Biases In General Circulation Models Research Articles

Comparative Analysis of Climate Change Impacts on Climatic Variables and Reference Evapotranspiration in Tunisian Semi-Arid Region

Systematic biases in general circulation models (GCM) and regional climate models (RCM) impede their direct use in climate change impact research. Hence, the bias correction of GCM-RCMs outputs is a primary step in such studies. This study compares the potential of two bias correction methods (the method from the third phase of the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP3) and Detrended Quantile Matching (DQM)) applied to the raw outputs of daily data of minimum and maximum air temperatures and precipitation, in the Cap-Bon region, from eight GCM-RCM combinations. The outputs of GCM/RCM combinations were acquired from the European branch of the coordinated regional climate downscaling experiment (EURO-CORDEX) dataset for historical periods and under two representative concentration pathway (RCP4.5 and RCP8.5) scenarios. Furthermore, the best combination of bias correction/GCM-RCM was used to assess the impact of climate change on reference evapotranspiration (ET0). Numerous statistical indicators were considered to evaluate the performance of the bias correction/historical GCM-RCMs compared to the observed data. Trends of the Hargreaves–Samani_ET0 model during the historical and projected periods were determined using the TFPMK method. A comparison of the bias correction methods revealed that, for all the studied model combinations, ISIMIP3 performs better in reducing biases in monthly precipitation. However, for Tmax and Tmin, the biases are greatly removed when the DQM bias correction method is applied. In general, better results were obtained when the HadCCLM model was used. Before applying bias correction, the set of used GCM-RCMs projected reductions in precipitation for most of the months compared to the reference period (1982–2006). However, Tmin and Tmax are expected to increase in all months and for the three studied periods. Hargreaves–Samani ET0 values obtained from the best combination (DQM/ HadCCLM) show that RCP8.5 (2075–2098) will exhibit the highest annual ET0 increase compared to the RCP4.5 scenario and the other periods, with a change rate equal to 11.85% compared to the historical period. Regarding spring and summer seasons, the change rates of ET0 are expected to reach 10.44 and 18.07%, respectively, under RCP8.5 (2075–2098). This study shows that the model can be used to determine long-term trends in ET0 patterns for diverse purposes, such as water resources planning, agricultural crop management and irrigation scheduling in the Cap-Bon region.

Evaluation of bias correction techniques for generating high-resolution daily temperature projections from CMIP6 models

Due to the considerable biases in general circulation models (GCMs) simulation, bias correction methods are required and widely applied to reduce the model biases for impact studies. This study evaluated the performance of two bias correction methods, quantile delta mapping (QDM) and scaled distribution mapping (SDM), for generating high-resolution daily maximum temperature (Tmax) and minimum temperature (Tmin) projections for Canada using the latest GCMs from the Coupled Model Intercomparison Project phase 6 (CMIP6). CMIP6 GCMs show overall consistency with observations before and after bias correction, with better performance on Tmax compared to Tmin. QDM shows better performance relative to observations while SDM shows superior skill in preserving the raw climate signals. QDM and SDM methods are effective in reducing the biases of Tmax and Tmin for all GCMs. Both methods show similar skills in reproducing monthly probability distribution and capturing seasonal spatial patterns over Canada. Multi-model ensemble means have good performance in simulating the monthly mean of Tmax and Tmin but poor performance for high and low quantiles as well as standard deviation. QDM and SDM corrected ensemble means have the best performance. This study presents a comprehensive assessment of bias correction methods applications for individual CMIP6 GCMs and their multi-model ensemble means for high-resolution daily temperature predictions for Canada, providing a reference significance for bias correction studies as well as technical support for further impact assessment and adaptation planning around the world.

