Analysis of UK precipitation extremes derived from Met Office gridded data

Transcription

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 34: (2014) Published online 6 November 2013 in Wiley Online Library (wileyonlinelibrary.com) DOI: /joc.3850 Analysis of UK precipitation extremes derived from Met Office gridded data I. R. Simpson a andp.d.jones a,b * a Climatic Research Unit, University of East Anglia, Norwich, UK b Center of Excellence for Climate Change Research, Department of Meteorology, King Abdulaziz University, Jeddah, Saudi Arabia ABSTRACT: In Simpson and Jones (2012) we introduced new, homogenized UK national and regional rainfall series derived from 5 km gridded daily and monthly precipitation data. The monthly series were extended back to 1766 for monthly England and Wales (EW) precipitation, to 1873 for monthly values for each of the five EW sub-regions, and to 1931 for daily values in all regions. Using data from those series, this paper provides analysis of how mean precipitation totals and extremes have changed over the respective periods. The results showed statistically significant upward trends in mean and extreme winter, spring and autumn precipitation for some Scottish regions, but trends over England and Wales were mostly insignificant, though England and Wales had a significant increase in winter precipitation over The trend in summer precipitation over has been statistically insignificant, though with a significant long-term downward trend for England and Wales over Prior observations of a trend towards drier summers and wetter winters have been complicated by a recent succession of wet summers and dry winters. Many of the observed changes in seasonal precipitation totals are most likely associated with changes in the North Atlantic Oscillation. KEY WORDS precipitation; area average; England and Wales; Scotland; Northern Ireland; extremes; NAO; gridded data Received 4 September 2012; Revised 10 September 2013; Accepted 27 September Introduction Many studies have identified long-term trends in mean and extreme UK precipitation during the 20th century (e.g. Maraun et al., 2008; Osborn et al., 2000; Jones and Conway, 1997), which are consistent with climate model projections of increased extreme precipitation, wetter winters and drier summers across the UK under enhanced greenhouse conditions. Increases in precipitation extremes are likely to result in increased flooding and drought, with strong implications for insurance claims (e.g. Fowler and Wilby, 2010). Previous work by Wigley et al. (1984), Wigley and Jones (1987), Gregory et al. (1991) and Jones and Conway (1997) used a rainfall series based on series for five separate spatially coherent regions in England and Wales (EW), with a maximum of seven well-spaced stations per region (hereafter the HadUKP series). The regions are defined in Table 1. This series was extended to cover Scotland (S) and Northern Ireland (NI) by Gregory et al. (1991). The original series was developed by Nicholas and Glasspoole (1931) to produce a homogeneous rainfall series for England and Wales (Wigley et al., 1984). The choice of England and Wales by Nicholas and Glasspoole (1931), particularly the extension back to 1725, when there were no rain gauges in Wales until the 19th century has always been somewhat surprising. The publication of a wheat price index by Beveridge for England and Wales * Correspondence to: P. D. Jones, Climatic Research Unit, University of East Anglia, Norwich NR4 7TJ, UK. p.jones@uea.ac.uk a year earlier may have been a factor (Beveridge, 1930; see also Gower, 1955). Alexander and Jones (2001) produced a method of updating the HadUKP series in near real-time and an updated analysis of changes in UK precipitation over the respective periods. This paper uses data extracted from the 5 km gridded dataset produced by the Met Office Hadley Centre (MOHC), further details of which are available in Perry and Hollis (2005a, 2005b). The MOHC monthly series spans and the daily series spans Daily data were extended back to 1931 using regressions against the HadUKP series, and monthly data were extended back to 1873 for the five England and Wales regions and to 1766 for England and Wales as a whole. The main advantage of this new series is that it is based on a more extensive network of rain gauges, thus reducing aspects caused, for example, by rain gauges being biased towards the drier parts of a region, but it has the downside of relying heavily on interpolation of the available digitized precipitation data (see Perry and Hollis, 2005a, 2005b for details of station numbers used). Further details of the methods used, and associated error estimates, are provided by Simpson and Jones (2012). The names and abbreviations of the UK regions are provided in Table Methods of assessing trends in extreme precipitation For this analysis, a number of indices are used to analyse changes in precipitation extremes over the UK over the 2013 Royal Meteorological Society

