Validation of climate model output using Bayesian statistical methods

The growing interest in and emphasis on high spatial resolution estimates of future climate has demonstrated the need to apply regional climate models (RCMs) to that problem. As a consequence, the need for validation of these models, an assessment of how well an RCM reproduces a known climate, has also grown. Validation is often performed by comparing RCM output to gridded climate datasets and/or station data. The primary disadvantage of using gridded climate datasets is that the spatial resolution is almost always different and generally coarser than climate model output. We have used a Bayesian statistical model derived from observational data to validate RCM output. We used surface air temperature (SAT) data from 109 observational stations in California, all with records of approximately 50 years in length, and created a statistical model based on this data. The statistical model takes into account the elevation of the station, distance from coastline, and the NOAA climate region in which the station resides. Analysis indicates that the statistical model provides reliable estimates of the mean monthly SAT at any given station. In our method, the uncertainty in the estimates produced by the statistical model are directly determined by obtaining probability density functions for predicted SATs. This statistical model is then used to estimate average SATs corresponding to each of the climate model grid cells. These estimates are compared to the output of the RCM to assess how well the RCM matches the observed climate as defined by the statistical model. Overall, the match between the RCM output and the statistical model is good, with some deficiencies likely due in part to the representation of topography in the RCM.     

Comparison of two bias correction methods for precipitation simulated with a regional climate model

This study evaluates the performance of two bias correction techniques—power transformation and gamma distribution adjustment—for Eta regional climate model (RCM) precipitation simulations. For the gamma distribution adjustment, the number of dry days is not taken as a fixed parameter; rather, we propose a new methodology for handling dry days. We consider two cases: the first case is defined as having a greater number of simulated dry days than the observed number, and the second case is defined as the opposite. The present climate period was divided into calibration and validation sets. We evaluate the results of the two bias correction techniques using the Kolmogorov-Smirnov nonparametric test and the sum of the differences between the cumulative distribution curves. These tests show that both correction techniques were effective in reducing errors and consequently improving the reliability of the simulations. However, the gamma distribution correction method proved to be more efficient, particularly in reducing the error in the number of dry days.

     

Present climate simulation over Korea with a regional climate model using a one-way double-nested system

Summary This study investigates the capability of the regional climate model RegCM3 to simulate surface air temperature and precipitation over the Korean Peninsula. The model is run in one-way double nested mode, with a 60 km grid point spacing “mother” domain encompassing the eastern regions of Asia and a 20 km grid point spacing nested domain covering the Korean Peninsula. The simulation spans the three-year period of 1 October 2000 through 30 September 2003 and the boundary conditions needed to run the mother domain experiment are provided from the NCEP reanalysis of observations. The model results are compared with a high density station observation dataset to examine the fine scale structure of the surface climate signal. The model shows a good performance in capturing both the sign and magnitude of the seasonal and inter-annual variations of the surface variables both over East Asia as a whole and over the Korean Peninsula in the nested system. Some persistent biases are however present. Surface temperature is systematically underestimated, especially over mountainous regions in the warm season. This feature may be due to the relatively coarse representation of the Korean topography. The simulated precipitation over the mother domain successfully reproduces the broad spatial pattern of observed precipitation over East Asia along with its seasonal evolution. On the other hand, fine scale details from the nested results show a varying level of quality for the different individual years. Because of the better resolved topographic forcing, the increased resolution of the nested model improves the spatial agreement with the fine scale observation fields for temperature and cold season precipitation. For summer monsoon precipitation the simulation of individual monsoon convective events and tropical storms is however more important than the topographic forcing, and therefore the performance of the nested system is more case-dependent.     

