct" (or to "adjust") them in order to correct some of their statistical properties (e.g., Christensen et al, 2008;Muerth et al, 2013).…”
Abstract. Climate simulations often suffer from statistical biases with respect to observations or reanalyses. It is therefore common to correct (or adjust) those simulations before using them as inputs into impact models. However, most bias correction (BC) methods are univariate and so do not account for the statistical dependences linking the different locations and/or physical variables of interest. In addition, they are often deterministic, and stochasticity is frequently needed to investigate climate uncertainty and to add constrained randomness to climate simulations that do not possess a realistic variability. This study presents a multivariate method of rank resampling for distributions and dependences (R 2 D 2 ) bias correction allowing one to adjust not only the univariate distributions but also their inter-variable and inter-site dependence structures. Moreover, the proposed R 2 D 2 method provides some stochasticity since it can generate as many multivariate corrected outputs as the number of statistical dimensions (i.e., number of grid cell × number of climate variables) of the simulations to be corrected. It is based on an assumption of stability in time of the dependence structure -making it possible to deal with a high number of statistical dimensions -that lets the climate model drive the temporal properties and their changes in time. R 2 D 2 is applied on temperature and precipitation reanalysis time series with respect to high-resolution reference data over the southeast of France (1506 grid cell). Bivariate, 1506-dimensional and 3012-dimensional versions of R 2 D 2 are tested over a historical period and compared to a univariate BC. How the different BC methods behave in a climate change context is also illustrated with an application to regional climate simulations over the 2071-2100 period. The results indicate that the 1d-BC basically reproduces the climate model multivariate properties, 2d-R 2 D 2 is only satisfying in the inter-variable context, 1506d-R 2 D 2 strongly improves inter-site properties and 3012d-R 2 D 2 is able to account for both. Applications of the proposed R 2 D 2 method to various climate datasets are relevant for many impact studies. The perspectives of improvements are numerous, such as introducing stochasticity in the dependence itself, questioning its stability assumption, and accounting for temporal properties adjustment while including more physics in the adjustment procedures.
“…3126 A. M. Jobst et al: Intercomparison of different uncertainty sources climate models and a separate higher resolution hydrological model (Maraun et al, 2010;Muerth et al, 2013).…”
Section: Introductionmentioning
confidence: 99%
“…As discussed by Muerth et al (2013), the hydro-climatic model chain typically consists of the following components: emission scenario, GCM, regional climate model (RCM) or statistical downscaling, bias correction, and hydrological model. All of these components constitute a potential uncertainty source, and as such all need to be examined to provide a truly comprehensive understanding of the uncertainty associated with hydrological impact assessments (Teutschbein and Seibert, 2010).…”
Abstract. As climate change is projected to alter both temperature and precipitation, snow-controlled mid-latitude catchments are expected to experience substantial shifts in their seasonal regime, which will have direct implications for water management. In order to provide authoritative projections of climate change impacts, the uncertainty inherent to all components of the modelling chain needs to be accounted for. This study assesses the uncertainty in potential impacts of climate change on the hydro-climate of a headwater sub-catchment of New Zealand's largest catchment (the Clutha River) using a fully distributed hydrological model (WaSiM) and unique ensemble encompassing different uncertainty sources: general circulation model (GCM), emission scenario, bias correction and snow model. The inclusion of snow models is particularly important, given that (1) they are a rarely considered aspect of uncertainty in hydrological modelling studies, and (2) snow has a considerable influence on seasonal patterns of river flow in alpine catchments such as the Clutha. Projected changes in river flow for the 2050s and 2090s encompass substantial increases in streamflow from May to October, and a decline between December and March. The dominant drivers are changes in the seasonal distribution of precipitation (for the 2090s +29 to +84 % in winter) and substantial decreases in the seasonal snow storage due to temperature increase. A quantitative comparison of uncertainty identified GCM structure as the dominant contributor in the seasonal streamflow signal (44-57 %) followed by emission scenario (16-49 %), bias correction (4-22 %) and snow model (3-10 %). While these findings suggest that the role of the snow model is comparatively small, its contribution to the overall uncertainty was still found to be noticeable for winter and summer.
“…Although most of the simulations are run on a high grid resolution, systematic biases in the regional climate models (RCMs) remain, due to errors related to (i) imperfect model representation of the physical processes or phenomena and (ii) to the parametrization and incorrect initialization of the models. Thus, even when using the highest resolution available, RCMs still require some adjustments (Christensen et al, 2008;Muerth et al, 2013). Therefore, bias correction methods continue to be used in impact studies -for example in hydrology (e.g.…”
Abstract. The CHASE-PL (Climate change impact assessment for selected sectors in Poland) Climate Projections -Gridded Daily Precipitation and Temperature dataset 5 km (CPLCP-GDPT5) consists of projected daily minimum and maximum air temperatures and precipitation totals of nine EURO-CORDEX regional climate model outputs bias corrected and downscaled to a 5 km × 5 km grid. Simulations of one historical period and two future horizons (2021-2050 and 2071-2100) assuming two representative concentration pathways (RCP4.5 and RCP8.5) were produced. We used the quantile mapping method and corrected any systematic seasonal bias in these simulations before assessing the changes in annual and seasonal means of precipitation and temperature over Poland. Projected changes estimated from the multi-model ensemble mean showed that annual means of temperature are expected to increase steadily by 1 • C until 2021-2050 and by 2 • C until 2071-2100 assuming the RCP4.5 emission scenario. Assuming the RCP8.5 emission scenario, this can reach up to almost 4 • C by 2071-2100. Similarly to temperature, projected changes in regional annual means of precipitation are expected to increase by 6 to 10 % and by 8 to 16 % for the two future horizons and RCPs, respectively. Similarly, individual model simulations also exhibited warmer and wetter conditions on an annual scale, showing an intensification of the magnitude of the change at the end of the 21st century. The same applied for projected changes in seasonal means of temperature showing a higher winter warming rate by up to 0.5 • C compared to the other seasons. However, projected changes in seasonal means of precipitation by the individual models largely differ and are sometimes inconsistent, exhibiting spatial variations which depend on the selected season, location, future horizon, and RCP. The overall range of the 90 % confidence interval predicted by the ensemble of multi-model simulations was found to likely vary between −7 % (projected for summer assuming the RCP4.5 emission scenario) and +40 % (projected for winter assuming the RCP8.5 emission scenario) by the end of the 21st century. Finally, this high-resolution bias-corrected product can serve as a basis for climate change impact and adaptation studies for many sectors over Poland. The CPLCP-GDPT5 dataset is publicly available at
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.