For the assessment of Mediterranean temperature under anthropogenically forced climate conditions canonical correlation models are established for the 1948–98 period between highly resolved ...Mediterranean temperatures and large-scale North-Atlantic–European 1000 hPa-/500 hPa-geopotential height fields. Predictor output from two different global general circulation model runs (ECHAM4/OPYC3 and HadCM3), both forced with B2 scenario assumptions according to the Special Report on Emission Scenarios (SRES), is used to assess Mediterranean temperature changes in the 21st century. The results show a temperature increase for the whole Mediterranean area for all months of the year in the period 2071–2100 compared to 1990–2019. The assessed temperature rise varies depending on region and season, but overall substantial temperature changes of partly more than 4 °C by the end of this century have to be anticipated under enhanced greenhouse warming conditions.
VALUE is an open European collaboration to intercompare downscaling approaches for climate change research, focusing on different validation aspects (marginal, temporal, extremes, spatial, ...process‐based, etc.). Here we describe the participating methods and first results from the first experiment, using “perfect” reanalysis (and reanalysis‐driven regional climate model (RCM)) predictors to assess the intrinsic performance of the methods for downscaling precipitation and temperatures over a set of 86 stations representative of the main climatic regions in Europe. This study constitutes the largest and most comprehensive to date intercomparison of statistical downscaling methods, covering the three common downscaling approaches (perfect prognosis, model output statistics—including bias correction—and weather generators) with a total of over 50 downscaling methods representative of the most common techniques.
Overall, most of the downscaling methods greatly improve (reanalysis or RCM) raw model biases and no approach or technique seems to be superior in general, because there is a large method‐to‐method variability. The main factors most influencing the results are the seasonal calibration of the methods (e.g., using a moving window) and their stochastic nature. The particular predictors used also play an important role in cases where the comparison was possible, both for the validation results and for the strength of the predictor–predictand link, indicating the local variability explained. However, the present study cannot give a conclusive assessment of the skill of the methods to simulate regional future climates, and further experiments will be soon performed in the framework of the EURO‐CORDEX initiative (where VALUE activities have merged and follow on).
Finally, research transparency and reproducibility has been a major concern and substantive steps have been taken. In particular, the necessary data to run the experiments are provided at http://www.value‐cost.eu/data and data and validation results are available from the VALUE validation portal for further investigation: http://www.value‐cost.eu/validationportal.
The largest and most comprehensive to date intercomparison of statistical downscaling methods is presented, with a total of over 50 downscaling methods representative of the most common approaches and techniques. Overall, most of the downscaling methods greatly improve raw model biases and no approach is superior in general, due to the large method‐to‐method variability. The main factors influencing the results are the seasonal calibration of the methods and their stochastic nature, for biases in the mean and variance.
The Mediterranean area is strongly vulnerable to future changes in temperature and precipitation, particularly concerning extreme events, and has been identified as a climate change hot spot. This ...study performs a comprehensive investigation of present-day and future Mediterranean precipitation extremes based on station data, gridded observations and simulations of the regional climate model (REMO) driven by the coupled global general circulation model ECHAM5/MPI-OM. Extreme value estimates from different statistical methods—quantile-based indices, generalized pareto distribution (GPD) based return values and data from a weather generator—are compared and evaluated. Dynamical downscaling reveals improved small-scale topographic structures and more realistic higher rainfall totals and extremes over mountain ranges and in summer. REMO tends to overestimate gridded observational data in winter but is closer to local station information. The dynamical–statistical weather generator provides virtual station rainfall from gridded REMO data that overcomes typical discrepancies between area-averaged model rainfall and local station information, e.g. overestimated numbers of rainy days and underestimated extreme intensities. Concerning future rainfall amount, strong summer and winter drying over the northern and southern Mediterranean, respectively, is confronted with winter wetting over the northern part. In contrast, precipitation extremes tend to increase in even more Mediterranean areas, implying regions with decreasing totals but intensifying extremes, e.g. southern Europe and Turkey in winter and the Balkans in summer. The GPD based return values reveal slightly larger regions of increasing rainfall extremes than quantile-based indices, and the virtual stations from the weather generator show even stronger increases.