Comparing precipitation bias correction methods for high-resolution regional climate simulations using COSMO-CLM

A new parametric bias correction method for precipitation with an extension for extreme values is compared to an empirical and an existing parametric method. The bias corrections are applied to the regional climate model COSMO-CLM (consortium for small-scale modelling – climate limited area modelling) with a resolution of 4.5 km for the time periods 1991–2000 and 2091–2100. In addition to a comparison in a cross-validation framework, a focus is laid on the investigation of extreme value correction and the effect of the bias correction on the climate change signal. According to the statistical methods used in this study, it was found that the empirical method outperforms both parametric alternatives. However, due to the limited length of the available time series, some outliers occurred, and all methods had problems correcting extreme values. The climate change signal is moderately influenced by all three methods, and the power of climate change detection is reduced. The largest effect was found for the number of dry days and the mean daily intensity, which are considerably altered after correction.     

Improved confidence in regional climate model simulations of precipitation evaluated using drought statistics from the ENSEMBLES models

An ensemble of regional climate model simulations from the European framework project ENSEMBLES is compared with observations of low precipitation events across a number of European regions. We characterize precipitation deficits in terms of two drought indices, the Standardized Precipitation Index and the self-calibrated Palmer Drought Severity Index. Models that robustly describe the observations for the period 1961–2000 in given regions are identified and an assessment of the overall performance of the ensemble is provided. The results show that in general, models capture the most severe drought events and that the ensemble mean model also performs well. Some regions that appear to be more problematic to simulate well are also identified. These are relatively small regions and have rather complex topographical features. The analysis suggests that assessment of future drought occurrence based on climate change experiments in general would appear to be robust. But due to the heterogeneous and often fine-scaled structure of drought occurrence, quantitative results should be used with great care, particularly in regions with complex terrain and limited information about past drought occurrence.     

The simulation of medicanes in a high-resolution regional climate model

Medicanes, strong mesoscale cyclones with tropical-like features, develop occasionally over the Mediterranean Sea. Due to the scarcity of observations over sea and the coarse resolution of the long-term reanalysis datasets, it is difficult to study systematically the multidecadal statistics of sub-synoptic medicanes. Our goal is to assess the long-term variability and trends of medicanes, obtaining a long-term climatology through dynamical downscaling of the NCEP/NCAR reanalysis data. In this paper, we examine the robustness of this method and investigate the value added for the study of medicanes. To do so, we performed several climate mode simulations with a high resolution regional atmospheric model (CCLM) for a number of test cases described in the literature. We find that the medicanes are formed in the simulations, with deeper pressures and stronger winds than in the driving global NCEP reanalysis. The tracks are adequately reproduced. We conclude that our methodology is suitable for constructing multi-decadal statistics and scenarios of current and possible future medicane activities.     

Convection-permitting regional climate simulation of warm-season precipitation over Eastern China

Climate Dynamics - Convection-permitting regional climate models have been shown to improve precipitation simulation in many aspects, such as the diurnal cycle, precipitation frequency, intensity...     

CWRF模式对长江流域极端降水气候事件的模拟能力评估

基于1980—2016年长江流域站点观测降水,评估了CWRF区域气候模式对长江流域面雨量和极端降水气候事件的模拟能力.结果表明:CWRF模式能较好地再现1980—2016年长江流域及不同分区降水空间分布及月/季面雨量年际变率,且在冬、春季表现较好,夏、秋季次之.CWRF模式对长江流域面雨量存在系统性高估,对面雨量的模拟...     

Linear methods for the statistical transformation of daily precipitation sums from regional climate models

This paper presents two new methods for the linear transformation of climate model precipitation data. For verification of the developed methods, daily precipitation sums from the REMO climate model were transformed to observed data from the meteorological stations in the Malse River basin in the Czech Republic. Both methods are based on an analysis of the gamma distribution. The comparison between the observed and transformed data indicates their suitable applicability. The precipitation amounts of transformed data comply with an equivalent probability distribution as the observed data, a distribution of dry days in time drifts towards realistic values. In addition, the efficiency of the formerly published method for nonlinear transformation and newly introduced linear methods is compared.     