2 ANALYSIS OF UK PRECIPITATION EXTREMES 2439 Table 1. The weights used to derive the national series for EW, S and NI. The NI weight is the multiplier to maintain homogeneity with the UK Met Office for the province (Jones and Conway, 1997). Region Weight South Scotland (SS) East Scotland (ES) North Scotland (NS) North West England and North Wales (NWE) North East England (NEE) a Central and Eastern England (CEE) South East England (SEE) South West England and South Wales (SWE) Northern Ireland (NI) a Includes a small part of Southeastern Scotland. respective periods for which data are available, for each of the four meteorological seasons (winter is defined as December to February, spring as March to May, summer as June to August and autumn as September to November): Percentiles of daily wet day rainfall totals (50, 90, 95 and 99), restricted to days with 1 mm precipitation or more. The maximum 5-day precipitation total. The wet day or simple rainfall intensity index (total precipitation/number of days with 1 mm precipitation or more). The consecutive dry day index (longest number of consecutive days with less than 1 mm precipitation). For maximum 5-day precipitation totals, the end day must fall within the season, whereas for the consecutive dry day index the entire period must fall within the season. The percentile thresholds of daily wet day rainfall totals have been calculated by re-arranging the daily precipitation totals from lowest to highest and using the following formula to determine the rank corresponding to the relevant percentile: n = PN (1) 100 where N is the number of values, P is the required percentile (e.g. for the 50th percentile, P = 50) and n is the rank corresponding to the required percentile. Trend lines have derived from linear regression against time, fitting the data to the equation y = A + Bx, and statistical significance has been estimated by determining the number of standard deviations of the trend relative to the mean value. In the samples considered, the number of observations is no lower than 81 (for the cases of winter daily precipitation indices) which makes this a close approximation of the true statistical significance. The analysis covers each of the UK regions individually and the national series for Scotland and England & Wales. Indices of mean and extreme precipitation are compared with the following three indices of the North Atlantic Oscillation (NAO): The Jones et al. (1997) and Osborn (2011) NAO series, derived from sea-level pressure series of SW Iceland and Gibraltar. This NAO series is intended primarily for the winter quarter but values have been used and calculated for the other three seasons. The Cornes et al. (2013) NAO series, based on pressure readings for London and Paris, which extends Table 2a. Trend values (mm) divided by the population standard deviation for each region, for each extreme precipitation index, for winter (DJF). 50th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day Table 2b. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Jones et al., 1997, Osborn, 2011) for winter (DJF) for each region for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day

3 2440 I. R. SIMPSON AND P. D. JONES Table 2c. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Cornes et al., 2013) for winter (DJF) for each region for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day (a) (b) NS 50th percentile winter precipitation Cornes et al. (2013) winter NAO Figure 1. (a) Seasonal winter (DJF) North Atlantic Oscillation index for , as per the series by Jones et al. (1997) and Osborn (2011) (left) and the Cornes et al. (2013) series (right). The red line shows a decadal moving average of the series. (b) The 50th percentile of NS precipitation (red) and the trend in the winter NAO (blue) stemming from Cornes et al. (2013), plotted as a time series (left) and scatter plot (right). The correlation is significant at the 99% level and is the strongest of the observed correlations between regional winter precipitation and the winter NAO index. further back in time. This, too, is primarily a winter NAO index but values for spring, summer and autumn have been calculated and used. The Folland et al. (2009) summer NAO index, which applies only to high summer (July and August), takes into account the issue that prevailing atmospheric circulation patterns differ at different times of the year, and thus a winter-based NAO index may not be as useful for the summer months as for the winter months. Precipitation values are correlated with NAO values for the corresponding season (e.g. spring precipitation values are correlated with the corresponding spring NAO values). 3. Trends in UK extreme daily precipitation 3.1. Daily winter precipitation during All regions of the UK have seen an increase in mean and extreme daily winter precipitation during the period, with positive trends for the 50th, 90th, 95th and 99th percentiles for wet day precipitation, maximum

