Short Communication United Kingdom daily precipitation intensity: improved early data, error estimates and an update from 2000 to 2006

Transcription

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 28: (8) Published online 13 February 8 in Wiley InterScience ( Short Communication United Kingdom daily precipitation intensity: improved early data, error estimates and an update from to 6 D. Maraun,* T. J. Osborn and N. P. Gillett Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, UK ABSTRACT: This paper updates the analysis by Osborn et al. of trends in the contribution of heavy events to precipitation in the UK. We spatially extended the previous analysis of 1 rain gauges to a set of 689 rain gauges covering almost the whole UK, and updated the results to November 6. For each station and season, we calculated ten time series of the contribution of ten precipitation amount categories to the total seasonal precipitation. A principal component analysis of post-1961 trends of all categories and stations is consistent with earlier results, namely, widespread shifts towards greater contribution from heavier precipitation categories during winter, and towards light and moderate categories during summer. Regional and UK average time series of the contribution from the category consisting of the heaviest events indicate that the increased winter intensity was sustained during the most recent ten years, but the trend did not continue at the rate reported previously for For summer, the decreasing contribution from the heaviest rainfall category reported for underwent a reversal during the most recent decade, returning towards the reference level of intensity. Confidence intervals for these regional and UK average time series were estimated by a bootstrap approach and indicate that the sparser observations from the first half of the th century are still sufficient to estimate UK average change. These longer records support the existence of a long-term increase in winter precipitation intensity, and similar trends have now also become evident in spring and (to a lesser extent) autumn. The summer rainfall intensity has exhibited changes that are more consistent with inter-decadal variability than any overall trend. Copyright 8 Royal Meteorological Society KEY WORDS observed climate; daily precipitation; precipitation intensity trends; climate change; decadal variability; UK Received 6 July 7; Revised 2 November 7; Accepted 3 November 7 1. Introduction Since the mid-19th century, a considerable increase in global, hemispheric and regional average surface temperatures has been observed (Brohan et al., 6; Trenberth et al., 7) and future warming owing to anthropogenic greenhouse gas emissions is very likely to happen (Meehl et al., 7). It is well understood that higher temperatures affect the hydrological cycle, leading, at the global scale, to higher atmospheric moisture content and increased evapotranspiration (Trenberth, ; Meehl et al., 7). Regionally varying contributions of these thermodynamic factors in relation to dynamical factors, i.e. changes in the advection and convergence of moisture, lead to a geographically complex response of mean precipitation to global warming. Most Atmosphere Ocean General Circulation Models (AOGCMs) predict an increase in annual mean precipitation in the equatorial region and at high to mid-latitudes, but a decrease from the sub-tropics towards the mid-latitudes * Correspondence to: D. Maraun, Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, UK. d.maraun@uea.ac.uk (Meehl et al., 7). Regional studies for Europe (Christensen et al., 7) simulate a decrease in annual mean precipitation for southern Europe, and an increase in northern Europe, with increases in winter over central and northern Europe, and decreases in summer over central and southern Europe. As surface run-off depends strongly on the rainfall intensity and frequency, changes in intense precipitation events rather than mean precipitation will impact more strongly on floods (and also agriculture, via, e.g. soil erosion). A multi-model study for the mid-latitudes predicts an increase in precipitation intensity, partly due to an increased atmospheric content of water vapour and partly due to atmospheric circulation; for northern Europe, the latter advective effects dominate (Emori and Brown, ; Meehl et al., ). Voss et al. (2) found trends in simulated precipitation extremes and Watterson and Dix (3) detected an increase of the gamma scale parameter for monthly precipitation extremes in the IPCC SRES scenario A2. These trends in extreme precipitation are predicted to accelerate in the period 99 relative to the period (Semenov and Bengtsson, 2). Copyright 8 Royal Meteorological Society