Modeling general circulation model bias via a combination of localized regression and quantile mapping methods

Abstract. General circulation model (GCM) outputs are a primary source of information for climate change impact assessments. However, raw GCM data rarely are used directly for regional-scale impact assessments as they frequently contain systematic error or bias. In this article, we propose a novel extension to standard quantile mapping that allows for a continuous seasonal change in bias magnitude using localized regression. Our primary goal is to examine the efficacy of this tool in the context of larger statistical downscaling efforts on the tropical island of Puerto Rico, where localized downscaling can be particularly challenging. Along the way, we utilize a multivariate infilling algorithm to estimate missing data within an incomplete climate data network spanning Puerto Rico. Next, we apply a combination of multivariate downscaling methods to generate in situ climate projections at 23 locations across Puerto Rico from three general circulation models in two carbon emission scenarios: RCP4.5 and RCP8.5. Finally, our bias-correction methods are applied to these downscaled GCM climate projections. These bias-correction methods allow GCM bias to vary as a function of a user-defined season (here, Julian day). Bias is estimated using a continuous curve rather than a moving window or monthly breaks. Results from the selected ensemble agree that Puerto Rico will continue to warm through the coming century. Under the RCP4.5 forcing scenario, our methods indicate that the dry season will have increased rainfall, while the early and late rainfall seasons will likely have a decline in total rainfall. Our methods applied to the RCP8.5 forcing scenario favor a wetter climate for Puerto Rico, driven by an increase in the frequency of high-magnitude rainfall events during Puerto Rico's early rainfall season (April to July) as well as its late rainfall season (August to November).

Global Impact of Cloud Longwave Scattering in an Atmosphere‐Only General Circulation Model Simulation

AbstractMost general circulation models (GCMs) neglect cloud longwave scattering in pursuit of computational efficiency. This study implements the 2‐/4‐stream (2/4S) method, a relatively fast cloud longwave scattering treatment, in Community Atmospheric Model version 5 (CAM5) to analyze the impact of cloud longwave scattering on the large‐scale circulation. Two 45‐years‐long integrations are performed with prescribed sea surface temperature (SST). In the experiment run, cloud longwave scattering is included using the 2/4S method; in the control run, clouds only absorb in the longwave. The results show that cloud longwave scattering acts to enhance the cloud longwave (greenhouse) effect by reducing outgoing longwave radiation (OLR) and enhancing downward longwave irradiance at the surface. The OLR reduction is most significant over the tropics, where surface temperatures and cloud elevations are high. The surface downward irradiance increase is most significant over polar areas and the Tibetan Plateau, where cloud elevations are low and the air below clouds is dry. Inclusion of cloud longwave scattering enhances the Walker circulation, suggestive of the importance of diabatic radiative heating in the tropical circulation. Inclusion of cloud longwave scattering also appears to shift the eddy‐driven jet poleward in the austral summer in the Southern Hemisphere, suggesting that the cloud longwave effect plays a role in shaping the jet position. Persistent equatorward jet biases in GCMs may be reduced if cloud longwave scattering is considered.

Intercomparison of Bias-Correction Data Sources and Their Influence on Watershed-Specific Downscaling Climate Projections

HighlightsBias correction from three historical data sources (NCDC, PRISM, and NEXRAD) were assessed.Bias correction removed watershed-average bias in general circulation model (GCM) data for a historical period.Subbasin-specific variability was detected in future projections for each bias-correction data source.Data source had less impact on uncertainty in future projections than GCM or a representative concentration pathway.Uncertainty from bias-correction data sources was higher for precipitation than for temperature.Abstract. Climate projections developed by general circulation models (GCM) are often used in watershed modeling applications to project future hydrologic changes. In many models, the climate projections are downscaled to individual map units represented by grid cells or subbasins. Uncertainty of downscaled climate projections are a product of uncertainties arising mainly from the model itself, from the representative concentration pathway (RCP), and from the downscaling procedure. Other sources of uncertainty may include the historical observations used for GCM bias correction and data aggregation from GCM grids to map (often subbasin) units. This study evaluated effects of three sources of historical data (ground-based weather station network, NCDC, and two gridded datasets, NEXRAD and PRISM) on historical variability, and shifts and uncertainty in precipitation and temperature projections. Climate projections from six GCMs and three RCPs were evaluated in 54 subbasins of the Smoky Hill River watershed in the U.S. Central Great Plains. Bias correction of GCM projections reduced bias of watershed-average annual precipitation in the historical period to near zero, but subbasin-specific variability remained in future projections with little difference among bias-correction data sources. For minimum and maximum temperatures, the GCM ensemble statistics for basin-average and subbasin-specific future projections were similar for all bias-correction data sources. Increase in RCP forcing was found to widen the uncertainty in future projections. Overall, the uncertainty due to data source selection was smaller than the uncertainty due to GCM model and RCP forcing selection. The results demonstrate that statistical downscaling is essential to account for local climate factors within a watershed, and that both weather station-based and gridded bias-correction data sources can be used effectively, but that future climate projections may inherit the historical bias in a selected data source. These inherent uncertainties associated with application of GCMs in hydrological and geospatial modeling should be carefully considered for understanding climate projections when building watershed models and interpreting the results. Keywords: Bias correction, Climate change, Downscaling, GCM, Uncertainty, Watershed.