4 ANALYSIS OF UK PRECIPITATION EXTREMES 2441 (a) (b) (c) (d) Figure 2. (a) Trends in the 50th percentile of precipitation (mm) for England and Wales (left) and Scotland (right) for each season for Seasonal values have been smoothed using a 10-point moving average. (b) As (a) but for the 90th percentile. (c) As (a) but for the 95th percentile. (d) As (a) but for the 99th percentile. (e) As (a) but for 5-day maximum precipitation. (f) As (a) but for the wet day index. (g) As (a) but for the consecutive dry day index. 5-day totals and the rainfall intensity index (Table 2a). The positive trend for the 50th percentile reaches the 95% significance level for the Scottish regions, and some indices reach 95% significance for NI and the England and Wales regions, with the exception of Central and Eastern England (CEE). The most significant increases have occurred in the 50th percentile and the rainfall intensity index, suggesting that mean precipitation and mean wet day totals have seen the most significant increases. The trend in the consecutive dry day index has been opposite to rainfall totals (as might be surmised), with the exception of East Scotland (ES) where the trend

5 2442 I. R. SIMPSON AND P. D. JONES (e) (f) (g) Figure 2. Continued Table 3a. Trend values (mm) divided by the population standard deviation for each region, for each extreme precipitation index, for spring (MAM). 50th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive. dry day in the consecutive dry day index has been upward in spite of significant increases in rainfall. The change in the consecutive dry index has not proved statistically significant in any regions of the UK. Table 2b shows the correlations between the winter NAO and indices of extreme precipitation across the UK regions, using the NAO indices provided by Jones et al. (1997) and Osborn (2011). This shows positive values for the Scottish regions, which are statistically significant for South Scotland (SS), North Scotland (NS) and S but, with the exception of the 50th percentile, not for ES. Table 2c shows the correlations between the NAO index (based on longer sea-level pressure series for London and Paris) provided by Cornes et al. (2013) and indices of extreme precipitation, which show more significant correlations, particularly for SS, NS, North West England and North

6 ANALYSIS OF UK PRECIPITATION EXTREMES 2443 Table 3b. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Jones et al., 1997; Osborn, 2011) for spring (MAM) for each region for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day Table 3c. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Cornes et al., 2013) for spring (MAM) for each region for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day Table 4a. Trend values (mm) divided by the population standard deviation for each region, for each extreme precipitation index, for summer (JJA). 50th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day Table 4b. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Jones et al., 1997; Osborn, 2011) for summer (JJA) for each region, for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day

7 2444 I. R. SIMPSON AND P. D. JONES Table 4c. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Cornes et al., 2013) for summer (JJA) for each region, for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day Table 4d. Correlations between observed precipitation values by each extreme precipitation index and the summer NAO (Folland et al., 2009) for high summer (JA) for each region, for th percentile th percentile th percentile th percentile Max 5 day total Wet day index Consecutive dry day Wales (NWE) and South West England and South Wales (SWE). Figure 1(a) shows the trend in both indices of the winter NAO, and Figure 1(b) shows the close relationship between the 50th percentile of NS precipitation and the Cornes et al. (2013) winter NAO. In particular, the peak in winter storminess and a strongly positive NAO in the 1990s, followed by a decline in the 2000s (Wang et al., 2008) corresponds with a peak in Scottish extreme precipitation values in the 1990s followed by a decline in the 2000s (Figure 2(a)). These results imply that the observed changes in precipitation across Northern Britain are at least partly associated with changes in the NAO Daily spring precipitation during Spring precipitation has shown an upward trend in most regions but fewer of the observed trends have been statistically significant (Table 3a). The most significant increases have occurred in the 50th percentile for NS and the wet day index for SS, NS and S, all of which are significant at the 99% level. The maximum 5-day totals only show a significant increase for S and NS. The correlations with the NAO, whether using the Osborn (2011) index (see Table 3b) or Cornes et al. (2013) index (see Table 3c) are weaker than in winter, although many of the positive correlations for NS are significant at the 99% level. The graphs of spring precipitation for England and Wales (Figure 2(a) (g)) do not show strong evidence of any long-term trend, but over Scotland most indices show an increase in both mean and extreme precipitation across Scotland during the 1980s and 1990s, followed by a decline since the late 1990s. The weaker correlations observed may be partly because these two indices of the NAO have been developed based on the behaviour of the NAO during the winter quarter Daily summer precipitation during Trends in both mean and extreme summer precipitation have been mostly negative except for North East England (NEE), CEE and South East England (SEE) (Table 4a) but do not generally reach the 95% significance level. There is a statistically significant reduction in the consecutive dry day index for SS, ES and S. The graphs Table 5a. Trend values (mm) divided by the population standard deviation for each region, for each extreme precipitation index, for autumn (SON). 50th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day

