Biases In Global Climate Models Research Articles

Projected near-surface wind speed and wind energy over Central Asia using dynamical downscaling with bias-corrected global climate models

Wind energy development in Central Asia can help alleviate drought and fragile ecosystems. Nevertheless, current studies mainly used the global climate models (GCMs) to project wind speed and energy. The simulated biases in GCMs remain prominent, which induce a large uncertainty in the projected results. To reduce the uncertainties of projected near-surface wind speed (NSW) and better serve the wind energy development in Central Asia, the Weather Research and Forecasting (WRF) model with bias-corrected GCMs was employed. Compared with the outputs of GCMs, dynamical downscaling acquired using the WRF model can better capture the high- and low-value centres of NSWS, especially those of Central Asia's mountains. Meanwhile, the simulated NSWS bias was also reduced. For future changes in wind speed and wind energy, under the Representative Concentration Pathway 4.5 (RCP4.5) scenario, NSWS during 2031–2050 is projected to decrease compared with that in 1986–2005. The magnitude of NSWS reduction during 2031–2050 will reach 0.1 m s−1, and the maximum reduction is projected to occur over the central and western regions (>0.2 m s−1). Furthermore, future wind power density (WPD) can reveal nonstationarity and strong volatility, although a downward trend is expected during 2031–2050. In addition, the higher frequency of wind speeds at the turbine hub height exceeding 3.0 m s−1 can render the plain regions more suitable for wind energy development than the mountains from 2031 to 2050. This study can serve as a guide in gaining insights into future changes in wind energy across Central Asia and provide a scientific basis for decision makers in the formulation of policies for addressing climate change.

Is Bias Correction in Dynamical Downscaling Defensible?

AbstractLocalized projections of 21st‐century hydroclimate variables obtained from downscaling Global Climate Model (GCM) output are central to informing regional impact assessments and infrastructure planning. Regional GCM biases can be significant and, for dynamical downscaling, can be addressed either before (a priori) or after (a posteriori) downscaling. However, a priori bias correction (APBC) has generally unexplored effects on climate change signals. Here we analyze dynamically downscaled solutions of CMIP6 GCMs over the Western U.S., with and without APBC, and quantify APBC's impact on climate change signals relative to other irreducible uncertainty sources. For temperature and precipitation, the uncertainty introduced by APBC is negligible compared to that arising from GCM choice or internal variability. Furthermore, APBC greatly reduces regional models' unrealistically high snow‐water‐equivalent (SWE) biases that result directly from GCM errors. We leverage this finding to encourage the dynamical downscaling community to adopt APBC as a standard operating procedure.

Understanding the Cascade: Removing GCM Biases Improves Dynamically Downscaled Climate Projections

AbstractPolarization surrounding bias correction (BC) in creating climate projections arises from its lack of physicality. Here, we perform and analyze 18 dynamical downscaling simulations (with and without BC) to better understand the physical impacts of BC, applied before downscaling, on regional climate output across the western United States. Without BC, downscaled precipitation is systematically and unrealistically wet biased compared to a hierarchy of observationally based datasets over the 1980–2014 period due to cascading mean‐state Global Climate Model (GCM) biases: (a) overly strong lower‐tropospheric lapse rates (5 K/km), (b) overly cold (2 K) tropospheric temperatures, and (c) anomalous mid‐tropospheric cyclonic vorticity advection. With BC, downscaled precipitation (snow) biases are virtually eliminated (halved). Identified GCM biases are common to the broader Coupled Model Intercomparison Project ensemble. Physical effects of BC on the quality of the regionalized projections, pending an evaluation of BC's distortion of the downscaled climate response, may motivate its broader application by dynamical downscalers.

Diagnosing the Radiation Biases in Global Climate Models Using Radiative Kernels

AbstractRadiation energy balance at the top of the atmosphere (TOA) is a critical boundary condition for the Earth climate. It is essential to validate it in the global climate models (GCM) on both global and regional scales. However, the comparison of overall radiation field is known to conceal compensating errors. Here we use a new set of radiative kernels to diagnose the radiation biases caused by different geophysical variables in the GCMs of the Coupled Model Intercomparison Project. We find although clouds remain a primary cause of radiation biases, the radiation biases caused by non‐cloud variables are of comparable magnitudes. Many GCMs have a cold air temperature bias and a moist tropospheric humidity bias, which lead to considerable biases in TOA radiation budget but are compensated by cloud biases. These findings signify the importance of validating the GCM simulations in terms of both the overall and component radiation biases.

