• Bader, J., , and M. Latif, 2003: The impact of decadal-scale Indian Ocean sea surface temperature anomalies on Sahelian rainfall and the North Atlantic Oscillation. Geophys. Res. Lett, 30 .2169, doi:10.1029/2003GL018426.

    • Search Google Scholar
    • Export Citation
  • Barnston, A. G., , and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev, 115 , 10831126.

    • Search Google Scholar
    • Export Citation
  • Beniston, M., 2004: The 2003 heat wave in Europe: A shape of things to come? Geophys. Res. Lett, 31 .L02202, doi:10.1029/2003GL018857.

  • Chelliah, M., , and C. F. Ropelewski, 2000: Reanalyses-based tropospheric temperature estimates: Uncertainties in the context of global climate change detection. J. Climate, 13 , 31873205.

    • Search Google Scholar
    • Export Citation
  • Chelliah, M., , and G. D. Bell, 2004: Tropical multidecadal and interannual climate variability in the NCEP–NCAR reanalysis. J. Climate, 17 , 17771803.

    • Search Google Scholar
    • Export Citation
  • Folland, C. K., , T. N. Palmer, , and D. E. Parker, 1986: Sahel rainfall and worldwide sea temperatures, 1901–85. Nature, 320 , 602607.

  • Giannini, A., , R. Saravanan, , and P. Chang, 2003: Oceanic forcing of Sahel rainfall on interannual to interdecadal time scales. Science, 302 , 10271030.

    • Search Google Scholar
    • Export Citation
  • Greatbatch, R. J., 2000: The North Atlantic Oscillation. Stochast. Environ. Res. Risk Assess, 14 , 213242.

  • Hoerling, M. P., , J. W. Hurrell, , and T. Xu, 2001: Tropical origin for recent North Atlantic climate change. Science, 292 , 9092.

  • Hurrell, J. W., , and C. K. Folland, 2002: A change in the summer atmospheric circulation over the North Atlantic. CLIVAR Exchanges, Vol. 7, No. 3/4, International CLIVAR Project Office, Southampton, United Kingdom, 52–54.

  • Hurrell, J. W., , Y. Kushnir, , G. Ottersen, , and M. Visbeck, 2003: An overview of the North Atlantic Oscillation. The North Atlantic Oscillation, Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 1–35.

  • Inoue, T., , and J. Matsumoto, 2004: A comparison of summer sea level pressure over east Eurasia between NCEP–NCAR reanalysis and ERA-40 for the period 1960–99. J. Meteor. Soc. Japan, 82 , 951958.

    • Search Google Scholar
    • Export Citation
  • Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-Year Reanalysis: Monthly means CD-ROM and documentation. Bull. Amer. Meteor. Soc, 82 , 247268.

    • Search Google Scholar
    • Export Citation
  • Levinson, D. H., , and A. M. Waple, 2004: State of the climate in 2003. Bull. Amer. Meteor. Soc, 85 , S1S72.

  • Lu, J., , R. J. Greatbatch, , and K. A. Peterson, 2004: On the trend in Northern Hemisphere winter atmospheric circulation during the last half of the twentieth century. J. Climate, 17 , 37453760.

    • Search Google Scholar
    • Export Citation
  • Manley, G., 1974: Central England temperatures: Monthly means 1659 to 1973. Quart. J. Roy. Meteor. Soc, 100 , 389405.

  • Pal, J. S., , F. Giorgi, , and X. Q. Bi, 2004: Consistency of recent European summer precipitation trends and extremes with future regional climate projections. Geophys. Res. Lett, 31 .L13202, doi:10.1029/2004GL019836.

    • Search Google Scholar
    • Export Citation
  • Parker, D. E., , L. V. Alexander, , and J. Kennedy, 2004: Global and regional climate in 2003. Weather, 59 , 145152.

  • Parker, D. E., , T. P. Legg, , and C. K. Folland, 1992: A new daily central England temperature series, 1772–1991. Int. J. Climatol, 12 , 317342.

    • Search Google Scholar
    • Export Citation
  • Rodwell, M. J., , and C. K. Folland, 2002: Atlantic air-sea interaction and seasonal predictability. Quart. J. Roy. Meteor. Soc, 128 , 14131443.

    • Search Google Scholar
    • Export Citation
  • Rodwell, M. J., , and C. K. Folland, 2003: Atlantic air-sea interaction and model validation. Ann. Geophys, 46 , 4756.

  • Sterl, A., 2004: On the (in)homogeneity of reanalysis products. J. Climate, 17 , 38663873.

  • Sturaro, G., 2003: A closer look at the climatological discontinuities present in the NCEP–NCAR reanalysis temperature due to the introduction of satellite data. Climate Dyn, 21 , 309316.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , and D. A. Paolino Jr., 1980: The Northern Hemisphere sea-level pressure data set: Trends, errors and discontinuities. Mon. Wea. Rev, 108 , 855872.

    • Search Google Scholar
    • Export Citation
  • Ward, M. N., 1998: Diagnosis and short-lead time prediction of summer rainfall in tropical North Africa at interannual and multidecadal timescales. J. Climate, 11 , 31673191.

