scholarly journals Adjusting spatial dependence of climate model outputs with cycle-consistent adversarial networks

2021 ◽  
Author(s):  
Bastien François ◽  
Soulivanh Thao ◽  
Mathieu Vrac

AbstractClimate model outputs are commonly corrected using statistical univariate bias correction methods. Most of the time, those 1d-corrections do not modify the ranks of the time series to be corrected. This implies that biases in the spatial or inter-variable dependences of the simulated variables are not adjusted. Hence, over the last few years, some multivariate bias correction (MBC) methods have been developed to account for inter-variable structures, inter-site ones, or both. As proof-of-concept, we propose to adapt a computer vision technique used for Image-to-Image translation tasks (CycleGAN) for the adjustment of spatial dependence structures of climate model projections. The proposed algorithm, named MBC-CycleGAN, aims to transfer simulated maps (seen as images) with inappropriate spatial dependence structure from climate model outputs to more realistic images with spatial properties similar to the observed ones. For evaluation purposes, the method is applied to adjust maps of temperature and precipitation from climate simulations through two cross-validation approaches. The first one is designed to assess two different post-processing schemes (Perfect Prognosis and Model Output Statistics). The second one assesses the influence of nonstationary properties of climate simulations on the performance of MBC-CycleGAN to adjust spatial dependences. Results are compared against a popular univariate bias correction method, a “quantile-mapping” method, which ignores inter-site dependencies in the correction procedure, and two state-of-the-art multivariate bias correction algorithms aiming to adjust spatial correlation structure. In comparison with these alternatives, the MBC-CycleGAN algorithm reasonably corrects spatial correlations of climate simulations for both temperature and precipitation, encouraging further research on the improvement of this approach for multivariate bias correction of climate model projections.

2021 ◽  
Author(s):  
Bastien François ◽  
Soulivanh Thao ◽  
Mathieu Vrac

Abstract Climate model outputs are commonly corrected using statistical univariate bias correction methods. Most of the time, those 1d-corrections do not modify the ranks of the time series to be corrected. This implies that biases in the spatial or inter-variable dependences of the simulated variables are not adjusted. Hence, over the last few years, some multivariate bias correction (MBC) methods have been developed to account for inter-variable structures, inter-site ones, or both. As proof-of-concept, we propose to adapt a computer vision technique used for Image-to-Image translation tasks (CycleGAN) for the adjustment of spatial dependence structures of climate model projections. The proposed algorithm, named MBC-CycleGAN, aims to transfer simulated maps (seen as images) with inappropriate spatial dependence structure from climate model outputs to more realistic images with spatial properties similar to the observed ones. For evaluation purposes, the method is applied to adjust maps of temperature and precipitation from climate simulations through two cross-validation approaches. The first one is designed to assess two different post-processing schemes (Perfect Prognosis and Model Output Statistics). The second one assesses the influence of nonstationary properties of climate simulations on the performance of MBC-CycleGAN to adjust spatial dependences. Results are compared against a popular univariate bias correction method, a ``quantile-mapping'' method, which ignores inter-site dependencies in the correction procedure, and two state-of-the-art multivariate bias correction algorithms aiming to adjust spatial correlation structure. In comparison with these alternatives, the MBC-CycleGAN algorithm reasonably corrects spatial correlations of climate simulations for both temperature and precipitation, encouraging further research on the improvement of this approach for multivariate bias correction of climate model projections.


2021 ◽  
Author(s):  
Bastien François ◽  
Soulivanh Thao ◽  
Mathieu Vrac

<p>Climate model outputs are commonly corrected using statistical univariate bias correction methods. Most of the time, those 1d-corrections do not modify the ranks of the time series to be corrected. This implies that biases in the spatial or inter-variable dependences of the simulated variables are not adjusted. Hence, over the last few years, some multivariate bias correction (MBC) methods have been developed to account for inter-variable structures, inter-site ones, or both. As proof-of-concept, we propose to adapt  a computer vision technique used for Image-to-Image translation tasks (CycleGAN) for the adjustment of spatial dependence structures of climate model projections. The proposed algorithm, named MBC-CycleGAN, aims to transfer simulated maps (seen as images) with inappropriate spatial dependence structure from climate model outputs to more realistic images with spatial properties similar to the observed ones. For evaluation purposes, the method is applied to adjust maps of temperature and precipitation from climate simulations through two cross-validation approaches. The first one is designed to assess two different post-processing schemes (Perfect Prognosis and Model Output Statistics). The second one assesses the influence of non-stationary properties of climate simulations on the performance of MBC-CycleGAN to adjust spatial dependences. Results are compared against a popular univariate bias correction method, a "quantile-mapping" method, which ignores inter-site dependencies in the correction procedure, and two state-of-the-art multivariate bias correction algorithms aiming to adjust spatial correlation structure. In comparison with these alternatives, the MBC-CycleGAN algorithm reasonably corrects spatial correlations of climate simulations for both temperature and precipitation, encouraging further research on the improvement of this approach for multivariate bias correction of climate model projections.</p>