Projected Changes in the Tibetan Plateau Snowpack Resulting From Rising Global Temperatures

AbstractThe snowpack is a critical component of the hydrologic cycle in cold regions, the change in which becomes important for proper planning and management of water. The Tibetan Plateau provides significant amount of water to most Asian rivers, and consequently the downstream population is dependent on its availability. Despite its importance, potential change in snowpack in this region due to climate change is poorly understood to date, largely because of remoteness and the orographic complexity of the area. This study inspects the impact of climate change on snowpack change over the Tibetan Plateau considering historical simulations (1981–2004), near future projections (2041–2064), and far future projections (2071–2094) from global climate models (GCMs) and regional climate models (RCMs) of derived temperature, precipitation, and snow water equivalent (SWE). A multivariate nesting bias correction approach (MRNBC) was employed to correct possible biases in GCM and RCM derived temperature, precipitation, and SWE jointly over multiple time scales to preserve interdependencies among the variables while enabling simulation of year‐to‐year persistence, which is of importance in water security assessments. MRNBC reduced the bias in model simulations significantly and improved projections of the snow climatology. The results indicate that the annual maximum spell of snow‐free days will increase over the region whereas the snowy day fraction will decrease in the future compared to the historical period. In addition, annual SWE are noted to be decreasing in both the near future and far future with respect to historical averages. Changes in SWE will result from warming temperatures and also from changes in precipitation, which will lead to more rainfall than snowfall thus affecting snowmelt processes.

Future changes in precipitation and temperature over the Yangtze River Basin in China based on CMIP6 GCMs

This paper explored the projected changes in precipitation, maximum temperature (Tmax), and minimum temperature (Tmin) over the Yangtze River Basin (YRB) based on 23 Global climate models (GCMs) from the Coupled Model Intercomparison Project phase 6 (CMIP6). The Empirical Quantile Mapping (EQM) approach was firstly applied to correct the biases in the GCMs. Next, the future changes were investigated by analyzing the multi-model ensemble (MME) of the bias-corrected dataset during 2025–2044 (near-term), 2045–2064 (mid-term), and 2081–2100 (long-term) periods, with reference to the baseline period 1995–2014, under three integrated scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) of the Shared Socioeconomic Pathways (SSPs) and the Representative Concentration Pathways (RCPs). Results show that: (1) the biases in CMIP6 GCMs can be effectively corrected by EQM, and the MME performs better than individual models for each climatic variable. (2) Precipitation over the YRB from 2025 to 2100 is projected to increase at the rate of 9.66 mm/decade, 13.45 mm/decade, and 21.01 mm/decade under SSP1‐2.6, SSP2‐4.5, and SSP5‐8.5, respectively. In the long-term, the annual precipitation is projected to increase by 10.41%, 10.66% and 15.80%, under SSP1‐2.6, SSP2‐4.5 and SSP5‐8.5, respectively. (3) Tmax (Tmin) over the YRB is projected to increase by 0.09 (0.07) °C/decade, 0.29 (0.27) °C/decade, and 0.66 (0.64) °C/decade under SSP1‐2.6, SSP2‐4.5, and SSP5‐8.5, respectively. In the long-term, Tmax (Tmin) averaged over the YRB is projected to increase by 1.75 (1.50) °C, 2.72 (2.54) °C and 5.04 (4.85) °C under SSP1‐2.6, SSP2‐4.5 and SSP5‐8.5, respectively. (4) Uncertainties in the projected precipitation and temperature over the YRB were reduced based on MME. But further researches, such as selecting the superior models with respect to the regional climate of the YRB from the CMIP6 and using the ensemble methods that assign weight based on the performance of each model, are still needed to provide more reliable climate projections.