Projected changes of extreme precipitation in the Mediterranean area up until the end of the 21st century are analysed by means of statistical downscaling. Generalized linear models are used as ...downscaling technique to assess different percentile‐based indices of extreme precipitation on a fine‐scale spatial resolution. In the region under consideration extreme precipitation is related to anomalies of the large‐scale circulation as well as to convective conditions. To account for this, predictor selection encompasses variables describing the large‐scale circulation (geopotential heights of the 700 hPa and 500 hPa levels, u‐ and v‐wind components of the 850 hPa level) as well as thermo‐dynamic parameters (specific humidity of the 850 hPa and 700 hPa levels, Showalter‐Index, convective inhibition). In the scope of the statistical downscaling approach a specific statistical ensemble technique is applied in order to allow for non‐stationarities in the predictors–predictand relationships. Consequently, the statistical ensembles include a range of possible future evolutions of extreme precipitation. Two different emission scenarios (A1B and B1), multiple runs for each scenario, and output of two different general circulation models (ECHAM5 and HadCM3) are applied to assess extreme precipitation under enhanced greenhouse warming conditions. The results yield mainly decreases over many parts of the Mediterranean area in spring. In summer increases are assessed around the Tyrrhenian Sea, the Ionian Sea, and the Aegean Sea, whereas decreases are projected for most of the western and northern Mediterranean regions. In autumn reductions of heavy rainfall occur over many parts of the western and central areas. In winter distinct increases are widespread in the Mediterranean area. Beyond the assessments using all predictors it is shown in the present contribution that different predictor variables can lead to varying statistical downscaling results. It points to distinct impacts of the change of specific atmospheric conditions on local extreme precipitation.
The objective of this study is to investigate the predictability of monthly climate variables in the Mediterranean area by using statistical models. It is a well-known fact that the future state of ...the atmosphere is sensitive to preceding conditions of the slowly varying ocean component with lead times being sufficiently long for predictive assessments. Sea surface temperatures (SSTs) are therefore regarded as one of the best variables to be used in seasonal climate predictions. In the present study, SST-regimes which have been derived and discussed in detail in Part I of this paper, are used with regard to monthly climate predictions for the Mediterranean area. Thus, cross-correlations with time lags from 0 up to 12 months and ensuing multiple regression analyses between the large-scale SST-regimes and monthly precipitation and temperature for Mediterranean sub-regions have been performed for the period 1950-2003. Statistical hindcast ensembles of Mediterranean precipitation including categorical forecast skill can be identified only for some months in different seasons and for some individual regions of the Mediterranean area. Major predictors are the tropical Atlantic Ocean and the North Atlantic Ocean SST-regimes, but significant relationships can also be found with tropical Pacific and North Pacific SST-regimes. Statistical hindcast ensembles of Mediterranean temperature with some categorical forecast skill can be determined primarily for the Western Mediterranean and the North African regions throughout the year. As for precipitation the major predictors for temperature are located in the tropical Atlantic Ocean and the North Atlantic Ocean, but some connections also exist with the Pacific SST variations.
Statistical downscaling methods (SDMs) are techniques used to downscale and/or bias‐correct climate model results to regional or local scales. The European network VALUE developed a framework to ...evaluate and inter‐compare SDMs. One of VALUE's experiments is the perfect predictor experiment that uses reanalysis predictors to isolate downscaling skill. Most evaluation papers for SDMs employ simple statistical diagnostics and do not follow a process‐based rationale. Thus, in this paper, a process‐based evaluation has been conducted for the more than 40 participating model output statistics (MOS, mostly bias correction) and perfect prognosis (PP) methods, for temperature and precipitation at 86 weather stations across Europe.
The SDMs are analysed following the so‐called “regime‐oriented” technique, focussing on relevant features of the atmospheric circulation at large to local scales. These features comprise the North Atlantic Oscillation, blocking and selected Lamb weather types and at local scales the bora wind and the western Iberian coastal‐low level jet.
