∗ Correspondence to: Dept. of Meteorology, Harry Pitt Building, University of Reading, Reading, RG6 6AL. E-mail:
The ability of the HiGEM climatemodel to represent high-impact, regional, precipitationevents is investigated in two ways. The first focusses on a case study of extremeregional accumulation of precipitation during the passage of a summer extra-tropical cyclone across southern England on 20 July 2007 that resulted in a national flooding emergency. The climatemodel is compared with a global Numerical Weather Prediction (NWP) model and higher resolution, nested limited area models. While the climatemodel does not simulate the timing and location of the cyclone and associated precipitation as accurately as the NWP simulations, the total accumulated precipitation in all models is similar to the rain gauge estimate across England and Wales. The regional accumulation over the event is insensitive to horizontal resolution for grid spacings ranging from 90km to 4km.
The frequency and magnitude of extreme weather events are likely to increase with global warming. However, it is not clear how these events might affect agricultural crops and whether yield losses resulting from severe droughts or heat stress will increase in the future. The aim of this paper is to analyse changes in the magnitude and spatial patterns of two impact indices for wheat: the probability of heat stress around ﬂowering and the severity of drought stress. To compute these indices, we used a wheat simulation model combined with high-resolution climate scenarios based on the output from the Hadley Centre regionalclimatemodel at 18 sites in England and Wales. Despite higher temperature and lower summer precipitation predicted in the UK for the 2050s, the impact of drought stress on simulated wheat yield is predicted to be smaller than that at present, because wheat will mature earlier in a warmer climate and avoid severe summer drought. However, the probability of heat stress around ﬂowering that might result in considerable yield losses is predicted to increase signiﬁcantly. Breeding strategies for the future climate might need to focus on wheat varieties tolerant to high temperature rather than to drought.
JJA 216 61.1 0.05 −0.33 0.59 1.63
SON 271 71.8 0.20 1.16 0.70 0.34
Table 1: The annual and seasonal average EWP precipitation (mm), the standard deviation SD (mm) as a measure of variability based on the period 1931-2014 (1932-2014 for DJF). In remaining columns show the precipitation trends (mm/year) based on a linear least squares fit for four relevant time windows (1931-2014, 1961-2006, 1961-2014 and the ERA-Interim period 1979-2014).
significant increase in very heavy and extremeprecipitationevents defined by 76.2 mm 59
day -1 and 154.9 mm day -1 respectively. It
always seen as increasing from year to year but rather as a climatological trend. These 61
results are also not spatially or seasonally homogeneous. The period examined was 62
which is a version of the NCEP global spectral model optimized for regional applications. The ability of the regional models to reproduce observed climatology given historical reanalysis as forcing was examined in Miller et al. (2009), who concluded that, while all models have limitations, they do a credible job overall. In total, we examine five dynamically downscaled model projections. Two methods of statistical downscaling are used: 1) bias correction with constructed analogs (BCCA) (Hidalgo et al. 2008; Maurer et al. 2010), which downscales fields by linearly combining the closest analogs in the historical record, and 2) bias correction with spatial disaggregation (BCSD) (Wood et al. 2002, 2004), which generates daily data from monthly GCM output by selecting a historical month and rescaling the daily precipitation to match the monthly value and so does not preserve the original global model sequence of daily precipitation. The his- torical month chosen is conditioned on monthly precip- itation amount, so the number of zero precipitation days can change as precipitation changes, but the precipitation intensity changes in BCSD are less directly connected to the GCM results than in the other methods. Maurer and Hidalgo (2008) compared results of using BCCA and BCSD and concluded that they have comparable skill in producing downscaled monthly temperature and precip- itation. In total, we analyze 4 model projections with BCCA and another 16 with BCSD.
Global warming, and consequently climate change are important topics studied extensively by researchers throughout the world in the recent decades where changes in climatic parameters are investigated. Considering large-scaled output of AOGCMs and low precision in computational cells, uncertainty analysis is one of the principles in hydrological studies. For this reason, the uncertainty due to precision of computational cells and in passing from global scale to regional scale through LARS- WG model and CRU institute, precipitation changes in Mashhad synoptic station located in Ghareghom basin were analyzed. The results showed enough ability of the model to simulate precipitation parameter in the base period. Downscaled output of HadCM3 generated by CRU with high precision shows gradual decreasing of precipitation trend for frequency and sum values. Comparing the downscaled output of the AOGCM with 2.5*3.75 resolution and the output of CRU with 0.5*0.5 resolution, the uncertainty is due to precision of computational cells from global to regional scale: the latter scale is closer to real values.