2 834 D. MARAUN ET AL. Furthermore, as the atmospheric moisture capacity reacts more strongly to temperature changes than evapotranspiration does, extreme precipitation is likely to increase more strongly than mean precipitation (Trenberth, ). This qualitative argument has been supported by model simulations, showing that trends in extremes are generally stronger than in mean precipitation (Semenov and Bengtsson, 2; Kharin and Zwiers, ). In northern and central Europe, precipitation extremes are projected to increase (Christensen et al., 7). An investigation of the IPCC A2 and B2 scenarios by Christensen and Christensen (3) came to the conclusion that the probability of severe summertime flooding in Europe might increase in spite of drier summers. For the UK, Ekström et al. () project an increase in extreme precipitation events on scales from 1 to days in Scotland, and a decrease in most of England and Wales, though this study does not discriminate between different seasons. These model experiments are complemented by observational studies. Alexander et al. (6) and Frich et al. (2) identified positive trends in indices of precipitation intensity in most regions of the extra-tropics. Station analysis (Klein Tank and Können, 3; Moberg and Jones, ) found trends in winter mean as well as moderately strong rainfall events over parts of Europe. For the UK, Osborn et al. () (O), and Osborn and Hulme (2) (OH2) found coherent patterns of positive trends in the contribution of heavy daily rainfall events for winter and negative trends for summer. The primary purpose of this paper is to present an extension to 6 of the O and OH2 studies, which analysed daily UK rain gauge observations up to 199 and, respectively. In so doing, we also take the opportunity to make two key enhancements to these earlier studies. First, in the new analysis, the number of rain gauge records is greatly expanded to 689 for the main analysis period (1961 6), compared with 146 used by OH2 for , and to 37 for the earliest analysis back to 19, compared with 32 used by OH2 back to 191. Second, the uncertainty in the regional and UK time series due to incomplete geographical coverage has been estimated and provides essential information for comparing recent changes with those diagnosed from the earlier, sparser rain gauge network. The analysis follows the non-parametric method described by O, with each daily rainfall assigned to one of ten categories according to its amount, and subsequent analysis (via trend fitting, principal component (PC) analysis and area averaging) is made of these category time series, with particular attention to the category containing the highest daily totals. This approach focuses on the precipitation intensity distribution, and disregards absolute precipitation values or an explicit analysis of the annual cycle. For a discussion of these issues, please refer to OH2, page 13. Since the purpose here is to report the updated results, we leave further interpretation (e.g. of the association with changing atmospheric circulation or humidity) and inter-comparison with other analysis methods (e.g. regional frequency analysis, Fowler and Kilsby, 3) to a subsequent paper. 2. Data and methods The daily precipitation data used in this analysis comprise an extension, both in time and spatial coverage, of the data used in O. The original source for these data is the UK Met Office, though we obtained them from the copy of the MIDAS (Met Office Integrated Data Archive System) held at the British Atmospheric Data Center (BADC, There were a number of stations for which some observations used by O (extracted from an earlier Met Office database) were not present in the MIDAS database, possibly due to data loss during transfer between databases. For these stations, the data gaps were filled with values from O if the two records were in close agreement over their common periods. The observations have already undergone quality control checks at the Met Office, though we undertook some additional screening for long runs with zero precipitation and for very high precipitation outliers. A subset of 689 stations of the whole MIDAS database for the UK has been selected on the basis of record length and completeness (supplementary information). Taking into account the relatively large spatial de-correlation lengths of around km for the measures investigated in this study (supplementary material), this set provides an almost complete coverage of the UK for the period from 196 to 6, and an almost complete coverage for England from 19 onwards. However, data are sparse for Scotland before 196 and especially before 194, thus the results for the UK average are biased towards the characteristics of England during these times. The number of stations still providing daily measurements and with relatively long records has reduced after about. Although coverage is mostly still good, this is not the case for the far north and west of Scotland (including Shetland, Orkney and Western Isles). These data have been analysed for the whole of the UK and separately for nine regions originally defined by Wigley et al. (1984) and Gregory et al. (1991). For details of the selected stations, quality criteria and coverage, please refer to the supplementary information. For each daily precipitation record, we defined ten categories of different rainfall intensities according to O, and time series representing the contribution of each category to the total rainfall: 1. For each station and for each of the 12 months separately, all daily data in the reference period (selected for consistency with O) were sorted into ascending order. 