2018 ◽  
Vol 22 (6) ◽  
pp. 3175-3196 ◽  
Author(s):  
Mathieu Vrac

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 (R2D2) bias correction allowing one to adjust not only the univariate distributions but also their inter-variable and inter-site dependence structures. Moreover, the proposed R2D2 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. R2D2 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 R2D2 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-R2D2 is only satisfying in the inter-variable context, 1506d-R2D2 strongly improves inter-site properties and 3012d-R2D2 is able to account for both. Applications of the proposed R2D2 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.


2018 ◽  
Author(s):  
Mathieu Vrac

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, while 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 (R2D2) bias correction allowing to adjust not only the univariate distributions, but also their inter-variable and inter-site dependence structures. Moreover, the proposed R2D2 method provides some stochasticity since it can generate as many multivariate corrected outputs as the number of statistical dimensions (i.e., number of grid-cells × number of climate variables) of the simulations to be corrected. It is based on an assumption of stability in time of the dependence structure – allowing to deal with a high number of statistical dimensions –, that lets the climate model drive the temporal properties and their changes in time. R2D2 is applied on temperature and precipitation reanalyses time series with respect to high-resolution reference data over South-East of France (1506 grid-cells). Bivariate, 1506-dimensional and 3012-dimensional versions of R2D2 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-R2D2 is only satisfying in the inter-variable context, 1506d-R2D2 strongly improves inter-site properties and 3012d-R2D2 is able to account for both. Applications of the proposed R2D2 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.


2012 ◽  
Vol 16 (2) ◽  
pp. 305-318 ◽  
Author(s):  
I. Haddeland ◽  
J. Heinke ◽  
F. Voß ◽  
S. Eisner ◽  
C. Chen ◽  
...  

Abstract. Due to biases in the output of climate models, a bias correction is often needed to make the output suitable for use in hydrological simulations. In most cases only the temperature and precipitation values are bias corrected. However, often there are also biases in other variables such as radiation, humidity and wind speed. In this study we tested to what extent it is also needed to bias correct these variables. Responses to radiation, humidity and wind estimates from two climate models for four large-scale hydrological models are analysed. For the period 1971–2000 these hydrological simulations are compared to simulations using meteorological data based on observations and reanalysis; i.e. the baseline simulation. In both forcing datasets originating from climate models precipitation and temperature are bias corrected to the baseline forcing dataset. Hence, it is only effects of radiation, humidity and wind estimates that are tested here. The direct use of climate model outputs result in substantial different evapotranspiration and runoff estimates, when compared to the baseline simulations. A simple bias correction method is implemented and tested by rerunning the hydrological models using bias corrected radiation, humidity and wind values. The results indicate that bias correction can successfully be used to match the baseline simulations. Finally, historical (1971–2000) and future (2071–2100) model simulations resulting from using bias corrected forcings are compared to the results using non-bias corrected forcings. The relative changes in simulated evapotranspiration and runoff are relatively similar for the bias corrected and non bias corrected hydrological projections, although the absolute evapotranspiration and runoff numbers are often very different. The simulated relative and absolute differences when using bias corrected and non bias corrected climate model radiation, humidity and wind values are, however, smaller than literature reported differences resulting from using bias corrected and non bias corrected climate model precipitation and temperature values.


2017 ◽  
Vol 21 (6) ◽  
pp. 2649-2666 ◽  
Author(s):  
Matthew B. Switanek ◽  
Peter A. Troch ◽  
Christopher L. Castro ◽  
Armin Leuprecht ◽  
Hsin-I Chang ◽  
...  

Abstract. Commonly used bias correction methods such as quantile mapping (QM) assume the function of error correction values between modeled and observed distributions are stationary or time invariant. This article finds that this function of the error correction values cannot be assumed to be stationary. As a result, QM lacks justification to inflate/deflate various moments of the climate change signal. Previous adaptations of QM, most notably quantile delta mapping (QDM), have been developed that do not rely on this assumption of stationarity. Here, we outline a methodology called scaled distribution mapping (SDM), which is conceptually similar to QDM, but more explicitly accounts for the frequency of rain days and the likelihood of individual events. The SDM method is found to outperform QM, QDM, and detrended QM in its ability to better preserve raw climate model projected changes to meteorological variables such as temperature and precipitation.