Most often precipitation is measured as point observations using rain gauges. These point measurements provide us with useful data for hydrological modelling. Depending on the purpose, point measurements can be good data sets for cal- culating precipitation indices for a given area. Mean prop- erties such as the mean annual precipitationcan be esti- mated fairly accurately from long time series of point mea- surements, since this property of precipitation is expected to change slowly in space unless topographical obstacles like mountains interfere. Other indices are less well estimated from point measurements. Extremeprecipitation properties from a single time series are less representative of a given area than the mean annual precipitation. These properties are often calculated from a small number of measurements, nor- mally one or a few per year, which means that they are af- fected by significant sampling error. Additionally, the fre- quency, true mean intensity and spatial distribution of the extremeevents that are recorded are not accurately known. Nonetheless, information on extremeevents for a given area is needed in hydrological modelling. Techniques such as the areal reduction factor (ARF) (Wilson, 1990; Sivapalan and Blöschl, 1998) have been introduced to extrapolate point pre- cipitation properties to catchment scale. The ARF can be calculated as a simple linear function of the area covered (Wilson, 1990), or by using more advanced models based on extensive analysis of observations (Sivapalan and Blöschl, 1998). In both cases the areal average precipitation index will
Abstract. Probability estimates of the future change of ex- treme precipitationevents are usually based on a limited number of available global climatemodel (GCM) or regionalclimatemodel (RCM) simulations. Since floods are related to heavy precipitationevents, this restricts the assessment of flood risks. In this study a relatively simple method has been developed to get a better description of the range of changes in extremeprecipitationevents. Five bias-corrected RCM simulations of the 1961–2100 climate for a single green- house gas emission scenario (A1B SRES) were available for the Rhine basin. To increase the size of this five-member RCM ensemble, 13 additional GCM simulations were anal- ysed. The climate responses of the GCMs are used to modify an observed (1961–1995) precipitation time series with an advanced delta change approach. Changes in the temporal means and variability are taken into account. It is found that the range of future change of extremeprecipitation across the five-member RCM ensemble is similar to results from the 13-member GCM ensemble. For the RCM ensemble, the time series modification procedure also results in a similar climate response compared to the signal deduced from the direct model simulations. The changes from the individual RCM simulations, however, systematically differ from those of the driving GCMs, especially for long return periods.
the basins reveal an underestimation of the larger daily pre- cipitation quantiles as compared with Spain02. Figure 5 also shows a smaller spread for the five‐model ensemble ENS2 than for full ‐ensemble ENS1, with no significant improve- ment of the results, although in some particular basins the ENS2 ensemble gets closer to the observed quantiles. Figure 5 shows how the spread increases quasi‐linearly as quantiles increase, showing the lack of consistency of RCMs in correctly reproducing the extreme values of precipita- tion. However, when considering spatially averaged values (Figure 6) the RCMs better reproduce the observed quantiles, particularly when considering the ENS2 ensemble. In this case, the RCMs show a distribution similar to Spain02 in almost all river basins. Note that the better results of the spatially averaged values reflect a limitation of the RCMs in simulating the intensity of precipitation at the small intrabasin scales. This is not surprising since the Spain02 resolution is slightly higher than that of the RCMs. It not clear whether the RCM precipitation should be interpreted as the precipitation on the center of the grid cell [Gutowski et al., 2007] or an average precipitation for the grid cell [Osborn and Hulme, 1998]. Using the latter interpretation, the Spain02 averages of station point values on smaller cells would lead naturally to more extremeprecipitation values than those provided by the RCMs. In any case, the RCMs cannot be expected to be skillful at their grid point scale [von Storch et al., 1993; Frei et al., 2003]. The spatial average over several grid points smoothes out the errors at the grid point scale leading to better estimates. The smallest basin considered (Baleares) contains 37 Spain02 grid points, that is, around 23 grid points in the native RCM grids.