2. Based on these values, we defined ten categories: Each category makes up % of the total rainfall amount for this month, from category 1 containing

3 UNITED KINGDOM DAILY PRECIPITATION INTENSITY 83 the lowest precipitation events making up % to category containing the highest precipitation events making up %. For instance, the January category 1 then is defined as the precipitation range that contains a sufficient number of the weakest rainfall events to make up % of the total January amount from Category, instead, is the range that contains the far fewer but much heavier events making up the same amount of %. 3. Using these categories, defined on data from the period, we calculated contribution time series for each season for the whole period of data available. These annual time series, one for each season, category and station, measure the percentage of precipitation from events falling into a certain category relative to the total precipitation of the respective season. During the reference period, the average contribution of each category is, by construction, %, but it is free to vary between individual years and outside the reference period. As the categories have been defined for each station and month individually, they allow for an analysis somewhat complementing an investigation based on globally fixed thresholds: in a rather dry area or time of the year, category consists of lower rainfall events compared to those in a wet regime. 3. Results Our focus is the variability and trends of heavy rainfall events, and so we concentrate mostly on variations in category. The supplementary information also shows the trends in mean precipitation, wet-day probability and mean wet-day amount. We calculated contribution time series for each of the 689 stations. However, for the subsequent analyses we selected different subsets based on the amount of missing values for the period and season of interest Inter-annual and decadal variability of the contribution from heavy precipitation events Figure 1 shows UK average time series for the contribution of category precipitation to the total amount of precipitation recorded in each season at a particular station. The UK average (as well as the regional averages shown below) is obtained from a different number of stations for each period (Table I). In the averaging procedure, every station has been weighted with the inverse of the number of stations within its spatial de-correlation length (supplementary material). This avoids a bias towards regions with clusters of closely spaced stations. The shadings depict 9% confidence intervals, which have been estimated based on a non-parametric bootstrap to represent the uncertainty associated with incomplete geographical coverage of observations (see Appendix for further details). It is interesting to note that the time series based on the sparser station sets (blue, green and red) in the period Figure 1. Average UK contribution of category for winter (DJF, upper panel) and summer (JJA, lower panel), calculated using stations with data available during 19 (blue), 19 (green), 194 (red) and (cyan). Table I. Number of stations in each region for each time period. Where a range is given it reflects the different numbers of stations used for different seasons. NS: North Scotland, ES: East Scotland, SS: South Scotland, NWE: Northwest England (and North Wales), NEE: Northeast England, NI: Northern Ireland, SWE: Southwest England (and South Wales), CEE: Central and East England, SEE: Southeast England. Region NS ES SS NWE NEE NI SWE CEE SEE UK agree very well with the time series based on the complete selection (cyan). This indicates that the pre- 196 time series are reliably representing the whole UK despite the sparsity of stations. The reason for this is the high spatial de-correlation length of the contribution measure of around 8 km. Motivated by this observation, we construct ten continuous regional time series for each season, using for each of the time intervals 19 19, , and the maximum number of stations available. The actual number of contributing stations are listed in Table I.