8 ANALYSIS OF UK PRECIPITATION EXTREMES 2445 Table 5b. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Jones et al., 1997; Osborn, 2011) for autumn (SON) for each region, for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day Table 5c. Correlations between observed precipitation values by each extreme precipitation index and the winter NAO (Cornes et al., 2013) for autumn (SON) for each region, for th percentile th percentile th percentile th percentile Max 5-day total Wet day index Consecutive dry day for England and Wales and Scotland (Figure 2(a) (g)) do not suggest a marked long-term trend, but for EW and S the summers of the 1980s and 1990s were relatively dry, followed by wetter summers since the turn of the century. The results suggest that the trend towards drier summers during has not exceeded the bounds of natural variability in most cases, and that there has been no significant increase in prolonged dry spells (as measured by the consecutive dry day index) some regions have seen a significant decrease. Correlations between summer precipitation and the Osborn (2011) index of the NAO (Table 4b) show mostly negative values but only eight of the correlations are statistically significant. The results from the Cornes et al. (2013) version (Table 4c) show mostly positive correlations, especially for S, SS, ES and NWE, and show a statistically significant negative correlation with the consecutive dry day index for most regions. Both of these indices are based on the Osborn (2011) and Cornes et al. (2013) definitions of the NAO and may not be the most appropriate indices to use for summer. The Folland et al. (2009) summer NAO index (see Table 4d), which applies only to July and August, produces strongly negative correlations with extreme rainfall indices except for NEE, CEE and SEE, and strongly positive correlations with the consecutive dry day index. The trends are of opposite sign relative to the Cornes et al. (2013) index because the positive phase of the summer NAO as identified by Folland et al. (2009) has an area of high pressure extending from the Azores towards the British Isles, a synoptic situation which is usually associated with below-average rainfall, especially in southern Britain, whereas a positive winter NAO, as defined by the indices used by Osborn (2011) and Cornes et al. (2013), is traditionally associated with low pressure over Iceland and mild wet westerly or southwesterly regimes over Britain Daily autumn precipitation during The observed trends in autumn precipitation are weak and mixed in signal (Table 5a), with the only statistically significant results being an increase in the maximum 5-day totals for SS and S. The trend in the Scottish regions is upward but does not reach the 95% significance level, suggesting that the increase in mean and extreme Table 6. Trend values (mm/year) for each region for each season, for monthly precipitation, for Season SS ES NS NWE NEE CEE SEE SWE S NI EW Winter Spring Summer Autumn Annual

9 2446 I. R. SIMPSON AND P. D. JONES (a) Figure 3. (a) England and Wales seasonal and annual precipitation (mm) using Met Office gridded monthly rainfall data, There is an upward trend in winter and a downward trend in summer. (b) Scotland seasonal and annual precipitation (mm) derived from Met Office gridded monthly rainfall data, There is an upward trend in winter and a downward trend in summer. autumn precipitation in Scotland is indistinguishable from natural short-term variability. The correlations between precipitation and the autumn NAO stemming from Osborn (2011) (Table 5b) and Cornes et al. (2013) (Table 5c) are not as strongly significant as for winter, again suggesting that a different index of the NAO may be more appropriate for the autumn. The graphs for England and Wales and Scotland (Figure 2(a) (g)) do not show any obvious changes over the period, except for a peak in the 50th percentile for Scotland during the 1980s followed by a slow decline in the 1990s and 2000s. 4. Trends in UK mean monthly precipitation Figure 3(a) shows the trends in England and Wales monthly precipitation for each of the four meteorological seasons over the period , whereas Figure 3(b) shows the trends for Scotland monthly

10 ANALYSIS OF UK PRECIPITATION EXTREMES 2447 (b) Figure 3. Continued precipitation over the period Trends and statistical significance for each region for each season over the period are given in Table 6 ( is chosen for consistency with the statistical significance analysis in Section 3) Long-term trends in winter precipitation Over the period (Table 6 ), the trend in winter precipitation has been statistically insignificant, except for NS and S, where an upward trend reaches the 95% Correction added on 6 December 2013 after original online publication: references to Table 5(a,b) have been corrected to Table 6. confidence level. The trend over (not shown) falls short at the 95% confidence level for all regions. Evidence for increased mean winter precipitation since 1931 is thus weaker than the evidence for increased daily intensity, suggesting an emphasis on heavier and/or more prolonged daily precipitation. Figure 3(b) suggests that the increase in Scottish winter precipitation has arisen via a step-change around the mid-1980s rather than a steady upward trend, most likely associated with a shift towards a more positive winter NAO. Extending the England and Wales regions to introduces a small positive trend for those regions, though none of them reach 95% significance. In contrast, extending the England and Wales series to

