- Email: [email protected]
Regional analysis of extreme temperature and precipitation indices for the Carpathian Basin from 1946 to 2001
Global and Planetary Change 57 (2007) 83 – 95 www.elsevier.com/locate/gloplacha
Regional analysis of extreme temperature and precipitation indices for the Carpathian Basin from 1946 to 2001 Judit Bartholy ⁎, Rita Pongrácz Department of Meteorology, Eötvös Loránd University, Pázmány P. st. 1/a, H-1117, Budapest, Hungary Available online 3 January 2007
Abstract Since human and natural systems may be especially affected by changes of extreme climate events, the main objective of our research is to detect the possible changes of intensity and frequency of these extreme events. Several climate extreme indices are analyzed and compared for Central/Eastern Europe (focusing on Hungary) for the 20th century based on the guidelines suggested by the joint WMO-CCl/CLIVAR Working Group on climate change detection. These climate extreme indices include the numbers of severe cold days, winter days, frost days, cold days, warm days, summer days, hot days, extremely hot days, cold nights, warm nights, hot nights, the intra-annual extreme temperature range, the heat wave duration, the number of wet days (using several threshold values defining extremes, i.e., 10 mm and 20 mm), the maximum number of consecutive dry days, the highest 1-day precipitation amount, the greatest 5-day rainfall total, the simple daily precipitation intensity, the numbers of moderate wet days and very wet days, the annual fraction due to extreme precipitation events, etc. Therefore, daily maximum, minimum and mean temperature observations (from 13 stations) and daily precipitation amounts (from 31 stations) are used in the present statistical analysis. Our results suggest that similarly to the global and continental trends, regional temperature of Central/Eastern Europe has become warmer during the second half of the 20th century. Specifically, the strongest increasing tendency is detected in case of the annual numbers of hot days, summer days, warm days, warm nights, and the heat wave duration index. Before this warming period in the last quarter of the 20th century, most of the indices exhibit a cooling period until the middle of the 1970's. Furthermore, regional intensity and frequency of extreme precipitation has increased (as shown by the annual fraction due to extreme precipitation events, and the numbers of moderate wet days, very wet days, and very heavy precipitation days) between 1976 and 2001, while the total precipitation has decreased (as shown by numbers of precipitation days exceeding 0.1 mm, 1 mm, and 5 mm) in the region. © 2006 Elsevier B.V. All rights reserved. Keywords: extreme climate index; daily precipitation; daily temperature; Carpathian Basin; tendency analysis
1. Introduction In the 1998–2002 period, more flood events occurred in the Carpathian Basin (mainly on rivers Tisza and
⁎ Corresponding author. Tel.: +36 1 3722945; fax: +36 1 3722904. E-mail addresses: [email protected] (J. Bartholy), [email protected] (R. Pongrácz). 0921-8181/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.gloplacha.2006.11.002
Danube) than in the previous 30–35 years. For instance, water level at many stations exceeded the previous historical record values during these recent floods (e.g., Autumn, 1998; Spring, 1999, 2000, 2001). Annual maximum water level exceeded the 3rd order flood warning level in 6 years out of the last decade on the headwaters of river Tisza (Bárdossy et al., 2003; Dezső et al., 2005). Because of the severe social and economical consequences of these recent floods, enhanced
84
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
public interest appeared to demand scientific analysis. In order to objectively study the recent climate change, especially, the change of the frequency of the extreme events, extreme precipitation and temperature indices can be used. In 1998, a joint WMO-CCl/CLIVAR Working Group (Karl et al., 1999) formed on climate change detection. One of its task groups aimed to identify the climate extreme indices, and completed a climate extreme analysis on all part of the world where
appropriate data was available (Frich et al., 2002). Some results of this working group also appeared in the third assessment report of IPCC (2001). For the European continent, Klein Tank and Können (2003) accomplished and summarized the extreme climate index analysis. In this paper, similar analyses for 13 temperature indices and 12 precipitation indices are presented for a finer spatial scale, focusing on the Carpathian Basin, located in Central/Eastern Europe. First, the database is briefly
Fig. 1. Geographical locations of meteorological stations in the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
described, then, results of the tendency analysis of extreme climate indices (suggested by the WMOCCl/CLIVAR Working Group), based on daily temperature and daily precipitation, are discussed in Sections 3 and 4, respectively. In the analysis, decadal trend coefficients are determined (using linear regression modeling), which represent the increasing or decreasing rate of the given parameter in 10 years. Hereafter, trend coefficient and tendency are used as equivalent terms. 2. Database For the evaluation of recent tendency of temperature and precipitation extremes in the Carpathian Basin, 13 and 31 meteorological stations have been
85
used (Fig. 1), respectively. Datasets for the 21 Hungarian stations were available from the Hungarian Meteorological Service (HMS) while datasets for the 11 stations in the neighbouring countries are freely available via Internet (Klein Tank, 2003) from the European Climate Assessment Dataset (ECAD). Daily precipitation, maximum, minimum, and mean temperature time series have been collected for the period 1946–2001. Stations have been selected on the base of considering the following general criteria used by the global NOAA NCDC datasets (Peterson and Vose, 1997), or the ECAD (Klein Tank et al., 2002b): (i) from the entire 1946–2001 period, data must be available for at least 40 years, (ii) missing data cannot be more than 10%, (iii) missing data from each year cannot exceed 20%, (iv) more than 3 months consecutive
Table 1 Summary of the trend analysis of extreme temperature indices for the Carpathian Basin based on 13 stations (warming and cooling trends are indicated by black and light grey color of the box, respectively)
Signs in parentheses indicate regional mean coefficients being not significant at 95% level. Numbers in parentheses indicate the stations with trend coefficients being not significant at 95% level. Regional means of the percentile values are as follows: 1.4 (10th percentile of Tmax), 27.8 (90th percentile of Tmax), − 3.6 (10th percentile of Tmin), 16.5 (90th percentile of Tmin).