Multidecadal simulation of coastal fog with a regional climate model

In order to model stratocumulus clouds and coastal fog, we have coupled the University of Washington boundary layer model to the regional climate model, RegCM (RegCM-UW). By comparing fog occurrences observed at various coastal airports in the western United States, we show that RegCM-UW has success at modeling the spatial and temporal (diurnal, seasonal, and interannual) climatology of northern California coastal fog. The quality of the modeled fog estimate depends on whether coast-adjacent ocean or land grid cells are used; for the model runs shown here, the oceanic grid cells seem to be most appropriate. The interannual variability of oceanic northern California summertime fog, from a multi-decadal simulation, has a high and statistically significant correlation with the observed interannual variability (r = 0.72), which indicates that RegCM-UW is capable of investigating the response of fog to long-term climatological forcing. While RegCM-UW has a number of aspects that would benefit from further investigation and development, RegCM-UW is a new tool for investigating the climatology of coastal fog and the physical processes that govern it. We expect that with appropriate physical parameterizations and moderate horizontal resolution, other climate models should be capable of simulating coastal fog. The source code for RegCM-UW is publicly available, under the GNU license, through the International Centre for Theoretical Physics.     

Dynamical downscaling of ERA-40 in complex terrain using the WRF regional climate model

Results from a first-time employment of the WRF regional climate model to climatological simulations in Europe are presented. The ERA-40 reanalysis (resolution 1°) has been downscaled to a horizontal resolution of 30 and 10?km for the period of 1961?C1990. This model setup includes the whole North Atlantic in the 30?km domain and spectral nudging is used to keep the large scales consistent with the driving ERA-40 reanalysis. The model results are compared against an extensive observational network of surface variables in complex terrain in Norway. The comparison shows that the WRF model is able to add significant detail to the representation of precipitation and 2-m temperature of the ERA-40 reanalysis. Especially the geographical distribution, wet day frequency and extreme values of precipitation are highly improved due to the better representation of the orography. Refining the resolution from 30 to 10?km further increases the skill of the model, especially in case of precipitation. Our results indicate that the use of 10-km resolution is advantageous for producing regional future climate projections. Use of a large domain and spectral nudging seems to be useful in reproducing the extreme precipitation events due to the better resolved synoptic scale features over the North Atlantic, and also helps to reduce the large regional temperature biases over Norway. This study presents a high-resolution, high-quality climatological data set useful for reference climate impact studies.     

Future extreme events in European climate: an exploration of regional climate model projections

This paper compares how well satellite versus weather station measurements of climate predict agricultural performance in Brazil, India, and the United States. Although weather stations give accurate measures of ground conditions, they entail sporadic observations that require interpolation where observations are missing. In contrast, satellites have trouble measuring some ground phenomenon such as precipitation but they provide complete spatial coverage of various parameters over a landscape. The satellite temperature measurements slightly outperform the interpolated ground station data but the precipitation ground measurements generally outperform the satellite surface wetness index. In India, the surface wetness index outperforms station precipitation but this may be reflecting irrigation, not climate. The results suggest that satellites provide promising measures of temperature but that ground station data may still be preferred for measuring precipitation in rural settings.     

Comparison of precipitation in the regional climate model BALTIMOS to radar observations

Observational data and simulations of the regional climate system Baltic integrated model system (BALTIMOS) were used to study precipitation in the Baltic Sea and its drainage basin with a special focus on the diurnal cycle. The study includes a general evaluation of BALTIMOS precipitation, showing that BALTIMOS has too many light rain events causing an overestimation of the total annual precipitation amount. The diurnal cycle as well as its spatial distribution was analysed. BALTIMOS captures the broad characteristics: a significant diurnal variability with an afternoon peak above land and weak variability with a nocturnal peak above sea. An algorithm to distinguish between frontal and convective precipitation was applied to examine the diurnal cycle more thoroughly. The local solar time of maximum rain in summer is about 1 to 2 h earlier in BALTIMOS than in radar observations of precipitation.     

Characteristics of sub-daily precipitation extremes in observed data and regional climate model simulations

Theoretical and Applied Climatology - The study compares characteristics of observed sub-daily precipitation extremes in the Czech Republic with those simulated by Hadley Centre Regional Model...     