An 18‐Year Climatology of Directional Stratospheric Gravity Wave Momentum Flux From 3‐D Satellite Observations

AbstractAtmospheric gravity waves (GWs) are key drivers of the atmospheric circulation, but their representation in general circulation models (GCMs) is challenging, leading to significant biases in middle atmospheric circulations. Unresolved GW momentum transport in GCMs must be parameterized, but global directional GW observations are needed to constrain this. Here we present an 18‐year climatology of directional stratospheric GW momentum flux (GWMF) from global AIRS/Aqua 3‐D satellite observations during 2002 to 2019. Striking hemispheric asymmetries are found at high latitudes, including dramatic reductions and reversals of GWMF during sudden stratospheric warmings. During Southern Hemisphere winter, a lateral convergence of GWMF toward 60°S is found that has no Northern Hemisphere counterpart. In the tropics, we find that zonal GWMF in AIRS measurements is strongly modulated by the semiannual oscillation (SAO) but not the quasi‐biennial oscillation (QBO). Our results provide guidance for future GW parameterizations needed to resolve long‐standing biases in GCMs.

Selection of general circulation models for the projections of spatio-temporal changes in temperature of Borneo Island based on CMIP5

A study has been conducted to evaluate the changes in temperature of Borneo Island for different representative concentration pathway (RCP) scenarios through statistical downscaling of an ensemble of general circulation models (GCMs) selected using entropy-based methods. A combination of past-performance and envelope approaches was used for the selection of GCM ensemble from a pool of 20 Coupled Model Intercomparison Project Phase 5 (CMIP5) GCMs. Ranking of the GCMs was done separately at each grid point by using three entropy-based feature selection methods. Climate Research Unit (CRU) gridded temperature data was used as dependent variable and the simulated temperature of different GCMs for the period 1961–2005 was used as independent variable for the selection of GCMs. The scores obtained by the GCMs at different CRU grids were aggregated using compromise programming method and then ranked for entire Borneo Island using group decision-making method. Linear scaling method was applied to remove the biases in GCMs, followed by random forest (RF)–based regression method to generate multi-model ensemble projections. The results revealed that FIO-ESM, MRI-CGCM3, GFDL-CM3, and IPSL-CM5A-MR are the most suitable GCMs for the projection of temperature in the study area. The projection of temperature using the selected GCMs indicated increase in temperature in the Borneo Island, particularly in the southwest regions. The minimum temperature was projected to increase more (3.3 to 4.7 °C) compared with maximum temperature (3.0 to 4.6 °C) and thus, the diurnal temperature range (DTR) was projected to decrease gradually until the end of the twenty-first century. Increase in average and maximum temperatures was projected more during the southwest (SW) monsoon compared with the northeast (NE) monsoon. The study indicated that large increase in temperature and decrease in DTR would have significant impact on ecology and bio-environment of the island which have one of the dense ecological diversities in the world.

Climate change uncertainties in seasonal drought severity-area-frequency curves: Case of arid region of Pakistan

The uncertainty assessment of the changes in drought characteristics due to climate change has caught the attention of the scientific community. This study used gauge-based gridded precipitation data obtained from Global Precipitation Climatology Centre (GPCC) to reconstruct historical droughts and downscale future precipitation projected by seven general circulation models (GCMs) of Coupled Model Inter-comparison Project phase 5 (CMIP5) under four Representative Concentration Pathways (RCP) scenarios, namely, RCP2.6, RCP4.5, RCP6.0 and RCP8.5. Support vector machine (SVM) and quantile mapping were used for downscaling and GCM bias correction, respectively. The model performances were assessed based on statistical measures. The historical and future projected precipitation data were finally used to characterize the seasonal droughts using Standardized Precipitation Index (SPI) for different crop growing periods. The drought severity-area-frequency (SAF) curves for the historical (1961–2010) and three future periods (2010–2039, 2040–2069, and 2070–2099) were developed. The uncertainty band of future drought SAF curves was estimated using Bayesian bootstrap (BB) at a 95% confidence level. As a result, SVM was successful in downscaling the precipitation of all selected CMIP5 GCMs. The seasonal ensemble of GCMs projected an increase in precipitation ranging from 8% to 41% under all scenarios. The historical SAF curves revealed that for equal drought severity, larger areas are affected by droughts having higher return periods. Future projections of droughts revealed the increase in affected area for lower severity and return period droughts and the decrease for higher severity and return period droughts. The uncertainty bands of drought SAF curves with higher return periods were found much wider compared to those with lower return periods which indicates more uncertainty in the projection of higher severity and return period droughts.