    • Search Google Scholar
    • Export Citation
  • Yang, S., , K-M. Lau, , and K-M. Kim, 2002: Variations of the East Asian jet stream and Asian–Pacific–American winter climate anomalies. J. Climate, 15 , 306325.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    The upper panels show the linear trend in SLP (hPa yr−1) over the period 1958–2002 in (a) the NCEP–NCAR reanalysis, (b) the ERA-40 reanalysis, and (c) Trenberth's SLP dataset. The lower panels show the time series of area mean SLP over (d) Africa and (f) Asia within the enframed areas in the top panels, and for the different datasets. (e) The corresponding time series for central England.

  • View in gallery

    The time series of SLP in the NCEP–NCAR reanalysis over Africa and Asia taken from Fig. 1.

  • View in gallery

    Same as in Fig. 1 except for 500-hPa geopotential height (gpm yr−1), and for the NCEP–NCAR and ERA-40 reanalyses only.

  • View in gallery

    Same as in Fig. 3 except for 2-m temperature (K yr−1).

  • View in gallery

    The (undetrended) interannual correlation between the different datasets during periods when each pair have overlapping availability (see text for details). Upper panels are for the NCEP–NCAR and ERA-40 reanalyses (a) SLP, (b) 500-hPa geopotential height, and (c) 2-m temperature. The lower panels are for Trenberth's dataset compared with (d) the NCEP–NCAR and (e) the ERA-40 reanalysis.

  • View in gallery

    The upper panels show the linear trend in 500-hPa temperature (K yr−1), 1958–2002, using (a) the NCEP–NCAR and (b) the ERA-40 reanalyses. (c), (d) The corresponding linear trends at 100 hPa. The (undetrended) interannual correlation between the two datasets over the same time period for (e) 500- and (f) 100-hPa temperature.

  • View in gallery

    The average for the entire latitude band from 17.5°N to the equator for (a) 500- and (b) 100-hPa temperature.

  • View in gallery

    The upper panels show the spatial pattern obtained by regressing the respective NAO indices against the summer mean SLP for (a) the NCEP–NCAR reanalysis, (b) the ERA-40 reanalysis, and (c) Trenberth's dataset. Negative contours are dashed and the zero and positive contours are solid. The contour interval is 5 hPa. (d) The corresponding NAO indices (NCEP–NCAR, dashed; ERA-40, solid; Trenberth, dot–dash).

  • View in gallery

    (a) The time series of summer mean central England temperature (CET) (solid line) and NAO index (dashed line) from Trenberth's dataset. (b) The sliding window (19-yr width) cross correlation between the two time series in (a). (c) The (interannual) correlation between summer mean SLP and CET over the 19-yr centered at 1921. Dashed contours indicate negative correlations; solid contours the zero line and positive correlations. The contour interval is 0.15 and the 5% significance level is 0.45, indicating correlations significantly different from zero at the 5% level near the regions of maximum positive and negative correlation.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 38 38 6
PDF Downloads 16 16 0

Discrepancies between Different Northern Hemisphere Summer Atmospheric Data Products

View More View Less
  • 1 Department of Oceanography, Dalhousie University, Halifax, Nova Scotia, Canada
© Get Permissions
Full access

Abstract

Northern Hemisphere summer (July–August) data from the NCEP–NCAR and ECMWF 40-yr Re-Analysis (ERA-40) reanalyses are compared with each other and with Trenberth's sea level pressure (SLP) dataset. Discrepancies in SLP and 500 hPa are mostly confined to a band connecting North Africa and Asia. In the NCEP–NCAR reanalysis, there is a negative offset in SLP over North Africa and Asia prior to the late 1960s, together with a similar problem in 500-hPa height, and in Trenberth's data there is a negative offset in SLP over Asia prior to the early 1990s. Both these offsets magnify the linear trend from 1958 to 2002 over North Africa and Asia in the NCEP–NCAR and Trenberth datasets. On the other hand, the interannual variability in the three datasets is highly correlated during the periods between these offsets. Compared to SLP and 500-hPa height, there is a more extensive area of discrepancy in 2-m temperature that extends eastward from North Africa across the subtropics into the Pacific, with an additional area of discrepancy over the Arctic and parts of the American continent. At 500 and 100 hPa, the biggest differences in the temperature time series are found in the Tropics, with a marked jump being evident in the late 1970s in the NCEP–NCAR, but not in the ERA-40, reanalysis that is almost certainly associated with the introduction of satellite data. On the other hand, all three datasets agree well over Europe. The summer North Atlantic Oscillation (NAO), defined here as the first EOF of summer mean SLP over the Euro-Atlantic sector, agrees well between the different datasets. The results indicate that the upward trend in the summer index in the 1960s is part of a longer-period interdecadal cycle, with relatively high index values also being found during the 1930s. The running cross correlation between the central England temperature record and the summer NAO shows a strong correlation throughout the last half of the twentieth century, but much reduced correlation in the early part of the twentieth century. It is not clear whether the change in correlation is real, or a data artifact, a topic that requires further research.