11 2448 I. R. SIMPSON AND P. D. JONES results in an upward trend that is significant at the 99% level. Figure 3(a) shows an erratic upward trend in England and Wales winter precipitation over the 241-year record, with a mean of near 200 mm until around 1860, increasing to nearer 250 mm since the 1920s, and this may explain the statistically significant upward trend noted since The trend may be due to higher sea surface temperatures and/or a change towards a more positive NAO. It is, however, possible that improved rainfall and snowfall recording may be a factor. In the HadUKP series, while seven sites per region were used from 1859 onwards, fewer sites were used from 1766 to 1858 due to limited availability of sites with long-standing and consistent records (Wigley et al., 1984) which means that the relationship between the HadUKP and Simpson and Jones (2012) series during is not guaranteed to be consistent with that from 1859 to present Long-term trends in spring precipitation Using the period (Table 6), spring precipitation has increased in most regions, with the largest increases occurring in Northern and Western Britain over the period. Increases reach the 99% significance level for SS, NS and S. Thus, unlike winter, the increases in mean daily intensity are matched or exceeded in terms of statistical significance by increases in the mean monthly amounts. The smallest trends are observed in CEE and SEE where they fall well short of statistical significance. Figure 3(b) suggests little or no trend in Scottish spring precipitation until approximately 1980 and then a sharp upward trend which as with the winter trend may be associated with the shift towards a positive NAO. Extending the period of coverage to (not shown) over the Scottish regions and Northern Ireland reduces the extent of the upward trend for the Scottish regions, but increases are significant at the 95% level for NI, and at the 99% level for SS, NS and S. Extending the coverage for the England and Wales sub-regions to produces statistically insignificant trends in precipitation for those regions. England and Wales as a whole saw a statistically insignificant trend in spring precipitation during , and Figure 3(a) illustrates the lack of a strong trend in spring precipitation in England and Wales Long-term trends in summer precipitation Over the period (Table 6), mean monthly summer precipitation shows a downward trend in most regions, but the trend falls short at the 95% significance level for all regions, suggesting that statistically insignificant trends in extreme summer precipitation have been matched by statistically insignificant trends in mean summer precipitation. The Scottish trends during are also downward, but only reach the 95% significance level for ES, while for NI the significance of the trend increases, reaching the 99% level. The England and Wales regions show a more significant decline in precipitation when the period is extended to , with NWE displaying a reduction that is significant at the 99% level, though the decline is not statistically significant for the other regions. England and Wales as a whole have had a downward trend that reaches the 99% level of significance over Figure 3(a) and (b) indicates that over England and Wales and also Scotland most of the observed decrease in summer precipitation has occurred since 1960, but the England and Wales series also suggests a small downward trend between 1766 and Long-term trends in autumn precipitation Over the period (Table 6), trends in autumn precipitation have been upward but only reach the 95% significance level for NS and S, implying that the trends are within the bounds of natural variability. This is similar to the existent, but not statistically significant, evidence for an increase in extreme precipitation in autumn. When the Scottish series is extended to cover (not shown), the trends remain significant at the 95% level for NS and S, and also becomes significant at the 95% level for SS. The England and Wales regions over show statistically insignificant downward trends in SEE and CEE, and weak upward trends in the other regions. Over , the trend in autumn precipitation over England and Wales falls well short at the 95% significance level. Figure 3(a) suggests no long-term trend in autumn precipitation over England and Wales, with the comparatively high decadal average of the 1990s and 2000s being skewed upwards by the recordbreaking wet autumn in Figure 3(b) suggests an erratic upward trend in Scottish autumn precipitation since approximately 1975, but not of sufficient magnitude to be distinguishable from short-term natural variability. 5. Conclusions The observed trends in UK precipitation are mostly consistent with the projections from climate models, with evidence of a trend towards wetter winters (especially in Scotland) and drier summers. The trends in mean precipitation have generally been less significant than the trends in extreme precipitation, and many years since 2000 have bucked the trend towards wetter winters and drier summers. Particularly in the case of the precipitation increases in winter, it is probable that much of the increase can be explained by the NAO, though it is also possible that anthropogenic forcing may have contributed to the observed changes to the NAO (Gillett et al., 2003). It is clear that the intensification and poleward shift of the Atlantic storm track during the 1980s and 1990s was too extreme to be due to anthropogenic forcing alone (Gillett, 2005). The results obtained here also strongly suggest