4 836 D. MARAUN ET AL. Figures 2 show the regionally averaged contribution of category events to the total seasonal precipitation for winter (DJF, Figure 2), spring (MAM, Figure 3), summer (JJA, Figure 4) and autumn (SON, Figure ) for the nine UK regions and the whole UK. The decadal filtered series (black solid lines) show considerable interdecadal variability. However, this variability should be interpreted carefully as the unfiltered time series are not distinguishable from white noise (although this might be due to a lack of power in the statistical test). In winter, the time series representing the whole UK (Figure 2, last panel) shows an upward trend from 194 onwards. Note that the high values in 19 and 193 result from a relatively small number of measurements (Table I). The strongest contribution to the overall trend occurred during the period , with a weakening during recent years that might either signify the onset of a downswing, a levelling off, or an intermittent interruption. Long-term positive trends are evident in some individual regions (East Scotland, Northern Ireland (a) 3 NS (b) 3 ES (c) (d) 3 4 SS 3 3 NEE (e) 4 3 NI (f) 3 3 NWE (g) 3 3 CEE (h) 3 3 SEE (i) 3 (j) 3 SWE UK Figure 2. DJF contribution of category for the nine regions plus UK (a j). The black solid line depicts a -year Gaussian filtered series. The regions are shown in Figure 6 and listed in Table I. This figure is available in colour online at

5 UNITED KINGDOM DAILY PRECIPITATION INTENSITY 837 and Northwest England), whereas, others show positive trends only in recent decades (North Scotland, South Scotland, Northeast England and Southwest England). The remaining regions show no trends in the contribution of heavy precipitation. The picture is similar in spring (Figure 3). During the last century, a positive trend can be seen in the UK time series. This trend is reflected in positive trends in North Scotland, Northern Ireland, Northwest and Northeast, as well as Central and East England and Southwest England, whereas, there are no clear trends in the other regions. The UK time series for summer (Figure 4, last panel) shows a distinct maximum in the late 196s, followed by a seesaw-like behaviour with an initially downward trend (reported by O) that might have been reversing during the last decade. The distinct maximum followed by a decline occurs in all English/Welsh regions except Northeast England, which seems to feature the same inter-annual behaviour, but an opposite trend since 196. Scotland shows no trends and Northern Ireland exhibits a high maximum in the late 193s, followed by a continuous downward trend. (a) 3 NS (b) 4 3 ES (c) (d) 3 SS 4 NEE 3 (e) 4 3 NI (f) 3 3 NWE (g) 3 CEE (h) 4 3 SEE (i) 3 3 SWE (j) UK Figure 3. As Figure 2, but for MAM. This figure is available in colour online at

6 838 D. MARAUN ET AL. (a) NS (b) 4 3 ES (c) (d) 4 3 SS 3 NEE (e) 4 3 NI (f) 4 3 NWE (g) 3 CEE (h) 4 3 SEE (i) 4 3 SWE (j) UK Figure 4. As Figure 2, but for JJA. This figure is available in colour online at During autumn (Figure ), the result is less consistent; with a slightly upward overall trend for the UK as a whole, but upward as well as vanishing or downward trends for individual regions Regional contribution trends for heavy precipitation events In order to highlight the changes in the contribution of the heaviest category during the most densely observed period, we fitted straight lines to the data for each station and averaged the resulting slopes for every region (Figure 6). The trends within each region are approximately Gaussian distributed; the standard deviation of the average is relatively high, making the individual region trends compatible with zero (for 1.96 standard deviations). This indicates high station-to-station variability. However, two remarkable features are the UK-wide positive trends in winter towards more intense rainfall (with very weak values in Northern Ireland and Northwest England) and the negative summer trends in all regions except North Scotland and Northeast England. Trends in the other seasons are not so homogeneous:

7 UNITED KINGDOM DAILY PRECIPITATION INTENSITY 839 (a) 3 NS (b) 4 ES 3 (c) 3 (e) 3 (g) (i) SS NI CEE SWE (d) (f) (h) (j) 4 NEE 3 4 NWE 3 3 SEE UK Figure. As Figure 2, but for SON. This figure is available in colour online at in spring, positive trends towards heavy precipitation events can be observed in Scotland, Northern Ireland and Southwest England, whereas the rest of England exhibits marginal negative trends. In autumn, trends are mainly positive towards heavier rainfall, with the notable exception of Northern Ireland and Northwest England Spatial pattern of contribution trends To get a spatially more detailed view,andalsotopresent the main spatial patterns of trends in all of the ten categories, we updated the PC analysis presented by O. We calculated PCs (based on the correlation matrix) of the matrix spanned by the ten categories (the dimensionality of the system) and the selection of stations (the set of observations), i.e. each matrix element contains the individual trend for the given category and station. For trends over , our results (not shown) are consistent with those of O, despite a five-fold increase in station numbers. The results of the analysis for the updated period are shown in Figure 7 for

8 84 D. MARAUN ET AL. 6N PCA provides more spatial details: Northeast England exhibits trends towards higher precipitation events even in summer, but this region stretches far into the Midlands and East Anglia. Positive trends dominate also along the southwest tips of Scotland, Wales and Cornwall. 8N 6N 4N 2N N NS NI SS ES NWE SWE NEE CEE SEE 8W 6W 4W 2W 2E Figure trends of category contribution for winter, spring, summer and autumn. The results are given as the absolute contribution change in percent over the whole period. This figure is available in colour online at winter (based on 448 stations) and summer (based on 446 stations). Please find the results for spring and autumn in the supplementary information. The right panels show the first PCs of trends for the ten categories. The left panels show the corresponding loadings, i.e. loosely speaking, the agreement of the trends at the individual station with the first PC. Positive loadings are plotted in blue (dark), negative ones in orange (pale). For winter, the first PC explains 33% of the total variance of the spatial contribution trends. It shows a shift from low and medium strength precipitation categories towards high precipitation categories. This pattern is representative of almost all the UK except the area around the Irish Sea and some parts of Cornwall and Devon. These local distinctions are not resolved by the regional averages in Section 3.2. In summer, the general behaviour is opposite: the first PC, explaining 29% of the total variability, shows an even more distinct shift between weak/medium and heavy precipitation events. But in summer, there are negative loadings across wide areas of Britain (with some notable exceptions). The exceptions are already partly captured by the regional analysis (Section 3.2), but the 4. Discussion and conclusions We have analysed a set of 689 daily precipitation records covering the UK. We updated and spatially extended the results from O and OH2 to investigate changes in the precipitation intensity distribution during the last century with a focus on trends during the recent decades. To this end we constructed seasonal intensity categories, which are normalized to the total seasonal precipitation, and thus, remove the influence of changes in the wetday probability, either due to trends or to spatial and seasonal variations. Consequently, we are able to study distributional changes in isolation. To study the essential temporal behaviour of heavy rainfall we calculated regionally averaged time series for the 19 6 period and corresponding confidence intervals. The agreement of regional averages based on a low number of stations with the confidence intervals of the denser network indicates a high reliability of the regional average time series reaching back to 19, which are based on a low number of stations only. Furthermore, to highlight the main spatial behaviour of all precipitation categories, we performed a PC analysis of post-1961 trends of all categories and stations. The agreement of the results for the post-196 period from the dense post-196 set of rain gauges with results based on the sparser sets of the earlier periods suggests a considerable reliability of the extension of OH back in time. Despite the five-fold number of stations, our results are consistent with O and OH2, yet the extension back to 19 and the update to 6 provide additional insight into the decadal variability of heavy precipitation. In fact, the negative summer trends of heavy precipitation contribution reported in the earlier papers are consistent with inter-decadal variability and might have reversed during the last decade. On the other hand, continuous positive trends in winter have been corroborated and, furthermore, become evident