Corresponding author address: Richard J. Greatbatch, Dept. of Oceanography, Dalhousie University, Halifax, NS B3H 4J1, Canada. Email: richard.greatbatch@dal.ca

Abstract

Northern Hemisphere summer (July–August) data from the NCEP–NCAR and ECMWF 40-yr Re-Analysis (ERA-40) reanalyses are compared with each other and with Trenberth's sea level pressure (SLP) dataset. Discrepancies in SLP and 500 hPa are mostly confined to a band connecting North Africa and Asia. In the NCEP–NCAR reanalysis, there is a negative offset in SLP over North Africa and Asia prior to the late 1960s, together with a similar problem in 500-hPa height, and in Trenberth's data there is a negative offset in SLP over Asia prior to the early 1990s. Both these offsets magnify the linear trend from 1958 to 2002 over North Africa and Asia in the NCEP–NCAR and Trenberth datasets. On the other hand, the interannual variability in the three datasets is highly correlated during the periods between these offsets. Compared to SLP and 500-hPa height, there is a more extensive area of discrepancy in 2-m temperature that extends eastward from North Africa across the subtropics into the Pacific, with an additional area of discrepancy over the Arctic and parts of the American continent. At 500 and 100 hPa, the biggest differences in the temperature time series are found in the Tropics, with a marked jump being evident in the late 1970s in the NCEP–NCAR, but not in the ERA-40, reanalysis that is almost certainly associated with the introduction of satellite data. On the other hand, all three datasets agree well over Europe. The summer North Atlantic Oscillation (NAO), defined here as the first EOF of summer mean SLP over the Euro-Atlantic sector, agrees well between the different datasets. The results indicate that the upward trend in the summer index in the 1960s is part of a longer-period interdecadal cycle, with relatively high index values also being found during the 1930s. The running cross correlation between the central England temperature record and the summer NAO shows a strong correlation throughout the last half of the twentieth century, but much reduced correlation in the early part of the twentieth century. It is not clear whether the change in correlation is real, or a data artifact, a topic that requires further research.

Corresponding author address: Richard J. Greatbatch, Dept. of Oceanography, Dalhousie University, Halifax, NS B3H 4J1, Canada. Email: richard.greatbatch@dal.ca

1. Introduction

Summer 2003 was exceptionally hot and dry in Europe, resulting in excess of 20 000 heat-related deaths and over U.S. $10 billion of agricultural losses (Parker et al. 2004). At Paris, France, temperatures exceeded 35°C on 10 consecutive days in early August, and the United Kingdom recorded its highest-ever daily maximum temperature of 38.5°C on 10 August (Levinson and Waple 2004). In Germany, it was the hottest summer since 1901 and in Switzerland summer 2003 is believed to have been the hottest since 1540 (Beniston 2004). Beniston (2004) notes the similarity between summer 2003 and model projections under increasing greenhouse gas forcing of summer conditions in Europe for the latter part of the twenty-first century.

While summer 2003 was clearly exceptional, many aspects of summer 2003 are consistent with recent trends in the atmospheric circulation over Europe during summer (e.g., Pal et al. 2004). Hurrell and Folland (2002) noted a shift toward more anticyclonic conditions over Europe during high summer that took place during the late 1960s. Associated with this shift, these authors note a warming trend in mean central England surface air temperature (CET; Manley 1974; Parker et al. 1992), and a roughly 15% reduction in summer precipitation over much of northern Europe since the late 1960s compared to the previous 20 yr (see also Pal et al. 2004). Hurrell and Folland (2002) relate these changes to a northward shift in the storm track over the North Atlantic Ocean and northern Europe that took place during the late 1960s. The shift in the storm track appears to be connected to an upward shift in the summer North Atlantic Oscillation (NAO) index that also took place during the late 1960s (Hurrell and Folland 2002). The well-documented Sahelian drought also began during the 1960s (Folland et al. 1986; Ward 1998), and Hurrell and Folland (2002) speculate that the regime shift over Europe may be connected with the onset of the drought. Some authors have related the drought to an upward trend in sea surface temperature (SST) in the tropical oceans, especially the Indian Ocean (Giannini et al. 2003; Bader and Latif 2003). Tropical SST trends have also been implicated in recent trends in the Northern Hemisphere winter atmospheric circulation (Hoerling et al. 2001; Lu et al. 2004). In addition, Chelliah and Bell (2004) have noted the close connection between what they call the “tropical multidecadal mode” and trends during the last half of the twentieth century not only in the Tropics, but also in the Northern Hemisphere extratropics in both the boreal winter and summer seasons.

Our original motivation for the work presented here was to try and determine the cause of the recent trend in the Northern Hemisphere summer circulation, following the approach used by Lu et al. (2004) to study the winter circulation trend; that is, using global, daily mean data to compute the forcing for a simple dynamical model, and then to use the model to determine the source of the trend. However, it soon became apparent that there are significant discrepancies between different data products [e.g., the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis and the European Centre for Medium-Range Weather Predictions (ECMWF) 40-yr Re-Analysis (ERA-40)] for Northern Hemisphere summer, thereby casting doubt on the usefulness of these products as model diagnostics. Chelliah and Ropelewski (2000) claim that linear trend analyses applied to reanalysis products have limited usefulness as tools to detect climate change, except when the signal is large and shows consistency among all datasets. As we show in section 3, there are differences in the linear trend for 1958–2002 during boreal summer between the different datasets, especially over North Africa and Asia. Yang et al. (2002) have noted a discrepancy over eastern Asia between the NCEP–NCAR reanalysis and Trenberth's sea level pressure dataset that is consistent with our findings (the details on the different datasets are given in section 2), an issue that has been explored further by Inoue and Matsumoto (2004). Inoue and Matsumoto point to a specific problem with the NCEP–NCAR reanalysis prior to the late 1960s. Here, we show that the problem over Asia noted by Inoue and Matsumoto is connected with a similar problem over North Africa, and in a band connecting the two regions. Problems with the data products over North Africa are particularly unfortunate given the possible connection between the drought in the Sahel region of Africa, and the recent, decadal trends over Europe (Hurrell and Folland 2002).