Long-term cloud fraction biases in CMIP5 GCMs over India during monsoon season

Using 24 years of cloud fraction (CF) data from the International Satellite Cloud Climatology Project (ISCCP) observations and their corresponding simulators in general circulation models (GCMs) from the Coupled Model Intercomparison Project phase 5 (CMIP5), we have analyzed cloud biases and their role on radiation over the Indian region (65–100° E and 5–40° N) for the monsoon season of June to September. The present study reports the spatial patterns of CFs and their biases in GCMs compared to observations. It is found that the simulated CFs are highly underestimated up to ~ 40%. Mean of total CF from ISCCP observations is 75% with at least 10% difference with simulated CFs. For high-topped clouds, this difference is about 3–4%. Except for high-topped clouds, other cloud types are not simulated realistically by CMIP5 models used in this study. Further, we investigated the individual cloud types classified based on cloud optical depth and cloud top pressure. We found that, in general, individual cloud types are poorly simulated by models, although some (Max Planck Institute Earth System Model, Low Resolution and Hadley Centre Global Environmental Model, version 2, Earth System) models convincingly simulate high-topped thin clouds. To assess the impact of cloud biases on the simulated radiative forcings, we studied shortwave and longwave cloud radiative forcings from CERES (Clouds and the Earth’s Radiant Energy System) observations and CMIP5 GCMs. It is noticed that the spatial patterns of biases in radiative forcings are similar to the patterns of biases in CFs for high-topped clouds, specifically over the oceanic regions. We find that the biases in cloud radiative forcings could potentially be caused due to the inefficacy of CMIP5 models in simulating high-topped anvil clouds (high-topped cirrus/stratocirrus clouds). The present study confirms that the uncertainty in simulating cloud fractions over the Indian region is still a prominent issue to be addressed in general circulation models.

Dynamical downscaling of regional climate: A review of methods and limitations

The traditional dynamical downscaling (TDD) method employs continuous integration of regional climate models (RCM) with the general circulation model (GCM) providing the initial and lateral boundary conditions. Dynamical downscaling simulations are constrained by physical principles and can generate a full set of climate information, providing one of the important approaches to projecting fine spatial-scale future climate information. However, the systematic biases of climate models often degrade the TDD simulations and hinder the application of dynamical downscaling in the climate-change related studies. New methods developed over past decades improve the performance of dynamical downscaling simulations. These methods can be divided into four groups: the TDD method, the pseudo global warming method, dynamical downscaling with GCM bias corrections, and dynamical downscaling with both GCM and RCM bias corrections. These dynamical downscaling methods are reviewed and compared in this paper. The merits and limitations of each dynamical downscaling method are also discussed. In addition, the challenges and potential directions in progressing dynamical downscaling methods are stated.

Hydrological Modeling to Evaluate Climate Model Simulations and Their Bias Correction

Abstract Variables simulated by climate models are usually evaluated independently. Yet, climate change impacts often stem from the combined effect of these variables, making the evaluation of intervariable relationships essential. These relationships can be evaluated in a statistical framework (e.g., using correlation coefficients), but this does not test whether complex processes driven by nonlinear relationships are correctly represented. To overcome this limitation, we propose to evaluate climate model simulations in a more process-oriented framework using hydrological modeling. Our modeling chain consists of 12 regional climate models (RCMs) from the Coordinated Downscaling Experiment–European Domain (EURO-CORDEX) forced by five general circulation models (GCMs), eight Swiss catchments, 10 optimized parameter sets for the hydrological model Hydrologiska Byråns Vattenbalansavdelning (HBV), and one bias correction method [quantile mapping (QM)]. We used seven discharge metrics to explore the representation of different hydrological processes under current climate. Specific combinations of biases in GCM–RCM simulations can lead to significant biases in simulated discharge (e.g., excessive precipitation in the winter months combined with a cold temperature bias). Other biases, such as exaggerated snow accumulation, do not necessarily impact temperature over the historical period to the point where discharge is affected. Our results confirm the importance of bias correction; when all catchments, GCM–RCMs, and discharge metrics were considered, QM improved discharge simulations in the vast majority of all cases. Additionally, we present a ranking of climate models according to their hydrological performance. Ranking GCM–RCMs is most meaningful prior to bias correction since QM reduces differences between GCM–RCM-driven hydrological simulations. Overall, this work introduces a multivariate assessment method of GCM–RCMs, which enables a more process-oriented evaluation of their simulations.