in spring. We decided not to assess the significance of the detected trends in this paper. The significance testing of trends is a highly non-trivial issue (e.g. Kallache et al., ): strictly speaking, such a test aims to discriminate continuous forcing from intrinsic variability, and hence, requires a reasonable stochastic model for this variability. Thus, and consistent with the discussion of the regionally averaged time series in Section 3.1, the observed trends might well be swings of multi-decadal variability. To identify which fraction has to be attributed to external factors, an analysis that explicitly models the intrinsic variability and incorporates external factors is necessary. For instance, the positive winter trends might partly be related to stronger frontal precipitation events due to

9 UNITED KINGDOM DAILY PRECIPITATION INTENSITY 841 (a) 6N 8N (b) 3 DJF, PC #1 (33.3%) 6N 4N 2N Contribution Trend N 8W 4W 2E Category (c) 6N 8N (d) JJA, PC #1 (28.6%) 6N 3 4N 2N N Contribution Trend 8W 4W 2E Category Figure 7. Leading PC of the trends in the contribution of ten different categories of precipitation (top: DJF, based on 448 stations, bottom: JJA, based on 446 stations). The bars show the dominant structure of trends and the dots indicate the level of agreement with the actual trend structure at each station; larger dots indicate stronger agreement, orange (pale) dots indicate changes are opposite in sign. This figure is available in colour online at an increased winter cyclone frequency at high latitudes (McCabe et al., 1), which results from a northward shift of storm tracks (Wang et al., 6). The major driver of this shift is the North Atlantic Oscillation (NAO, Hurrell and van Loon, 1997; Gulev et al., 1). Haylock and Goodess (4) already identified a link between the inter-annual variability of extreme winter precipitation in Europe and the NAO. We plan to assess the impacts of forced climate change, large-scale atmospheric circulation and synoptic scale weather types on extreme precipitation in the UK in a subsequent study. Acknowledgements This study was supported by the NERC Flood Risk from Extreme Events programme (NE/E2412/1). We thank

10 842 D. MARAUN ET AL. colleagues at the BADC and the Met Office for assistance in obtaining data. The analysis was done with software written in R ( Appendix: Bootstrap Confidence Intervals The regional averages of the contribution time series are subject to several types of uncertainties. The most important is the incomplete spatial coverage due to a limited set of stations, especially for the pre-196 periods. Since the underlying distribution of contribution values is non-gaussian and asymmetric (no values lower than zero), we estimated confidence intervals using a non-parametric bootstrap based on the dense post-196 network of rain gauges. For every region and for every point in time, we identified the closest analogue to the corresponding contribution value in the post-196 section of the time series. Then we constructed a bootstrap ensemble with resampling from all the contribution values of all stations within the individual region at the time of the analogue. Each value in the bootstrap ensemble is calculated from the same number of stations as used for the original value of interest. From the bootstrap ensemble, we estimated 9% confidence levels. The skewness of the contribution value distribution depends mostly on its mean (=closeness to zero), rather than on other parameters. Thus, we identify analogues by comparing the mean contribution values only. The analogue-approach does not imply stationarity of the precipitation process, but implicitly assumes a stationary station-to-station variability for a given mean value. References Alexander LV, Zhang X, Peterson TC, Caesar J, Gleason B, Klein Tank AMG, Haylock M, Collins D, Trewin B, Rahimzadeh F, Tagipour A, Rupa Kumar K, Revadekar J, Griffiths G, Vincent L, Stephenson DB, Burn J, Aguilar E, Brunet M, Taylor M, New M, Zhai P, Rusticucci M, Vazquez-Aguirre JL. 6. Global observed changes in daily climate extremes of temperature and precipitation. Journal of Geophysical Research 111: D9. Brohan P, Kennedy JJ, Harris I, Tett SFB, Jones PD. 6. Uncertainty estimates in regional and global observed temperature changes: a new dataset from 18. Journal of Geophysical Research 111: D126. Christensen JH, Christensen OB. 3. Severe summer flooding in Europe. Nature 421: 8. Christensen JH, Hewitson B, Busuioc A, Chen A, Gao X, Held I, Jones R, Kolli RK, Kwon W-T, Laprise R, Magaña Rueda V, Mearns L, Menéndez CG, Räisänen J, Rinke A, Sarr A, Whetton P. 7. Regional climate projections. In Climate Change 7: The Physical Science Basis, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon S, Qin D, Manning M, Chen C, Marquis M, Averyt KB, Tignor M, Miller HL (eds). Cambridge University Press: Cambridge, New York. Ekström M, Fowler HJ, Kilsby CG, Jones PD.. New estimates of future changes in extreme rainfall across the UK using regional climate model integrations. 