In section 3, the linear trend in some basic variables such as SLP, geopotential height, and air temperature are compared. In section 4 the summer NAO index is computed and compared in the different datasets, and a sliding window correlation between the NAO index and the CET time series is examined. Section 5 provides a summary and conclusions.

2. The data

The data used here come from three different main sources: the NCEP–NCAR reanalysis (e.g., Kistler et al. 2001), the ECMWF reanalysis (ERA-40; information available online at http://data.ecmwf.int/data), and Trenberth's SLP dataset (Trenberth and Paolino 1980; information online at http://dss.ucar.edu/datasets/ds010.1/). The latter is included for comparison purposes (while recognizing the much more comprehensive nature of the reanalysis products). The NCEP–NCAR and ERA-40 reanalyses have global coverage and a grid resolution of 2.5° × 2.5°, with the NCEP–NCAR data covering the period from 1948 to 2003, and the ERA-40 data from 1958 to 2002. Trenberth's SLP data cover the region north of 12.5°N at a grid resolution of 5° × 5° from 1901 to 2003. For all datasets, the months of July and August are averaged to represent the summer climate. We concentrate on July and August since these are the boreal summer months that show particularly enhanced warmth during the 15-yr period 1989–2003 in the CET record (see Fig. 8b in Parker et al. 2004), although the results we show are not greatly changed by including June. CET is taken from the Hadley Centre archive in the United Kingdom (information online at http://www.met-office.gov.uk/research/hadleycentre/CR-data/Monthly/HadleyCET-act.txt). The CET is a reliable, quality controlled temperature record that extends back several centuries.

3. The multidecadal trend

Figures 1a–c show the linear trend over the time period 1958–2002 in mean SLP for July and August (hereafter the summer) computed from each of the three datasets. We choose 1958–2002 since this is the period for which the data from all three datasets overlap. There are three regions that stand out: the Euro-Atlantic sector (particularly the center of action over Greenland), the Asian continent, and North Africa. As we shall see in section 4, the trend in the Euro-Atlantic sector reflects the upward trend in the summer NAO index and is in general agreement with the upward transition in the summer NAO index noted by Hurrell and Folland (2002). However, there are some differences between the different datasets over Asia and Africa. For example, all three datasets show an upward trend in SLP over Asia, but the magnitude and location of the maximum trend varies from dataset to dataset. Likewise, there is a pronounced upward trend in SLP over North Africa in the NCEP–NCAR reanalysis that is less pronounced in the other datasets.

Figures 1d,f show the time series of summer mean SLP averaged within the green boxes over North Africa and central Asia visible in the upper panels, and Fig. 1e shows the time series of summer mean SLP over central England. Over central England, all three datasets show very close agreement (the three curves are almost indistinguishable), no doubt reflecting the very good data coverage in this area. Over North Africa on the other hand (Fig. 1d), there is a clear offset in the NCEP–NCAR reanalysis from the other datasets up until the late 1960s, at which time there is a sudden upward jump in the NCEP–NCAR time series that brings it into line with the other time series. It is clear that it is the early, negative offset in the SLP time series over North Africa that explains the pronounced trend over this area in Fig. 1a, but not in Figs. 1b,c. In Fig. 1f, there is also the indication of a similar problem with the NCEP–NCAR reanalysis over Asia, a suggestion that is supported by overlapping the two time series in Fig. 2. Indeed, by comparing different data products with local station data, Inoue and Matsumoto (2004) conclude that the NCEP–NCAR reanalysis has a problem over Asia prior to the late 1960s [a problem that is also evident in boreal winter; Yang et al. (2002)]. After the upward jump in the late 1960s, the NCEP–NCAR and ERA-40 time series show generally good agreement over both North Africa and Asia, although there is the suggestion of a further upward offset in the NCEP–NCAR time series that begins in the mid-1970s. By comparison, the time series from Trenberth's data shows a downward offset over Asia (Fig. 1f) prior to the early 1990s that clearly dominates the trend in this dataset over Asia shown in Fig. 1c. Sudden jumps in the late 1960s in difference fields between the ERA-40 and NCEP–NCAR reanalyses have been noted previously by Sterl (2004). Sterl suggests that these jumps could arise from the use of different data streams in the two reanalyses. He notes that the ERA-40 reanalysis was conducted later than the NCEP–NCAR reanalysis, and that new datasets were used in the former. Sudden jumps in difference fields between the two reanalysis products during the 1970s could also be related to the introduction of satellite data (Sturaro 2003; Sterl 2004).