12 ANALYSIS OF UK PRECIPITATION EXTREMES 2449 that when comparing atmospheric circulation with precipitation, it is better to develop separate indices for the NAO for each season (e.g. see Folland et al., 2009, for a possible alternative summer NAO index). Thus, there is a scope for future work in determining how much of the observed changes in precipitation can be explained by changes in atmospheric circulation. As of 2012, it has proved harder to determine a detectable anthropogenic influence on the recent behaviour of UK precipitation than with temperature, and a series of wet summers starting in 2007 appear to have offset the significant downward trend in summer precipitation noted in many earlier studies. Similarly, the trend towards a strongly positive winter NAO and increased precipitation in Northwestern Britain has reversed during the past decade. Acknowledgements This work was supported by funding from the Natural Environment Research Council grant NE/F006888/1 and a CASE award 23 from the UK Meteorological Office. Matthew Perry and Dan Hollis produced the gridded precipitation data and provided extended series to the end of David Parker, Tim Osborn, Keith Briffa, John Caesar and Elizabeth Good provided useful assistance and comments. References Alexander LV, Jones PD Updated precipitation series for the U.K. and discussion of recent extremes. Atmos. Sci. Lett. 1, DOI: /asle Beveridge WH Wheat measures in the Winchester Rolls. Econ. History 5: Cornes RC, Jones PD, Briffa KR, Osborn TJ Estimates of the North Atlantic Oscillation back to 1692 using a Paris- London westerly index. Int. J. Climatol. 33: , DOI: / joc Folland CK, Knight J, Linderholm HW, Fereday D, Ineson S, Hurrell JW The summer North Atlantic oscillation: past, present and future. J. Climate 22(5): Fowler HJ, Wilby RL Detecting changes in seasonal precipitation extremes using regional climate model projections: implications for managing fluvial flood risk. Water Resour. Res. 46: W03525, DOI: /2008WR Gillett NP, Zwiers FW, Weaver AJ, Stott PA Detection of human influence on sea-level pressure. Nature 422(6929): Gillett NP Northern Hemisphere circulation. Nature 437(7058): 496. Gower JC A note on the periodogram of the Beveridge Wheat Price Index. J. Roy. Statist. Soc. Ser. B (Methodological) 17: Gregory JM, Jones PD, Wigley TML Precipitation in Britain: an analysis of area-average data updated to Int. J. Climatol. 11: Jones PD, Conway D Precipitation in the British Isles: an update of area-average analysis to Int. J. Climatol. 17: Jones PD, Jónsson T, Wheeler D Extension to the North Atlantic Oscillation using early instrumental pressure observations from Gibraltar and SW Iceland. Int. J. Climatol. 17: Maraun D, Osborn TJ, Gillett NP United Kingdom daily precipitation intensity: improved early data, error estimates and an update from 2000 to Int. J. Climatol. 28: Nicholas FJ, Glasspoole J General monthly rainfall over England and Wales 1727 to British Rainfall 1931: Osborn T, Hulme M, Jones PD, Basnett TA Observed trends in the daily intensity of United Kingdom precipitation. Int. J. Climatol. 20: Osborn T Winter 2009/2010 temperatures and a record-breaking North Atlantic Oscillation index. Weather 66(1): Perry M, Hollis D. 2005a. The development of a new set of long-term climate averages for the UK. Int. J. Climatol. 25: Perry M, Hollis D. 2005b. The generation of monthly gridded datasets for a range of climate variables across the UK. Int. J. Climatol. 25: Simpson IR, Jones PD Updated precipitation series for the UK derived from Met Office gridded data. Int. J. Climatol. 32: Wang XL, Swail VR, Zwiers FW, Feng Y Trends and variability of storminess in the Northeast Atlantic region, Climate Dynam. 33(7 8): Wigley TML, Jones PD England and Wales precipitation: A discussion of recent changes in variability and an update to J. Climatol. 7: Wigley TML, Lough JM, Jones PD Spatial patterns of precipitation in England and Wales and a revised, homogeneous England and Wales precipitation series. J. Climatol. 4: 1 25.