A growing ﬁeld of research aims to characterise the contribution of anthropogenic emissions to the likelihood of extreme weather and climateevents. These analyses can be sensitive to the shapes of the tails of simulated distributions. If tails are found to be unrealistically short or long, the anthropogenic signal emerges more or less clearly, respectively, from the noise of possible weather. Here we compare the chance of daily land-surface precipitation and near-surface temperature extremes generated by three Atmospheric Global Climate Models typically used for event attribution, with distributions from six re- analysis products. The likelihoods of extremes are compared for area-averages over grid cell and regional sized spatial domains. Results suggest a bias favouring overly strong attribution estimates for hot and cold eventsover many regions of Africa and Australia, and a bias favouring overly weak attribution estimates over regions of North America and Asia. For rainfall, results are more sensitive to geographic location. Although the three models show similar results over many regions, they do disagree over others. Equally, results highlight the discrepancy amongst reanalyses products. This emphasises the importance of using multiple reanalysis and/or observation products, as well as multiple models in event attribution studies.
Global warming is expected to alter the frequency, intensity, and risk of extremeprecipitationevents. How- ever, global climate models in general do not correctly reproduce the frequency and intensity distribution of precipitation, especially at the regional scale. We present an analogue method to detect the occurrence of extremeprecipitationevents without relying on modeled precipitation. Our approach is based on the use of composites to identify the distinct large-scale atmospheric conditions associated with widespread out- breaks of extremeprecipitationevents across local scales. The development of composite maps, exemplified in the south-central United States and the Western United States, is achieved through the joint analysis of 27-yr (1979–2005) CPC gridded station data and NASAs Modern Era Retrospective-analysis for Research and Applications (MERRA). Various circulation features and moisture plumes associated with extreme pre- cipitation events are examined. This analogue method is evaluated against the MERRA reanalysis with a success rate of around 80% in detecting extremeevents within one or two days. When applied to the climate
Similar to the above discussion, the simulation with the ERA40 LBC and the EMAN CPPS captures well the first spell (13-20 November 1996), as displayed in Fig. 8. It is worth mentioning that point-by-point matching of the simulated rain with station observations or gridded data is not yet possible, however the overall evolution of the system is well shown by RegCM3. Here we see intense rain in the Wjeh area on 17 November and in the Jeddah area on 19 November. The system is initiated in eastern Sudan (Fig. 8a), crosses the Red Sea (producing rain in western Saudi Arabia), and decays after 19 November. Overall, this life cycle is similar to the one obtained from NNRP2 with GFC, except that the intensity and the embedded areas are slightly different. It is noted that RegCM3 with EMAN is also used for the simulation of precipitationover eastern Africa and tropical Indian Ocean (Davis et al., 2009). They concluded that EMAN provides the most realistic simulation in reproducing spatial distribution of convective rainfall, compared to GFC and GAS, however it overestimates the total rainfall amount with respect to the observations.