The general impression from the time series plotted in Figs. 1d–f is that, apart from the discrepancies noted above, the interannual variability often agrees well between the different datasets. We can check this by computing correlations between the various time series, the results of which are shown in Table 1. All of the correlations shown are significantly different from zero at the 1% level. After 1967, the NCEP–NCAR and ERA-40 reanalyses are particularly highly correlated, with a correlation coefficient of 0.9 over North Africa and 0.8 over Asia. The correlations with Trenberth's data are also in the 0.7–0.8 range over Asia, but it is clear that Trenberth's data agree less well with both NCEP–NCAR and ERA-40 over North Africa.

Next, we turn to 500-hPa geopotential height (HGT500). Figures 3a,b show the linear trend in the NCEP–NCAR and ERA-40 reanalyses, respectively, again for the period 1958–2002 and for summer (July–August) mean values. Once again, the two spatial patterns generally agree with each other over the Euro-Atlantic sector. In particular, both show a lowering trend in HGT500 over Greenland and north of Norway that can be related to the downward trend in SLP noted in Fig. 1 over Greenland. However, while the centers of action appear in similar places over Asia, there is a much more pronounced trend over Asia in the NCEP–NCAR than in the ERA-40 dataset. The bottom panels show the time series over North Africa, central England, and Asia—the values for Africa and Asia being averaged over the green boxes shown in the upper panels. (Note that the two panels are not exactly the same as in Fig. 1 because the spatial patterns, although roughly barotropic, slightly shift their locations with altitude.) Again, we find very good agreement over central England, but we find a big departure between the datasets over Asia during the 1960s. In fact, the time series of HGT500 from the NCEP–NCAR reanalysis over Asia shows the same “dip” as the corresponding SLP time series in the early 1960s (Fig. 1f), suggesting that the problem in the 1960s is part of the same problem we encountered earlier with the NCEP–NCAR dataset. Interestingly, after the mid-1970s, the two time series over Asia become almost identical (see also Table 2). Over North Africa, the discrepancy between the two datasets seems less severe at 500 hPa than at SLP, although a similar (if less severe) negative offset exists in the NCEP–NCAR reanalysis prior to 1970 to that found over Asia. One other point to note is that before 1977, there is the suggestion of the opposite trend in the ERA-40 data over Asia compared to the NCEP–NCAR data (Fig. 3c), which explains the difference between Figs. 3a,b over Asia.

The linear trend in summer 2-m air temperature (2mT), also between 1958 and 2002, is shown in Figs. 4a,b. There is a warming trend of 0.02 K yr−1 over central England that is found in both the NCEP–NCAR and ERA-40 reanalyses (see also Fig. 4d). This compares with a warming trend of 0.04 K yr−1 in the CET time series over the same time period (the reduced magnitude in the reanalyses is probably because, in the reanalyses, the temperature is averaged over a 2.5° grid box). Looking at the time series in Fig. 4c, we see the same “dip” over North Africa that we noted in the SLP time series (Fig. 1d), suggesting that it arises from the same cause, and over Asia, there is a notable offset between the two data products. Looking at Table 2, we see that the correlation between the two time series is higher over North Africa than over Asia during the period 1967–2002. Both datasets also show a marked upward trend in temperature within the African green box.

So far we have seen that there are discrepancies between the two datasets over North Africa and Asia. To appreciate the true spatial extent of the discrepancies, Fig. 5 shows the (undetrended) correlation between the different datasets, and for different fields, during the time periods the datasets overlap (so, when comparing either NCEP–NCAR or Trenberth's data with ERA-40, the time period is 1958–2002; for NCEP–NCAR and Trenberth's data, the time period is 1948–2003). Comparing the NCEP–NCAR and ERA-40 reanalyses, we see from Figs. 5a,b that Asia and North Africa are indeed regions of low correlation (the 10% significance level is about 0.3). Furthermore, there is a region of reduced correlation that connects these two regions, suggesting that the problem with the NCEP–NCAR reanalysis noted by Inoue and Matsumoto (2004) is not confined to Asia, but in fact extends in a band connecting Asia and North Africa. There is also the suggestion of reduced correlation in SLP over western North America that may be due to the Rocky Mountains and the extrapolation technique used to compute SLP. In 2mT the discrepancies spread more widely across North Africa, southern Asia, and into the Pacific Ocean, and there is also a region of reduced correlation over the Arctic and parts of the American continent. Comparing Trenberth's SLP with the other two SLP datasets (Figs. 5d,e), North Africa and Asia again stand out, and also the western part of North America, as in Fig. 5a.

Finally, we turn to the temperature at 500 and 100 hPa as revealed by the NCEP–NCAR and ERA-40 reanalyses. Looking at the trend in the two datasets (Figs. 6a,b), there are clearly many features in common at 500 hPa, notably a region of cooling that extends from Africa to Asia, and regions of warming over the British Isles and over North America, although there are also differences in detail between the two datasets. At 100 hPa, the NCEP–NCAR reanalysis (Fig. 6c) shows a strong warming trend in the Tropics that is not found in the ERA-40 data (Fig. 6d), even though both datasets show a cooling trend over middle and high latitudes. Looking at the correlation between the two time series (Figs. 6e,f), we again find much better agreement at middle latitudes than in the Tropics. It turns out that at 500 hPa, there are significant differences between the datasets in different locations in the Tropics and subtropics, but when averaged around the globe, the two datasets agree quite well (see Fig. 7a). This is not true at 100 hPa, however, where the averaged temperature in the NCEP–NCAR reanalysis shows a strong upward shift in the late 1970s (Fig. 7b). We speculate that this is due to the introduction of satellite data (Sturaro 2003).