Global marine heatwave events using the new CMIP6 multi-model ensemble: from shortcomings in present climate to future projections

In recent years, research related to the occurrence of marine heatwave (MHW) events worldwide has been increasing, reporting severe impacts on marine ecosystems which led to losses of marine biodiversity or changes in world fisheries. Many of these studies, based on regional and global coupled models, show relevant biases in the MHW properties when compared with observations. In this study, the MHW frequency of occurrence, the duration and mean intensity over the global oceans are characterized, taking advantage of the new global climate model (GCM) dataset, from the Coupled Model Project Intercomparison Phase 6 (CMIP6). The MHWs result for the historical period are compared with observations, and the future projected changes are characterized under three socioeconomic pathways (SSPs) (SSP1, SSP2 and SSP5), for the middle and end of century (2041–2070 and 2071–2100). The results show a reasonable agreement between the modeled and observed MHW property trends, indicating increases in the frequency, duration and intensity of MHWs along the historical period. For the period 1982–2014, both the ∼2 mean observed events per year and the mean intensity of 0.35 °C above the threshold are underestimated by the multi-model ensemble (MME) mean by 21% and 31%, respectively, while the observed duration of ∼12 d are overestimated by 100%. The future MHWs are expected to increase in duration and intensity, where a near permanent MHW occurs with reference to the historical climate conditions, mainly by the end of the 21st century. The future MHWs intensity, projected by the MME mean, increases in the range of 0.2 °C to 1.5 °C, from the least to the most severe pathways. The GCMs biases obtained with CMIP6 revealed to be in line with the CMIP5 biases, reinforcing the need to use high spatial resolution models to characterize MHW.

Statistically downscaled precipitation sensitivity to gridded observation data and downscaling technique

AbstractFuture climate projections illuminate our understanding of the climate system and generate data products often used in climate impact assessments. Statistical downscaling (SD) is commonly used to address biases in global climate models (GCM) and to translate large‐scale projected changes to the higher spatial resolutions desired for regional and local scale studies. However, downscaled climate projections are sensitive to method configuration and input data source choices made during the downscaling process that can affect a projection's ultimate suitability for particular impact assessments. Quantifying how changes in inputs or parameters affect SD‐generated projections of precipitation is critical for improving these datasets and their use by impacts researchers. Through analysis of a systematically designed set of 18 statistically downscaled future daily precipitation projections for the south‐central United States, this study aims to improve the guidance available to impacts researchers. Two statistical processing techniques are examined: a ratio delta downscaling technique and an equi‐ratio quantile mapping method. The projections are generated using as input results from three GCMs forced with representative concentration pathway (RCP) 8.5 and three gridded observation‐based data products. Sensitivity analyses identify differences in the values of precipitation variables among the projections and the underlying reasons for the differences.Results indicate that differences in how observational station data are converted to gridded daily observational products can markedly affect statistically downscaled future projections of wet‐day frequency, intensity of precipitation extremes, and the length of multi‐day wet and dry periods. The choice of downscaling technique also can affect the climate change signal for variables of interest, in some cases causing change signals to reverse sign. Hence, this study provides illustrations and explanations for some downscaled precipitation projection differences that users may encounter, as well as evidence of symptoms that can affect user decisions.

Evaluation of sea salt aerosols in climate systems: global climate modeling and observation-based analyses**The data that support the findings of this study are available from the corresponding author upon reasonable request.

Sea salt aerosols (SSA), one of the most abundant aerosol species over the global oceans, play important roles for Earth’s climate. State-of-the-art SSA parameterizations in global climate models (GCMs) are typically modeled using near-surface wind speed, sea surface temperature (SST), and precipitation. However, these have non-trivial biases in CMIP3 and CMIP5 GCMs over the tropical Pacific Ocean that can contribute to biases in the simulated SSA. This study investigates the impacts of falling ice radiative effects on the biases of the aforementioned modeled parameters and the resulting modeled SSA biases. We compare the CMIP5 modeled SSA against satellite observations from MISR and MODIS using a pair of sensitivity experiments with falling ice radiative effects on and off in the CESM1-CAM5 model. The results show that when falling ice radiative effects are not taken into account, models have weaker surface wind speeds, warmer SSTs, excessive precipitation, and diluted sea surface salinity (SSS) over the Pacific trade-wind regions, leading to underestimated SSA. In the tropical Pacific Ocean, the inclusion of falling ice radiative effects leads to improvements in the modeled near-surface wind speeds, SSTs, and precipitation through cloud-precipitation-radiation-circulation coupling, which results in more representative patterns of SSA and reduces the SSA biases by ∼10% to 15% relative to the satellite observations. Models including falling ice radiative effects in CMIP5 produce smaller biases in SSA than those without falling ice radiative effects. We suggest that one of the causes of these biases is likely the failure to account for falling ice radiative effects, and these biases in turn affect the direct and indirect effects of SSA in the GCMs.