Simulating Eastern- and Central-Pacific Type ENSO Using a Simple Coupled Model

Severe biases exist in state-of-the-art general circulation models (GCMs) in capturing realistic central-Pacific (CP) El Nino structures. At the same time, many observational analyses have emphasized that thermocline (TH) feedback and zonal advective (ZA) feedback play dominant roles in the development of eastern-Pacific (EP) and CP El Nino–Southern Oscillation (ENSO), respectively. In this work, a simple linear air–sea coupled model, which can accurately depict the strength distribution of the TH and ZA feedbacks in the equatorial Pacific, is used to investigate these two types of El Nino. The results indicate that the model can reproduce the main characteristics of CP ENSO if the TH feedback is switched off and the ZA feedback is retained as the only positive feedback, confirming the dominant role played by ZA feedback in the development of CP ENSO. Further experiments indicate that, through a simple nonlinear control approach, many ENSO characteristics, including the existence of both CP and EP El Nino and the asymmetries between El Nino and La Nina, can be successfully captured using the simple linear air–sea coupled model. These analyses indicate that an accurate depiction of the climatological sea surface temperature distribution and the related ZA feedback, which are the subject of severe biases in GCMs, is very important in simulating a realistic CP El Nino.

Brief communication: Drought likelihood for East Africa

Abstract. The East Africa drought in autumn of year 2016 caused malnutrition, illness and death. Close to 16 million people across Somalia, Ethiopia and Kenya needed food, water and medical assistance. Many factors influence drought stress and response. However, inevitably the following question is asked: are elevated greenhouse gas concentrations altering extreme rainfall deficit frequency? We investigate this with general circulation models (GCMs). After GCM bias correction to match the climatological mean of the CHIRPS data-based rainfall product, climate models project small decreases in probability of drought with the same (or worse) severity as 2016 ASO (August to October) East African event. This is by the end of the 21st century compared to the probabilities for present day. However, when further adjusting the climatological variability of GCMs to also match CHIRPS data, by additionally bias-correcting for variance, then the probability of drought occurrence will increase slightly over the same period.

Assessing the extent of non-stationary biases in GCMs

General circulation models (GCMs) are the main tools for estimating changes in the climate for the future. The imperfect representation of climate models introduces biases in the simulations that need to be corrected prior to their use for impact assessments. Bias correction methods generally assume that the bias calculated over the historical period does not change and can be applied to the future. This study investigates this assumption by considering the extent and nature of bias non-stationarity using 20th century precipitation and temperature simulations from six CMIP5 GCMs across Australia. Four statistics (mean, standard deviation, 10th and 90th quantiles) in monthly and seasonal biases are obtained for three different time window lengths (10, 25 and 33years) to examine the properties of bias over time. This approach is repeated for two different phases of the Interdecadal Pacific Oscillation (IPO), which is known to have strong influences on the Australian climate. It is found that bias non-stationarity at decadal timescales is indeed an issue over some of Australia for some GCMs. When considering interdecadal variability there are significant difference in the bias between positive and negative phases of the IPO. Regional analyses confirmed these findings with the largest differences seen on the east coast of Australia, where IPO impacts tend to be the strongest. The nature of the bias non-stationarity found in this study suggests that it will be difficult to modify existing bias correction approaches to account for non-stationary biases. A more practical approach for impact assessments that use bias correction maybe to use a selection of GCMs where the assumption of bias non-stationarity holds.