4. Summer NAO

The NAO is the most important mode of variability (monthly time scales and longer) in the atmospheric circulation over the Euro-Atlantic sector (see Hurrell et al. 2003; Greatbatch 2000), and is the only teleconnection pattern that persists throughout the year in the Northern Hemisphere (Barnston and Livezey 1987). The NAO is weaker in summer than in winter, but nevertheless contributes more than 20% of the variance in summer seasonal mean SLP over the Euro-Atlantic sector according to Hurrell et al. (2003). The trend since the 1960s toward increased anticyclonicity over Europe was noted earlier (Hurrell and Folland 2002; Rodwell and Folland 2002, 2003) and is associated with an upward transition of the summer NAO index that occurred during the 1960s. In this section the summer NAO pattern and index are computed and compared using the different datasets.

Figure 8 shows the spatial patterns associated with the leading EOF and the corresponding principal component (PC) time series (defined here to be the NAO indices) for the NCEP–NCAR and ERA-40 reanalyses and Trenberth's SLP dataset, and computed from SLP averaged over July and August in each year. To avoid using data over North Africa (and so avoid the problems noted in section 3), the EOFs are computed over the domain 40°–70°N and 90°W−30°E, instead of 20°–70°N and 90°W−40°E as in Hurrell et al. (2003). In each case, the EOFs are computed using the whole time series of available data (NCEP–NCAR, 1948–2003; ERA-40, 1958–2002; Trenberth, 1901–2003). Despite the different lengths of the datasets, the spatial patterns are very similar in all three cases, and, during the time of overlap, the PC time series are also almost identical (correlation of 0.95 in all cases, exceeding the 1% significant level). The percentage of variance accounted for in each dataset is also similar (33%, 28%, and 30% for NCEP–NCAR, ERA-40, and Trenberth's data, respectively; the corresponding percentages for the second EOF are 21%, 21% and 19%, respectively). The time series confirm the upward transition of the NAO index during the 1960s, but the longer record available in Trenberth's dataset shows that the upward trend in the 1960s is really part of a multidecadal vacillation, with relatively high values of the index also being found during the 1930s. Figure 9 shows the time series of CET averaged over July and August, the July–August NAO index computed using Trenberth's data, and the sliding window (of 19-yr width) cross correlation between the two time series. After about 1950, the NAO accounts for about 60% or more of the variance in CET (correlation around 0.8 which is significantly different from zero at the 1% level), with positive NAO summers being associated with warmer than normal summers. This is not unexpected given that the positive phase of the NAO is associated with a high pressure anomaly over the British Isles. However, during the first half of the century the correlation is somewhat lower, and specifically for the 19 yr centered around 1920, the correlation drops to zero. The (interannual) correlation between summer mean SLP and the CET record over the period 1912–30, centered around 1920, is shown in Fig. 9c. The resulting pattern, while different from the NAO pattern (Fig. 8c), is, nevertheless, consistent with warm summer weather, and hence with the CET time series, since it is associated with anomalous southerly winds over the United Kingdom during anomalously warm summers. Nevertheless, it is not clear if the change in the relationship between the NAO and CET during the early part of the twentieth century is real, a topic that remains for future research.

5. Conclusions

The exceptional summer of 2003 in Europe (Beniston 2004) has raised interest in the summer season, especially since recent trends during the Northern Hemisphere summer are consistent with model projections under increasing greenhouse gas concentration in the atmosphere (Pal et al. 2004). Here we have documented some discrepancies between the trend in Northern Hemisphere summer (defined as July and August) as seen in a number of different variables in three different datasets, in particular, sea level pressure (SLP), 500-hPa height, and temperature at 2 m, 500 hPa, and 100 hPa in the NCEP–NCAR and ERA-40 reanalyses and Trenberth's SLP dataset. Over the last 50 yr, the NCEP–NCAR reanalysis shows an upward trend in SLP over North Africa, and all three datasets show an upward trend in SLP over Asia, and Europe, although over Asia the trend is larger for the NCEP–NCAR and Trenberth datasets. In the NCEP–NCAR reanalysis, the difference is accounted for by a negative offset of about 5–10 hPa (about 5–10 times larger than the magnitude of the interannual variability) over North Africa prior to the late 1960s, and also a similar problem at the same time over Asia that had been noted earlier by Yang et al. (2002) and Inoue and Matsumoto (2004). In Trenberth's SLP data there is a downward offset of about 5 hPa prior to 1994 over Asia that magnifies the trend in that dataset. On the other hand, the interannual variability in the three datasets is highly correlated during the periods between these offsets. The problem with SLP in the NCEP–NCAR reanalysis prior to the late 1960s is also evident in 500-hPa height data, especially over Asia. The discrepancies in SLP and 500-hPa height between the different datasets are confined mostly to a band between North Africa and Asia, but in 2-m temperature, there is a more extensive area of discrepancy extending eastward from North Africa across the subtropics into the Pacific, with an additional area of discrepancy over the Arctic and parts of the American continent. On the other hand, there is strong agreement between the different datasets over Europe, both reanalyses indicating the same warming trend. At 500 and 100 hPa, the biggest differences are found in the Tropics. In particular, the NCEP–NCAR reanalysis shows an upward trend in the Tropics at 100 hPa that is related to a strong upward jump in the late 1970s. The latter is almost certainly associated with the introduction of satellite data (Sturaro 2003; Sterl 2004).