Evaluation of CORDEX-RCMS and their driving GCMs of CMIP5 in simulation of Indian summer monsoon rainfall and its future projections

Climate models are widely used for global and regional assessment of climate change. The present study aims to assess the ability of regional climate models (RCMs) of the Coordinated Regional Climate Downscaling Experiment (CORDEX) and their driving global climate models (GCMs) of Coupled Models Intercomparison Project phase 5 (CMIP5) in simulating the Indian summer monsoon rainfall (ISMR) over India (1979–2005). The assessment of CORDEX-RCMs driven by the boundary conditions from GCMs is necessary to know the capability of RCMs in the simulation of ISMR over India. The spatiotemporal analysis of ISMR is performed for past periods to understand its characteristics while attempt is made for future projected changes over the homogeneous rainfall region of India. The study reveals that RCM REMO2009 (MPI) and its driving GCM MPI-ESM-LR are close to the observed rainfall of Global Precipitation Climatology Centre (GPCC). Further, it is noticed that the biases in REMO2009 (MPI) and GCMs viz. MPI-ESM-LR and GFDL-CM3 are of comparable amplitude making them suitable for future projection of ISMR for 2016–2045 under Representative Concentration Pathways (RCPs) 4.5 and 8.5 at 99% and 95% confidence levels. In the simulation of REMO2009 (MPI) and its driving GCM (MPI-ESM-LR) under RCPs 4.5 and 8.5, an excess of rainfall is possible over the parts of Peninsular India (PI) and West Central India (WCI) while a deficit over the North West India (NWI). The simulation of the GCM, GFDL-CM3 depicts an excess of rainfall over NWI, PI, and deficit over WCI under both emission scenarios.

A Multimodel Study on Warm Precipitation Biases in Global Models Compared to Satellite Observations

AbstractThe cloud‐to‐precipitation transition process in warm clouds simulated by state‐of‐the‐art global climate models (GCMs), including both traditional climate models and a high‐resolution model, is evaluated against A‐Train satellite observations. The models and satellite observations are compared in the form of the statistics obtained from combined analysis of multiple‐satellite observables that probe signatures of the cloud‐to‐precipitation transition process. One common problem identified among these models is the too‐frequent occurrence of warm precipitation. The precipitation is found to form when the cloud particle size and the liquid water path (LWP) are both much smaller than those in observations. The too‐efficient formation of precipitation is found to be compensated for by errors of cloud microphysical properties, such as underestimated cloud particle size and LWP, to an extent that varies among the models. However, this does not completely cancel the precipitation formation bias. Robust errors are also found in the evolution of cloud microphysical properties from nonprecipitating to drizzling and then to raining clouds in some GCMs, implying unrealistic interaction between precipitation and cloud water. Nevertheless, auspicious information is found for future improvement of warm precipitation representations: the adoption of more realistic autoconversion scheme in the high‐resolution model improves the triggering of precipitation, and the introduction of a sophisticated subgrid variability scheme in a traditional model improves the simulated precipitation frequency over subtropical eastern ocean. However, deterioration in other warm precipitation characteristics is also found accompanying these improvements, implying the multisource nature of warm precipitation biases in GCMs.

The effect of GCM biases on global runoff simulations of a land surface model

Abstract. Global climate model (GCM) outputs feature systematic biases that render them unsuitable for direct use by impact models, especially for hydrological studies. To deal with this issue, many bias correction techniques have been developed to adjust the modelled variables against observations, focusing mainly on precipitation and temperature. However, most state-of-the-art hydrological models require more forcing variables, in addition to precipitation and temperature, such as radiation, humidity, air pressure, and wind speed. The biases in these additional variables can hinder hydrological simulations, but the effect of the bias of each variable is unexplored. Here we examine the effect of GCM biases on historical runoff simulations for each forcing variable individually, using the JULES land surface model set up at the global scale. Based on the quantified effect, we assess which variables should be included in bias correction procedures. To this end, a partial correction bias assessment experiment is conducted, to test the effect of the biases of six climate variables from a set of three GCMs. The effect of the bias of each climate variable individually is quantified by comparing the changes in simulated runoff that correspond to the bias of each tested variable. A methodology for the classification of the effect of biases in four effect categories (ECs), based on the magnitude and sensitivity of runoff changes, is developed and applied. Our results show that, while globally the largest changes in modelled runoff are caused by precipitation and temperature biases, there are regions where runoff is substantially affected by and/or more sensitive to radiation and humidity. Global maps of bias ECs reveal the regions mostly affected by the bias of each variable. Based on our findings, for global-scale applications, bias correction of radiation and humidity, in addition to that of precipitation and temperature, is advised. Finer spatial-scale information is also provided, to suggest bias correction of variables beyond precipitation and temperature for regional studies.