2020 ◽  
Author(s):  
Flavio Maria Emanuele Pons ◽  
Davide Faranda

Abstract. The description and analysis of compound extremes affecting mid and high latitudes in the winter requires an accurate estimation of snowfall. Such variable is often missing for in-situ observations, and biased in climate model outputs, both in magnitude and number of events. While climate models can be adjusted using bias correction (BC), snowfall presents additional challenges compared to other variables, preventing from applying traditional univariate BC methods. We extend the existing literature on the estimation of the snowfall fraction from near-surface temperature, which usually involves binary thresholds or fitting parametric nonlinear functions. We show that, combining breakpoint search algorithms to define threshold temperatures and segmented regression models, it is possible to obtain accurate out-of-sample estimates of snowfall over Europe in ERA5 reanalysis, and to perform effective BC on the IPSL-WRF high resolution EURO-CORDEX climate model only relying on bias adjusted temperature and precipitation. This method offers a feasible way to reconstruct or adjust snowfall observations without requiring multivariate or conditional bias correction and stochastic generation of unobserved events.


2016 ◽  
Author(s):  
Matthew B. Switanek ◽  
Peter A. Troch ◽  
Christopher L. Castro ◽  
Armin Leuprecht ◽  
Hsin-I. Chang ◽  
...  

Abstract. Commonly used bias correction methods such as quantile mapping (QM) assume the function of error correction values between modelled and observed distributions are stationary or time-invariant. This article finds that this function of the error correction values cannot be assumed to be stationary. As a result, QM lacks justification to inflate/deflate various moments of the climate change signal. Previous adaptations of QM, most notably quantile delta mapping (QDM), have been developed that do not rely on this assumption of stationarity. Here, we outline a methodology called scaled distribution mapping (SDM), which is conceptually similar to QDM, but more explicitly accounts for the frequency of rain days and the likelihood of individual events. The SDM method is found to outperform QM, QDM and detrended QM in its ability to better preserve raw climate model projected changes to meteorological variables such as temperature and precipitation.


2011 ◽  
Vol 8 (4) ◽  
pp. 7919-7945 ◽  
Author(s):  
I. Haddeland ◽  
J. Heinke ◽  
F. Voß ◽  
S. Eisner ◽  
C. Chen ◽  
...  

Abstract. Due to biases in the output of climate models, a bias correction is often needed to make the output suitable for use in hydrological simulations. In most cases only the temperature and precipitation values are bias corrected. However, often there are also biases in other variables such as radiation, humidity and wind speed. In this study we tested to what extent it is also needed to bias correct these variables. Responses to radiation, humidity and wind estimates from two climate models for four large-scale hydrological models are analysed. For the period 1971–2000 these hydrological simulations are compared to simulations using meteorological data based on observations and reanalysis; i.e. the baseline simulation. In both forcing datasets originating from climate models precipitation and temperature are bias corrected to the baseline forcing dataset. Hence, it is only effects of radiation, humidity and wind estimates that are tested here. The direct use of climate model outputs result in substantial different evapotranspiration and runoff estimates, when compared to the baseline simulations. A simple bias correction method is implemented and tested by rerunning the hydrological models using bias corrected radiation, humidity and wind values. The results indicate that bias correction can successfully be used to match the baseline simulations. Finally, historical (1971–2000) and future (2071–2100) model simulations resulting from using bias corrected forcings are compared to the results using non-bias corrected forcings. The relative changes in simulated evapotranspiration and runoff are relatively similar for the bias corrected and non bias corrected hydrological projections, although the absolute evapotranspiration and runoff numbers are often very different. The simulated relative and absolute differences when using bias corrected and non bias corrected climate model radiation, humidity and wind values are, however, smaller than literature reported differences resulting from using bias corrected and non bias corrected climate model precipitation and temperature values.


2016 ◽  
Vol 29 (19) ◽  
pp. 7045-7064 ◽  
Author(s):  
Alex J. Cannon

Abstract Univariate bias correction algorithms, such as quantile mapping, are used to address systematic biases in climate model output. Intervariable dependence structure (e.g., between different quantities like temperature and precipitation or between sites) is typically ignored, which can have an impact on subsequent calculations that depend on multiple climate variables. A novel multivariate bias correction (MBC) algorithm is introduced as a multidimensional analog of univariate quantile mapping. Two variants are presented. MBCp and MBCr respectively correct Pearson correlation and Spearman rank correlation dependence structure, with marginal distributions in both constrained to match observed distributions via quantile mapping. MBC is demonstrated on two case studies: 1) bivariate bias correction of monthly temperature and precipitation output from a large ensemble of climate models and 2) multivariate correction of vertical humidity and wind profiles, including subsequent calculation of vertically integrated water vapor transport and detection of atmospheric rivers. The energy distance is recommended as an omnibus measure of performance for model selection. As expected, substantial improvements in performance relative to quantile mapping are found in each case. For reference, characteristics of the MBC algorithm are compared against existing bivariate and multivariate bias correction techniques. MBC performs competitively and fills a role as a flexible, general purpose multivariate bias correction algorithm.


Sign in / Sign up

Export Citation Format

Share Document