At 500 hPa, both reanalyses show an upward trend in geopotential height over the British Isles and a corresponding downward trend over Greenland and the Arctic (Fig. 3). A downward trend is also found in SLP over Greenland in all three datasets, and a weaker upward trend over Europe. These changes in 500-hPa height and SLP are connected to an upward trend in the summer NAO index. (The spatial pattern of the trend shown in Fig. 3 projects strongly onto the pattern obtained by regressing 500-hPa height against the detrended summer NAO index shown in Fig. 8.) The NAO index was computed for all three datasets as the principal component time series associated with the first EOF in summer mean SLP over the Euro-Atlantic sector (the exact region used to compute the EOF is given in section 4). The spatial pattern of SLP associated with the summer NAO, its time series, and the percentage of variance explained (about 30%) show excellent agreement between the three datasets. The running cross correlation between the central England temperature record and the summer NAO shows a significant correlation (approaching 0.8 and significantly different from zero at the 1% level) throughout the last half of the twentieth century, but much reduced correlation in the early part of the twentieth century. It is not clear whether the change in correlation is real, or a data artifact—a topic that requires further research.

Acknowledgments

This research was funded by NSERC and CFCAS in support of the Canadian CLIVAR Research Network. We are grateful to Dr. Hai Lin for originally pointing out the discrepancy between the trend in the NCEP–NCAR and ERA-40 reanalyses over Asia, and to Dr. Jian Lu for his help in the early stages of this work. Comments from Simon Blessing, Dr. Holger Pohlmann, and two anonymous reviewers are also much appreciated.

REFERENCES

  • Bader, J., , and M. Latif, 2003: The impact of decadal-scale Indian Ocean sea surface temperature anomalies on Sahelian rainfall and the North Atlantic Oscillation. Geophys. Res. Lett, 30 .2169, doi:10.1029/2003GL018426.

    • Search Google Scholar
    • Export Citation
  • Barnston, A. G., , and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev, 115 , 10831126.

    • Search Google Scholar
    • Export Citation
  • Beniston, M., 2004: The 2003 heat wave in Europe: A shape of things to come? Geophys. Res. Lett, 31 .L02202, doi:10.1029/2003GL018857.

  • Chelliah, M., , and C. F. Ropelewski, 2000: Reanalyses-based tropospheric temperature estimates: Uncertainties in the context of global climate change detection. J. Climate, 13 , 31873205.

    • Search Google Scholar
    • Export Citation
  • Chelliah, M., , and G. D. Bell, 2004: Tropical multidecadal and interannual climate variability in the NCEP–NCAR reanalysis. J. Climate, 17 , 17771803.

    • Search Google Scholar
    • Export Citation
  • Folland, C. K., , T. N. Palmer, , and D. E. Parker, 1986: Sahel rainfall and worldwide sea temperatures, 1901–85. Nature, 320 , 602607.

  • Giannini, A., , R. Saravanan, , and P. Chang, 2003: Oceanic forcing of Sahel rainfall on interannual to interdecadal time scales. Science, 302 , 10271030.

    • Search Google Scholar
    • Export Citation
  • Greatbatch, R. J., 2000: The North Atlantic Oscillation. Stochast. Environ. Res. Risk Assess, 14 , 213242.

  • Hoerling, M. P., , J. W. Hurrell, , and T. Xu, 2001: Tropical origin for recent North Atlantic climate change. Science, 292 , 9092.

  • Hurrell, J. W., , and C. K. Folland, 2002: A change in the summer atmospheric circulation over the North Atlantic. CLIVAR Exchanges, Vol. 7, No. 3/4, International CLIVAR Project Office, Southampton, United Kingdom, 52–54.

  • Hurrell, J. W., , Y. Kushnir, , G. Ottersen, , and M. Visbeck, 2003: An overview of the North Atlantic Oscillation. The North Atlantic Oscillation, Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 1–35.

  • Inoue, T., , and J. Matsumoto, 2004: A comparison of summer sea level pressure over east Eurasia between NCEP–NCAR reanalysis and ERA-40 for the period 1960–99. J. Meteor. Soc. Japan, 82 , 951958.

    • Search Google Scholar
    • Export Citation
  • Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-Year Reanalysis: Monthly means CD-ROM and documentation. Bull. Amer. Meteor. Soc, 82 , 247268.

    • Search Google Scholar
    • Export Citation
  • Levinson, D. H., , and A. M. Waple, 2004: State of the climate in 2003. Bull. Amer. Meteor. Soc, 85 , S1S72.

  • Lu, J., , R. J. Greatbatch, , and K. A. Peterson, 2004: On the trend in Northern Hemisphere winter atmospheric circulation during the last half of the twentieth century. J. Climate, 17 , 37453760.