Characterizing and Understanding Cloud Ice and Radiation Budget Biases in Global Climate Models and Reanalysis

Abstract The authors present an observationally based evaluation of the vertically resolved cloud ice water content (CIWC) and vertically integrated cloud ice water path (CIWP) as well as radiative shortwave flux downward at the surface (RSDS), reflected shortwave (RSUT), and radiative longwave flux upward at top of atmosphere (RLUT) of present-day global climate models (GCMs), notably twentieth-century simulations from the fifth phase of the Coupled Model Intercomparison Project (CMIP5), and compare these results to those of the third phase of the Coupled Model Intercomparison Project (CMIP3) and two recent reanalyses. Three different CloudSat and/or Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) combined ice water products and two methods are used to remove the contribution from the convective core ice mass and/or precipitating cloud hydrometeors with variable sizes and falling speeds so that a robust observational estimate can be obtained for model evaluations. The results show that, for annual mean CIWC and CIWP, there are factors of 2–10 (either over- or underestimate) in the differences between observations and models for a majority of the GCMs and for a number of regions. Most of the GCMs in CMIP3 and CMIP5 significantly underestimate the total ice water mass because models only consider suspended cloud mass, ignoring falling and convective core cloud mass. For the annual means of RSDS, RLUT, and RSUT, a majority of the models have significant regional biases ranging from −30 to 30 W m−2. Based on these biases in the annual means, there is virtually no progress in the simulation fidelity of RSDS, RLUT, and RSUT fluxes from CMIP3 to CMIP5, even though there is about a 50% bias reduction improvement of global annual mean CIWP from CMIP3 to CMIP5. It is concluded that at least a part of these persistent biases stem from the common GCM practice of ignoring the effects of precipitating and/or convective core ice and liquid in their radiation calculations.

Southern Annular Mode Dynamics in Observations and Models. Part I: The Influence of Climatological Zonal Wind Biases in a Comprehensive GCM

Abstract A common bias among global climate models (GCMs) is that they exhibit tropospheric southern annular mode (SAM) variability that is much too persistent in the Southern Hemisphere (SH) summertime. This is of concern for the ability to accurately predict future SH circulation changes, so it is important that it be understood and alleviated. In this two-part study, specifically targeted experiments with the Canadian Middle Atmosphere Model (CMAM) are used to improve understanding of the enhanced summertime SAM persistence. Given the ubiquity of this bias among comprehensive GCMs, it is likely that the results will be relevant for other climate models. Here, in Part I, the influence of climatological circulation biases on SAM variability is assessed, with a particular focus on two common biases that could enhance summertime SAM persistence: the too-late breakdown of the Antarctic stratospheric vortex and the equatorward bias in the SH tropospheric midlatitude jet. Four simulations are used to investigate the role of each of these biases in CMAM. Nudging and bias correcting procedures are used to systematically remove zonal-mean stratospheric variability and/or remove climatological zonal wind biases. The SAM time-scale bias is not alleviated by improving either the timing of the stratospheric vortex breakdown or the climatological jet structure. Even in the absence of stratospheric variability and with an improved climatological circulation, the model time scales are biased long. This points toward a bias in internal tropospheric dynamics that is not caused by the tropospheric jet structure bias. The underlying cause of this is examined in more detail in Part II of this study.

Technical Note: Bias correcting climate model simulated daily temperature extremes with quantile mapping

Abstract. When applying a quantile mapping-based bias correction to daily temperature extremes simulated by a global climate model (GCM), the transformed values of maximum and minimum temperatures are changed, and the diurnal temperature range (DTR) can become physically unrealistic. While causes are not thoroughly explored, there is a strong relationship between GCM biases in snow albedo feedback during snowmelt and bias correction resulting in unrealistic DTR values. We propose a technique to bias correct DTR, based on comparing observations and GCM historic simulations, and combine that with either bias correcting daily maximum temperatures and calculating daily minimum temperatures or vice versa. By basing the bias correction on a base period of 1961–1980 and validating it during a test period of 1981–1999, we show that bias correcting DTR and maximum daily temperature can produce more accurate estimations of daily temperature extremes while avoiding the pathological cases of unrealistic DTR values.