Resolution effects on regional climate model simulations of seasonal precipitation over Europe

We analyze a set of nine regional climate model simulations for the period 1961–2000 performed at 25 and 50 km horizontal grid spacing over a European domain in order to determine the effects of horizontal resolution on the simulation of precipitation. All of the models represent the seasonal mean spatial patterns and amount of precipitation fairly well. Most models exhibit a tendency to over-predict precipitation, resulting in a domain-average total bias for the ensemble mean of about 20% in winter (DJF) and less than 10% in summer (JJA) at both resolutions, although this bias could be artificially enhanced by the lack of a gauge correction in the observations. A majority of the models show increased precipitation at 25 km relative to 50 km over the oceans and inland seas in DJF, JJA, and ANN (annual average), although the response is strongest during JJA. The ratio of convective precipitation to total precipitation decreases over land for most models at 25 km. In addition, there is an increase in interannual variability in many of the models at 25 km grid spacing. Comparison with gridded observations indicates that a majority of models show improved skill in simulating both the spatial pattern and temporal evolution of precipitation at 25 km compared to 50 km during the summer months, but not in winter or on an annual mean basis. Model skill at higher resolution in simulating the spatial and temporal character of seasonal precipitation is found especially for Great Britain. This geographic dependence of the increased skill suggests that observed data of sufficient density are necessary to capture fine-scale climate signals. As climate models increase their horizontal resolution, it is thus a key priority to produce high quality fine scale observations for model evaluation.     

Convective and stratiform precipitation characteristics in an ensemble of regional climate model simulations

We apply a recently proposed algorithm for disaggregating observed precipitation data into predominantly convective and stratiform, and evaluate biases in characteristics of parameterized convective (subgrid) and stratiform (large-scale) precipitation in an ensemble of 11 RCM simulations for recent climate in Central Europe. All RCMs have a resolution of 25 km and are driven by the ERA-40 reanalysis. We focus on mean annual cycle, proportion of convective precipitation, dependence on altitude, and extremes. The results show that characteristics of total precipitation are often better simulated than are those of convective and stratiform precipitation evaluated separately. While annual cycles of convective and stratiform precipitation are reproduced reasonably well in most RCMs, some of them consistently and substantially overestimate or underestimate the proportion of convective precipitation throughout the year. Intensity of convective precipitation is underestimated in all RCMs. Dependence on altitude is also simulated better for stratiform and total precipitation than for convective precipitation, for which several RCMs produce unrealistic slopes. Extremes are underestimated for convective precipitation while they tend to be slightly overestimated for stratiform precipitation, thus resulting in a relatively good reproduction of extremes in total precipitation amounts. The results suggest that the examined ensemble of RCMs suffers from substantial deficiencies in reproducing precipitation processes and support previous findings that climate models’ errors in precipitation characteristics are mainly related to deficiencies in the representation of convection.     

Multiyear simulation of the African climate using a regional climate model (RegCM3) with the high resolution ERA-interim reanalysis

This study examines the ability of the latest version of the International Centre for Theoretical Physics (ICTP) regional climate model (RegCM3) to reproduce seasonal mean climatologies, annual cycle and interannual variability over the entire African continent and different climate subregions. The new European Center for Medium Range Weather Forecast (ECMWF) ERA-interim reanalysis is used to provide initial and lateral boundary conditions for the RegCM3 simulation. Seasonal mean values of zonal wind profile, temperature, precipitation and associated low level circulations are shown to be realistically simulated, although the regional model still shows some deficiencies. The West Africa monsoon flow is somewhat overestimated and the Africa Easterly Jet (AEJ) core intensity is underestimated. Despite these biases, there is a marked improvement in these simulated model variables compared to previous applications of this model over Africa. The mean annual cycle of precipitation, including single and multiple rainy seasons, is well captured over most African subregions, in some cases even improving the quality of the ERA-interim reanalysis. Similarly, the observed precipitation interannual variability is well reproduced by the regional model over most regions, mostly following, and sometimes improving, the quality of the ERA-interim reanalysis. It is assessed that the performance of this model over the entire African domain is of sufficient quality for application to the study of climate change and climate variability over the African continent.