    • Search Google Scholar
    • Export Citation
  • Manley, G., 1974: Central England temperatures: Monthly means 1659 to 1973. Quart. J. Roy. Meteor. Soc, 100 , 389405.

  • Pal, J. S., , F. Giorgi, , and X. Q. Bi, 2004: Consistency of recent European summer precipitation trends and extremes with future regional climate projections. Geophys. Res. Lett, 31 .L13202, doi:10.1029/2004GL019836.

    • Search Google Scholar
    • Export Citation
  • Parker, D. E., , L. V. Alexander, , and J. Kennedy, 2004: Global and regional climate in 2003. Weather, 59 , 145152.

  • Parker, D. E., , T. P. Legg, , and C. K. Folland, 1992: A new daily central England temperature series, 1772–1991. Int. J. Climatol, 12 , 317342.

    • Search Google Scholar
    • Export Citation
  • Rodwell, M. J., , and C. K. Folland, 2002: Atlantic air-sea interaction and seasonal predictability. Quart. J. Roy. Meteor. Soc, 128 , 14131443.

    • Search Google Scholar
    • Export Citation
  • Rodwell, M. J., , and C. K. Folland, 2003: Atlantic air-sea interaction and model validation. Ann. Geophys, 46 , 4756.

  • Sterl, A., 2004: On the (in)homogeneity of reanalysis products. J. Climate, 17 , 38663873.

  • Sturaro, G., 2003: A closer look at the climatological discontinuities present in the NCEP–NCAR reanalysis temperature due to the introduction of satellite data. Climate Dyn, 21 , 309316.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., , and D. A. Paolino Jr., 1980: The Northern Hemisphere sea-level pressure data set: Trends, errors and discontinuities. Mon. Wea. Rev, 108 , 855872.

    • Search Google Scholar
    • Export Citation
  • Ward, M. N., 1998: Diagnosis and short-lead time prediction of summer rainfall in tropical North Africa at interannual and multidecadal timescales. J. Climate, 11 , 31673191.

    • Search Google Scholar
    • Export Citation
  • Yang, S., , K-M. Lau, , and K-M. Kim, 2002: Variations of the East Asian jet stream and Asian–Pacific–American winter climate anomalies. J. Climate, 15 , 306325.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

The upper panels show the linear trend in SLP (hPa yr−1) over the period 1958–2002 in (a) the NCEP–NCAR reanalysis, (b) the ERA-40 reanalysis, and (c) Trenberth's SLP dataset. The lower panels show the time series of area mean SLP over (d) Africa and (f) Asia within the enframed areas in the top panels, and for the different datasets. (e) The corresponding time series for central England.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 2.
Fig. 2.

The time series of SLP in the NCEP–NCAR reanalysis over Africa and Asia taken from Fig. 1.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 3.
Fig. 3.

Same as in Fig. 1 except for 500-hPa geopotential height (gpm yr−1), and for the NCEP–NCAR and ERA-40 reanalyses only.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 4.
Fig. 4.

Same as in Fig. 3 except for 2-m temperature (K yr−1).

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 5.
Fig. 5.

The (undetrended) interannual correlation between the different datasets during periods when each pair have overlapping availability (see text for details). Upper panels are for the NCEP–NCAR and ERA-40 reanalyses (a) SLP, (b) 500-hPa geopotential height, and (c) 2-m temperature. The lower panels are for Trenberth's dataset compared with (d) the NCEP–NCAR and (e) the ERA-40 reanalysis.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 6.
Fig. 6.

The upper panels show the linear trend in 500-hPa temperature (K yr−1), 1958–2002, using (a) the NCEP–NCAR and (b) the ERA-40 reanalyses. (c), (d) The corresponding linear trends at 100 hPa. The (undetrended) interannual correlation between the two datasets over the same time period for (e) 500- and (f) 100-hPa temperature.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 7.
Fig. 7.

The average for the entire latitude band from 17.5°N to the equator for (a) 500- and (b) 100-hPa temperature.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 8.
Fig. 8.

The upper panels show the spatial pattern obtained by regressing the respective NAO indices against the summer mean SLP for (a) the NCEP–NCAR reanalysis, (b) the ERA-40 reanalysis, and (c) Trenberth's dataset. Negative contours are dashed and the zero and positive contours are solid. The contour interval is 5 hPa. (d) The corresponding NAO indices (NCEP–NCAR, dashed; ERA-40, solid; Trenberth, dot–dash).

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Fig. 9.
Fig. 9.

(a) The time series of summer mean central England temperature (CET) (solid line) and NAO index (dashed line) from Trenberth's dataset. (b) The sliding window (19-yr width) cross correlation between the two time series in (a). (c) The (interannual) correlation between summer mean SLP and CET over the 19-yr centered at 1921. Dashed contours indicate negative correlations; solid contours the zero line and positive correlations. The contour interval is 0.15 and the 5% significance level is 0.45, indicating correlations significantly different from zero at the 5% level near the regions of maximum positive and negative correlation.

Citation: Journal of Climate 19, 7; 10.1175/JCLI3643.1

Table 1.

Correlation between the summer mean time series of interannually varying SLP from the different datasets over Asia and North Africa for the time periods shown.

Table 1.
Table 2.

As in Table 1 but for 500-hPa height (HGT500) and 2-m temperature (2mT).

Table 2.
Save