2. Future estimates and use in impact studies. Journal of Hydrology 3: Emori S, Brown SJ.. Dynamic and thermodynamic changes in mean and extreme precipitation under changed climate. Geophysical Research Letters 32: L1776. Fowler HJ, Kilsby CG. 3. A regional frequency analysis of United Kingdom extreme rainfall from International Journal of Climatology 23: Frich P, Alexander LV, Deall-Marta P, Gleason B, Haylock M, Klein Tank AMG, Peterson T. 2. Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate Research 19: Gregory JM, Jones PD, Wigley TML Precipitation in Britain: an analysis of area-averaged data updated to International Journal of Climatology 11: Gulev SK, Zulina O, Grigoriev S. 1. Extratropical cyclone variability in the Northern Hemisphere winter from the NCEP/NCAR reanalysis data. Climate Dynamics 17: Haylock MR, Goodess CM. 4. Interannual variability of extreme European winter rainfall and links with mean large-scale circulation. International Journal of Climatology 24: Hurrell JW, van Loon H Decadal variations in climate associated with the North Atlantic Oscillation. Climatic Change 36: Kallache M, Rust HW, Kropp J.. Trend assessment: applications for hydrology and climate research. Nonlinear Processes in Geophysics 12: 1 2. Kharin VV, Zwiers FW.. Estimating extremes in transient climate change simulations. Journal of Climate 18: Klein Tank AMG, Können GP. 3. Trends in indices of daily temperature and precipitation extremes in Europe, Journal of Climate 16: McCabe GJ, Clark MP, Serreze MC. 1. Trends in Northern Hemisphere surface cyclone frequency and intensity. Journal of Climate 14: Meehl GA, Arblaster JM, Tebaldi C.. Understanding future patterns of precipitation extremes in climate model simulations. Geophysical Research Letters 32: L Meehl GA, Stocker TF, Collins WD, Friedlingstein P, Gaye AT, Gregory JM, Kitoh A, Knutti R, Murphy JM, Noda A, Raper SCB, Watterson IG, Weaver AJ, Zhao Z-C. 7. Global climate projections. In Climate Change 7: The Physical Science Basis, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon S, Qin D, Manning M, Chen C, Marquis M, Averyt KB, Tignor M, Miller HL (eds). Cambridge University Press: Cambridge, New York. Moberg A, Jones PD.. Trends in indices for extremes in daily temperature and precipitation in central and Western Europe, International Journal of Climatology : Osborn TJ, Hulme M. 2. Evidence for trends in heavy rainfall events over the UK. Philosophical Transactions of the Royal Society London A 36: Osborn TJ, Hulme M, Jones PD, Basnett TA.. Observed trends in the daily intensity of United Kingdom precipitation. International Journal of Climatology : Semenov VA, Bengtsson L. 2. Secular trends in daily precipitation characteristics: greenhouse gas simulation with a coupled AOGCM. Climate Dynamics 19: Trenberth KE.. Uncertainty in Hurricanes and global warming. Science 38: Trenberth KE, Jones PD, Ambenje P, Bojariu R, Easterling D, Klein Tank AMG, Parker D, Rahimzadeh F, Renwick JA, Rusticucci M, Soden B, Zhai P. 7. Observations: surface and atmospheric climate change. In Climate Change 7: The Physical Science Basis, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Solomon S, Qin D, Manning M, Chen C, Marquis M, Averyt KB, Tignor M, Miller HL (eds). Cambridge University Press: Cambridge, New York. Voss R, May W, Roeckner E. 2. Enhanced resolution modeling study on anthropogenic climate change: changes in the extremes of the hydrological cycle. International Journal of Climatology 22: Wang XL, Swail VR, Zwiers FW. 6. Climatology and changes of extratropical storm tracks and cyclone activity: comparison of ERA- 4 with NCEP/NCAR Reanalysis for Journal of Climate 19: Watterson IG, Dix MR. 3. Simulated changes due to global warming in daily precipitation means and extremes and their interpretation using the gamma distribution. Journal of Geophysical Research 8: D Wigley T, Lough J, Jones PD Spatial patterns of precipitation in England and Wales and a revised, homogeneous England and Wales precipitation series. Journal of Climatology 4: 1.