Ensemble Forecasts: Probabilistic Seasonal Forecasts Based on a Model Ensemble

Ensembles of general circulation model (GCM) integrations yield predictions for meteorological conditions in future months. Such predictions have implicit uncertainty resulting from model structure, parameter uncertainty, and fundamental randomness in the physical system. In this work, we build probabilistic models for long-term forecasts that include the GCM ensemble values as inputs but incorporate statistical correction of GCM biases and different treatments of uncertainty. Specifically, we present, and evaluate against observations, several versions of a probabilistic forecast for gridded air temperature 1 month ahead based on ensemble members of the National Centers for Environmental Prediction (NCEP) Climate Forecast System Version 2 (CFSv2). We compare the forecast performance against a baseline climatology based probabilistic forecast, using average information gain as a skill metric. We find that the error in the CFSv2 output is better represented by the climatological variance than by the distribution of ensemble members because the GCM ensemble sometimes suffers from unrealistically little dispersion. Lack of ensemble spread leads a probabilistic forecast whose variance is based on the ensemble dispersion alone to underperform relative to a baseline probabilistic forecast based only on climatology, even when the ensemble mean is corrected for bias. We also show that a combined regression based model that includes climatology, temperature from recent months, trend, and the GCM ensemble mean yields a probabilistic forecast that outperforms approaches using only past observations or GCM outputs. Improvements in predictive skill from the combined probabilistic forecast vary spatially, with larger gains seen in traditionally hard to predict regions such as the Arctic.

Growth and decay of the equatorial Atlantic SST mode by means of closed heat budget in a coupled general circulation model

Tropical Atlantic variability is strongly biased in coupled General Circulation Models (GCM). Most of the models present a mean Sea Surface Temperature (SST) bias pattern that resembles the leading mode of inter-annual SST variability. Thus, understanding the causes of the main mode of variability of the system is crucial. A GCM control simulation with the IPSL-CM4 model as part of the CMIP3 experiment has been analyzed. Mixed layer heat budget decomposition has revealed the processes involved in the origin and development of the leading inter-annual variability mode which is defined over the Equatorial Atlantic (hereafter EA mode). In comparison with the observations, it is found a reversal in the anomalous SST evolution of the EA mode: from west equator to southeast in the simulation, while in the observations is the opposite. Nevertheless, despite the biases over the eastern equator and the African coast in boreal summer, the seasonality of the inter-annual variability is well reproduced in the model. The triggering of the EA mode is found to be related to vertical entrainment at the equator as well as to upwelling along South African coast. The damping is related to the air-sea heat fluxes and oceanic horizontal terms. As in the observation, this EA mode exerts an impact on the West African and Brazilian rainfall variability. Therefore, the correct simulation of EA amplitude and time evolution is the key for a correct rainfall prediction over tropical Atlantic. In addition to that, identification of processes which are responsible for the tropical Atlantic biases in GCMs is an important element in order to improve the current global prediction systems.

A new dynamical downscaling approach with GCM bias corrections and spectral nudging

AbstractTo improve confidence in regional projections of future climate, a new dynamical downscaling (NDD) approach with both general circulation model (GCM) bias corrections and spectral nudging is developed and assessed over North America. GCM biases are corrected by adjusting GCM climatological means and variances based on reanalysis data before the GCM output is used to drive a regional climate model (RCM). Spectral nudging is also applied to constrain RCM‐based biases. Three sets of RCM experiments are integrated over a 31 year period. In the first set of experiments, the model configurations are identical except that the initial and lateral boundary conditions are derived from either the original GCM output, the bias‐corrected GCM output, or the reanalysis data. The second set of experiments is the same as the first set except spectral nudging is applied. The third set of experiments includes two sensitivity runs with both GCM bias corrections and nudging where the nudging strength is progressively reduced. All RCM simulations are assessed against North American Regional Reanalysis. The results show that NDD significantly improves the downscaled mean climate and climate variability relative to other GCM‐driven RCM downscaling approach in terms of climatological mean air temperature, geopotential height, wind vectors, and surface air temperature variability. In the NDD approach, spectral nudging introduces the effects of GCM bias corrections throughout the RCM domain rather than just limiting them to the initial and lateral boundary conditions, thereby minimizing climate drifts resulting from both the GCM and RCM biases.