Acknowledgements. This work was funded by the Swiss Fed- eral Office for the Environment (FOEN), the Swiss Federal Office of Energy (SFOE), and the Swiss Federal Nuclear Safety Inspectorate (ENSI) in the framework of the project EXAR: Understanding Extreme Flooding Events Aare-Rhein in Switzer- land as well as by the Swiss National Science Foundation (project 200021_143219 “EXTRA-LARGE”). We acknowledge the World Climate Research Programme’s Working Group on RegionalClimate and the Working Group on Coupled Modelling, former coordinating body of CORDEX and responsible panel for CMIP5. We also thank the climate modelling groups (listed in Table S1 of this paper) for producing and making available their model output. We also acknowledge the Earth System Grid Federation infrastructure, an international effort led by the U.S. De- partment of Energy’s Program for ClimateModel Diagnosis and Intercomparison, the European Network for Earth System Modelling, and other partners in the Global Organisation for Earth System Science Portals (GO-ESSP). Computation facilities for the CCC400 simulations were provided by the Swiss National Su- percomputing Centre (CSCS). The Twentieth Century Reanalysis Project is supported by the U.S. Department of Energy (DOE) Office of Science Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program, the Office of Biological and Environmental Research (BER), and the National Oceanic and Atmospheric Administration Climate Program Office. Edited by: Ricardo Trigo
Abstract Extremeprecipitationeventsover India have resulted in loss of human lives and damaged infrastructures, food crops, and lifelines. The inability of climate models to credibly project precipitation extremes in India has not been helpful to longer-term hazards resilience policy. However, there have been claims that ﬁner-resolution and regionalclimate models may improve projections. The claims are examined as hypotheses by comparing models with observations from 1951 –2005. This paper evaluates the reliability of the latest generation of general circulation models (GCMs), Coupled Model Intercomparison Project Phase 5 (CMIP5), speci ﬁcally a subset of the better performing CMIP5 models (called “BEST-GCM”). The relative value of ﬁner-resolution regionalclimate models (RCMs) is examined by comparing Coordinated RegionalClimate Downscaling Experiment (CORDEX) South Asia RCMs ( “CORDEX-RCMs”) versus the GCMs used by those RCMs to provide boundary conditions, or the host GCMs ( “HOST-GCMs”). Ensemble mean of BEST-GCMs performed better for most of the extremeprecipitation indices than the CORDEX-RCMs or their HOST-GCMs. Weaker performance shown by ensemble mean of CORDEX-RCMs is largely associated with their high intermodel variation. The CORDEX-RCMs occasionally exhibited slightly superior skills compared to BEST-GCMs; on the whole RCMs failed to signi ﬁcantly outperform GCMs. Observed trends in the extremes were not adequately captured by any of the model ensembles, while neither the GCMs nor the RCMs were determined to be adequate to inform hydrologic design.
We analyze the period 1982-99, discarding the years 1979-1981 for model spin up, and retaining years available in both observational and climatemodel data. We are working with extremes, so we adopt a relatively conservative spin-up period to ensure that the models water cycles are adequately spun up. Our region of interest is the upper Mississippi region, defined here as the region bounded by 37°-47°N, 89°-99°W, highlighted in Figure 1. This is the same definition used in previous analyses (Gutowski et al. 2007, 2008, 2010). Our analysis focuses on the winter season (December – February), when synoptic dynamics are more important than in the warmer months, when smaller scale convective events may be more important. This assumption here is that these events will be governed more by the resolved circulation (e.g., Schumacher and Johnson 2005, 2006, Gutowski et al. 2008)
and can be estimated from the disaggregated and raw sim- ulated data. This term accounts for most of the spatial res- olution bias. The difference between the disaggregated and observed values (Y 4km ∗ − X) can also be estimated from the data. However, this latter term contains three sources of er- rors: (i) the CRCM simulation bias (physical bias), (ii) the disaggregation model bias and (iii) the difference in resolu- tion between the 4 km pixel and the observation. The purpose of the validation exercise is to evaluate (i); (ii) and (iii) can be seen as noise. For point (ii), Gagnon (2012) showed, for convective events, that the disaggregation can underestimate the 4 km daily precipitation by up to 50 % at worst. Since the calibration of the disaggregation model was performed using 4 years only, the bias cannot be expressed as a function of p over 40 years. The exact value of point (iii) is unknown, but its order of magnitude can be roughly estimated from the analyses of the disaggregation results at 12, 8 and 4 km (pur- ple, green, and blue pixels of Fig. 1, respectively). The last term (X − T ) is the difference between the observed and the real value. It is caused by measurement errors. Measurement errors are more important for solid than for liquid precipita- tion (Goodison et al., 1998; Yang et al., 1999; Fortin et al., 2008). As only precipitation from May to October are an- alyzed, this term is assumed negligible compared to other terms and is not considered here. It is thereby assumed that the left-hand side of Eq. (6) (Y 45km − T ) is approximately