Heat balance and eddies in the Peru-Chile current system

The Peru-Chile current System (PCS) is a region of persistent biases in global climate models. It has strong coastal upwelling, alongshore boundary currents, and mesoscale eddies. These oceanic phenomena provide essential heat transport to maintain a cool oceanic surface underneath the prevalent atmospheric stratus cloud deck, through a combination of mean circulation and eddy flux. We demonstrate these behaviors in a regional, quasi-equilibrium oceanic model that adequately resolves the mesoscale eddies with climatological forcing. The key result is that the atmospheric heating is large (>50 W m−2) over a substantial strip >500 km wide off the coast of Peru, and the balancing lateral oceanic flux is much larger than provided by the offshore Ekman flux alone. The atmospheric heating is weaker and the coastally influenced strip is narrower off Chile, but again the Ekman flux is not sufficient for heat balance. The eddy contribution to the oceanic flux is substantial. Analysis of eddy properties shows strong surface temperature fronts and associated large vorticity, especially off Peru. Cyclonic eddies moderately dominate the surface layer, and anticyclonic eddies, originating from the nearshore poleward Peru-Chile Undercurrent (PCUC), dominate the subsurface, especially off Chile. The sensitivity of the PCS heat balance to equatorial intra-seasonal oscillations is found to be small. We demonstrate that forcing the regional model with a representative, coarse-resolution global reanalysis wind product has dramatic and deleterious consequences for the oceanic circulation and climate heat balance, the eddy heat flux in particular.

Temperature and humidity biases in global climate models and their impact on climate feedbacks

A comparison of AIRS and reanalysis temperature and humidity profiles to those simulated from climate models reveals large biases. The model simulated temperatures are systematically colder by 1–4 K throughout the troposphere. On average, current models also simulate a large moist bias in the free troposphere (more than 100%) but a dry bias in the boundary layer (up to 25%). While the overall pattern of biases is fairly common from model to model, the magnitude of these biases is not. In particular, the free tropospheric cold and moist bias varies significantly from one model to the next. In contrast, the response of water vapor and tropospheric temperature to a surface warming is shown to be remarkably consistent across models and uncorrelated to the bias in the mean state. We further show that these biases, while significant, have little direct impact on the models' simulation of water vapor and lapse‐rate feedbacks.

Air flow influences on local climate: comparison of a regional climate model with observations over the United Kingdom

The relationships between synoptic-scale air flow variability and surface temperature and precipitation simulated by the Hadley Centre's HadRM2 regional climate model (RCM; horizon- tal resolution 50 km) were compared with observed relationships for the UK. The aims of the work were to test how well HadRM2 replicated observed relationships between atmospheric circulation and surface climate at a daily scale and to compare the regional model's performance with the global climate model (GCM) within which the HadRM2 is nested (the HadCM2 GCM). Synoptic-scale air flow variability over the UK was measured on a daily time scale by 3 indices: geostrophic flow strength, vorticity and direction. The frequency distribution of these 3 UK air flow indices is well sim- ulated by HadRM2 in most seasons, with significant differences most noticeable in winter vorticity (the model has a bias to cyclonic vorticity) and in spring flow direction. Although HadCM2 also pro- duced a reasonable simulation of the major frequency distribution features, HadRM2 represents a significant improvement; in 8 of the 12 distributions the simulations were not significantly different from the observations. The relationship between temperature and air flow indices was very well sim- ulated by HadRM2, representing a marked improvement over HadCM2. However, similar relation- ships for precipitation were slightly less accurate in the RCM than the GCM, the main feature being overestimation of precipitation totals by between 20 and 50% in all seasons in the 2 regions studied. This is possibly due to propagation of GCM biases through the RCM boundary conditions, and may be overcome by running HadRM2 using observed rather than GCM data for the boundary conditions. Overall, the results show that more confidence could be placed in future UK climate change scenar- ios generated using HadRM2 rather than HadCM2 alone.