Increasing the detectability of external influence on precipitation by correcting feature location in GCMs

AbstractUnderstanding how precipitation varies as the climate changes is essential to determining the true impact of global warming. This is a difficult task not only due to the large internal variability observed in precipitation but also because of a limited historical record and large biases in simulations of precipitation by general circulation models (GCMs). Here we make use of a technique that spatially and seasonally transforms GCM fields to reduce location biases and investigate the potential of this bias correction to study historical changes. We use two versions of this bias correction—one that conserves intensities and another that conserves integrated precipitation over transformed areas. Focussing on multimodel ensemble means, we find that both versions reduce RMS error in the historical trend by approximately 11% relative to the Global Precipitation Climatology Project (GPCP) data set. By regressing GCMs' historical simulations of precipitation onto radiative forcings, we decompose these simulations into anthropogenic and natural time series. We then perform a simple detection and attribution study to investigate the impact of reducing location biases on detectability. A multiple ordinary least squares regression of GPCP onto the anthropogenic and natural time series, with the assumptions made, finds anthropogenic detectability only when spatial corrections are applied. The result is the same regardless of which form of conservation is used and without reducing the dimensionality of the fields beyond taking zonal means. While “detectability” is dependent both on the exact methodology and the confidence required, this nevertheless demonstrates the potential benefits of correcting location biases in GCMs when studying historical precipitation, especially in cases where a signal was previously undetectable.

Reliability of CMIP5 GCM simulations in reproducing atmospheric circulation over Europe and the North Atlantic: a statistical downscaling perspective

ABSTRACTReliability of GCM (general circulation model) simulations is a significant predictor selection criteria and seems to be essential for outcomes of the statistical downscaling. This article investigates the reliability of a wide ensemble of CMIP5 climate models in reproducing seasonal atmospheric circulation patterns over Europe and the North Atlantic, namely sea level pressure (SLP) and 500 hPa geopotential height (G500). GCMs were evaluated with respect to 1971–2000 NCEP/NCAR reanalysis data by means of MAE, correlation coefficient and standard deviation. An attempt was made to determine the range of biases introduced into the statistical models by biases in the GCM simulations. For this purpose, canonical correlation analysis (CCA) models linking large‐scale atmospheric circulation and air temperature in Poland were driven by historical GCM simulations expressed as anomalies with reference to the reanalysis data.It was shown that reliability of GCMs varies considerably, both seasonally and on an inter‐model basis, thus becoming unfavourable in the context of statistical downscaling application. Overall, GCM ability to reproduce G500 values is better than for SLP. The most important finding from the statistical downscaling perspective is that SLP and G500 biases may introduce considerable air temperature biases (T‐GCMbiases) into the downscaled reconstructions or projections of climate change. Seasonally, the highest T‐GCMbiases were found in DJF, when a significant overestimation of air temperature results from a GCM tendency to overestimate the meridional pressure gradient. A similar situation appears to be a significant factor in MAM and SON. Multiyear average GCM biases do not, however, explain a significant part of the inter‐model variability of T‐GCMbias, thus highlighting the need for a more in‐depth evaluation. Nevertheless, good GCM performance seems to ensure that T‐GCMbias is not of a considerable value. Overall, the results of the article emphasise the necessity for undertaking an application‐oriented assessment of GCM reliability prior to any statistical downscaling approach.

Errors in climate model daily precipitation and temperature output: time invariance and implications for bias correction

Abstract. When correcting for biases in general circulation model (GCM) output, for example when statistically downscaling for regional and local impacts studies, a common assumption is that the GCM biases can be characterized by comparing model simulations and observations for a historical period. We demonstrate some complications in this assumption, with GCM biases varying between mean and extreme values and for different sets of historical years. Daily precipitation and maximum and minimum temperature from late 20th century simulations by four GCMs over the United States were compared to gridded observations. Using random years from the historical record we select a "base" set and a 10 yr independent "projected" set. We compare differences in biases between these sets at median and extreme percentiles. On average a base set with as few as 4 randomly-selected years is often adequate to characterize the biases in daily GCM precipitation and temperature, at both median and extreme values; 12 yr provided higher confidence that bias correction would be successful. This suggests that some of the GCM bias is time invariant. When characterizing bias with a set of consecutive years, the set must be long enough to accommodate regional low frequency variability, since the bias also exhibits this variability. Newer climate models included in the Intergovernmental Panel on Climate Change fifth assessment will allow extending this study for a longer observational period and to finer scales.