Abstract. Meltwater from the Greenland Ice Sheet con- tributed 1.7–6.12 mm to global sea level between 1993 and 2010 and is expected to contribute 20–110 mm to future sea level rise by 2100. These estimates were produced by re- gional climate models (RCMs) which are known to be ro- bust at the ice sheet scale but occasionally miss regional- and local-scale climate variability (e.g. Leeson et al., 2017; Med- ley et al., 2013). To date, the fidelity of these models in the context of short-period variability in time (i.e. intra-seasonal) has not been fully assessed, for example their ability to sim- ulate extreme temperature events. We use an event identifi- cation algorithm commonly used in extreme value analysis, together with observations from the Greenland Climate Net- work (GC-Net), to assess the ability of the MAR (Modèle At- mosphérique Régional) RCM to reproduce observed extreme positive-temperature events at 14 sites around Greenland. We find that MAR is able to accurately simulate the frequency and duration of these events but underestimates their mag- nitude by more than half a degree Celsius/kelvin, although this bias is much smaller than that exhibited by coarse-scale Era-Interim reanalysis data. As a result, melt energy in MAR output is underestimated by between 16 and 41 % depend- ing on global forcing applied. Further work is needed to pre- cisely determine the drivers of extreme temperature events, and why the model underperforms in this area, but our find- ings suggest that biases are passed into MAR from bound- ary forcing data. This is important because these forcings are common between RCMs and their range of predictions of past and future ice sheet melting. We propose that examin- ing extremeevents should become a routine part of global and regionalclimatemodel evaluation and that addressing
Chapter 1: Introduction
Climate variability is one of the major drivers of both hydrological processes and human decisions regarding the water cycle. Precipitation, a direct connection between climate and the water cycle, affects the overall hydroclimate and agroclimate (agriculture related climate), particularly relative to drought, flood and streamflow (Groisman et al. 2004). In recent decades, heavy and extremeprecipitationevents tend to occur more frequently over most land areas on the Earth (IPCC 2007). As an important indicator of climate change, however, changes in heavy or extremeprecipitationevents in terms of occurrence frequency and intensity can occur independently or disproportionally from changes in the mean or total precipitation (Gershunov 1998; Groisman et al. 1999; Meehl et al. 2000). For instance, the intensity of extremeprecipitationeventscan increase under a stabilized or even decreased total precipitation (Easterling et al. 2000; Alpert et al. 2002; Groisman et al. 2005). More importantly, the variability of heavy and extremeprecipitationeventscan have greater influence than changes in the total precipitation (Katz and Brown 1992). High and extremeprecipitationeventscan lead to hydrological hazards, i.e., high streamflow and flooding (Kunkel et al. 1999b; Groisman et al. 2001; Kunkel 2003a). Flood and storm events account for an annual average of nearly 50% of natural disasters in the world, posing severe threats to human beings and property (Salvadori et al. 2007). Studying high and extremeprecipitationevents are of importance not only in improving the understanding of the characteristics and changes of climatic and hydrologic extremes, but also for providing better adaptation and mitigation advice to decision makers.
are known to be robust at the ice-sheet scale, but occasionally miss regional and local scale climate variability (e.g. Leeson et al., 2017, Medley et al., 2013). To date, the fidelity of these models in the context of short period variability in time (i.e. intra- seasonal) has not been fully assessed, for example their ability to simulate extreme temperature events. We use an event identification algorithm commonly used in Extreme Value Analysis, together with observations from the Greenland Climate Network (GC-Net), to assess the ability of the MAR RCM to reproduce observed extreme positive temperature events at 14 15
2008). There are a couple of considerations that must be made in order to define a hypothesis for testing.
When mean annual temperature departures are examined for each year from 1949 to 2008 (Figure 7) we see an obvious stepwise shift from mostly negative to mostly positive anomalies around 1976. This shift corresponds to a known shift in the Pacific Decadal Oscillation (PDO) from its negative to its positive phase. The positive phase of the PDO is characterized by increased southerly flow and warm air advection into Alaska during winter, resulting in mostly positive temperature anomalies (Alaska Climate Research Center, 2008). Based on the occurrence of this phase shift in the earlier half of the period, we could expect to find the greatest changes in frequency of extremes in that period. However, looking at annual temperature trends within each sub-period as shown in Figures 8 and 9, we clearly see a greater increase in average annual temperature in the latter half of the observing period (1950 to 2008). Based on this pattern of accelerated warming, we could hypothesize that the changes in frequency of extremeevents would be greater in the latter half of the period of record.