The representation of the local weather response to the selected features depends strongly on the method class. As expected, MOS is unable to generate process sensitivity when it is not simulated by the predictors (ERA‐Interim). Moreover, MOS often suffers from an inflation effect when a predictor is used for more than one station. The PP performance is very diverse and depends strongly on the implementation. Although conditioned on predictors that typically describe the large‐scale circulation, PP often fails in capturing the process sensitivity correctly. Stochastic generalized linear models supported by well‐chosen predictors show improved skill to represent the sensitivities.
The VALUE perfect predictor experiment SDMs are examined following the so‐called “regime‐oriented” technique, focused on relevant atmospheric circulation features and processes, from large to local scales. Overall, SDMs show a reasonable performance representing the large spectra of atmospheric phenomena analysed, although the PP methods reveal a more differentiated behaviour than MOS. As expected, MOS methods are unable to generate process sensitivity when it is not simulated by the predictors (ERA‐Interim). Although PP methods are conditioned on predictors typically representing the large‐scale circulation, many PP methods frequently fail in capturing the process sensitivity. In the figure, winter NAO conditioned biases for precipitation for the raw model outputs and the different MOS and PP methods. In (c, d) boxes span the 25–75% range, the whiskers the maximum value within 1.5 times the interquartile range, values outside that range are plotted individually; average results over the different PRUDENCE regions are indicated by a coloured horizontal bar (see the colours in the bottom legend).
In the present study, nonstationarities in predictor–predictand relationships within the framework of statistical downscaling are investigated. In this context, a novel validation approach is ...introduced in which nonstationarities are explicitly taken into account. The method is based on results from running calibration periods. The (non)overlaps of the bootstrap confidence interval of the mean model performance (derived by averaging the performances of all calibration/verification periods) and the bootstrap confidence intervals of the individual model errors are used to identify (non)stationary model performance. The specified procedure is demonstrated for mean daily precipitation in the Mediterranean area using the bias to assess model skill. A combined circulation‐based and transfer function–based approach is employed as a downscaling technique. In this context, large‐scale seasonal atmospheric regimes, synoptic‐scale daily circulation patterns, and their within‐type characteristics, are related to daily station‐based precipitation. Results show that nonstationarities are due to varying predictors–precipitation relationships of specific circulation configurations. In this regard, frequency changes of circulation patterns can damp or increase the effects of nonstationary relationships. Within the scope of assessing future precipitation changes under increased greenhouse warming conditions, the identification and analysis of nonstationarities in the predictors–precipitation relationships leads to a substantiated selection of specific statistical downscaling models for the future assessments. Using RCP4.5 scenario assumptions, strong increases of daily precipitation become apparent over large parts of the western and northern Mediterranean regions in winter. In spring, summer, and autumn, decreases of precipitation until the end of the 21st century clearly dominate over the entire Mediterranean area.
Key Points
Statistical downscaling of future precipitation changes using RCP4.5 scenario
Introduction of a novel validation approach considering nonstationarities
Application of a combined circulation‐based and transfer function– based approach
The determination of specific sea surface temperature (SST) patterns from large-scale gridded SST-fields has widely been done. Often principal component analysis (PCA) is used to condense the ...SST-data to major patterns of variability. In the present study SST-fields for the period 1950-2003 from the area 20°S to 60°N are analysed with respect to SST-regimes being defined as large-scale oceanic patterns with a regular and at least seasonal occurrence. This has been done in context of investigations on seasonal predictability of Mediterranean regional climate with large-scale SST-regimes as intended predictors in statistical model relationships. The SST-regimes are derived by means of a particular technique including multiple applications of s-mode PCA. Altogether 17 stationary regimes can be identified, eight for the Pacific Ocean, five for the Atlantic Ocean, two for the Indian Ocean, and two regimes which show a distinct co-variability within different ocean basins. Some regimes exist, with varying strength and spatial extent, throughout the whole year, whereas other regimes are only characteristic for a particular season. Several regimes show dominant variability modes, like the regimes associated with El Niño, with the Pacific Decadal Oscillation or with the North Atlantic Tripole, whereas other regimes describe little-known patterns of large-scale SST variability. The determined SST-regimes are subsequently used as predictors for monthly precipitation and temperature in the Mediterranean area. This subject is addressed in Part II of this paper.