86
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
missing values are not allowed. Originally, data series from 43 meteorological stations from HMS have been tested for including to our database. Based on the metadata containing the station history, and on the ratio of missing data, the number of Hungarian stations had to be reduced to 21. Inhomogeneity of the daily temperature and precipitation time series originating from the ECAD was analyzed by Wijngaard et al. (2003). In case of temperature, two characteristics of the diurnal temperature range (annual means and day-to-day absolute differences) were evaluated. The results demonstrated that among the 11 meteorological stations, used in our analysis, only station no. 5 (Cluj Napoca, Romania) is ‘suspect’ in both characteristics, while stations no. 1 (Arad, Romania) and no. 10 (Zagreb, Croatia) are ‘suspect’ in one characteristic, and ‘useful’ in the other. In case of precipitation, all the stations selected for our analysis were confirmed ‘useful’, which means that no inhomogeneity could be found. Extreme events (e.g., local floods and droughts, heat waves, local cold spells) often occur on local scale, but on the other hand, they are all important part of largescale climate patterns. However, local extremes could disappear in case of a spatial data interpolation. Therefore, maps presented in this paper use station data instead of gridded database.
3. Tendency analysis of extreme temperature indices On the base of our previous study of time series of mean temperature and extreme temperature parameters, a strong warming tendency was detected from the middle of the 1970's (Pongrácz and Bartholy, 2000). Therefore, the entire 1961–2001 period has been separated into two subperiods, namely, 1961–1975 and 1976–2001. The tendency analysis has been accomplished for these subperiods. The extreme temperature indices used in our analysis are listed and defined in Table 1. Furthermore, Table 1 summarizes the increasing (+) and decreasing (−) tendencies of the indices for the entire 41 years and for the two subperiods (15 and 26 years). The sign of the trend coefficients is not directly indicative of a warming or cooling tendency. For instance, negative coefficients of the number of cold days (Tx10) and positive coefficients of the number of hot days (Tx30GE) both indicate warming climate. Therefore, warming tendencies are shown in black boxes, while cooling tendencies in light grey. The trend coefficients of the index ETR (intra-annual extreme temperature range) are in white since they do not imply either warming or cooling tendency by themselves. The detected regional mean trend coefficients are significant at 95% level, except indices Tn90 (in 1961–1975), FD (in 1976–2001), and Tn20GT (in 1961–2001). These
Fig. 2. Distribution of decadal trend coefficients of extreme temperature indices for the Carpathian Basin. For each index, warming and cooling tendencies are indicated by grey and white background, respectively. In case of ETR, hatched background indicates that the sign of the trend coefficient is not directly indicative of a warming or cooling tendency.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
exceptions are indicated by using parentheses around the sign of the overall regional tendency. Furthermore, numbers in parentheses below the signs, represent the stations where the decadal trend coefficients are not significant at 95% level. In general, all the 13 stations exhibit significant tendency values for each extreme index, more than 2 stations with not significant trend can be detected only in a few cases (ETR in 1961–2001 and 1976–2001, Tx35GE in 1961–2001 and 1961–1975, Tn20GT in 1961–2001 and 1961–1975). Warming tendencies (in black) are dominant in the table. The regional climate of the Carpathian Basin show a warming tendency for most indices when the entire 41 years are considered (except HWDI — heat wave duration index), however cooling tendencies in
87
some indices are found in one of the other subperiod. Note that four extreme indices (Tn10 — number of cold nights, Tn-10LT — number of severe cold days, FD — number of frost days, Tn90 — number of warm nights) indicate warming tendencies both for the 1961–1975 and 1976–2001 subperiods. Based on the trend coefficients of HWDI, Tx90, SU, Tx30GE, Tx35GE, Tn20GT, the cooling tendencies until the middle of the 1970's is followed by a warming climate in the last quarter of the 20th century. Opposite tendency can be detected in case of two indices (Tx10, Tx0LT) using regional scale average. However, these cooling trend coefficients of the last decades are small (the regional means of the decadal trends are 0.39 and 0.36, respectively).
Fig. 3. Tendency of the heat wave duration index (HWDI) in the Carpathian Basin during the last quarter of the 20th century. Trend coefficients greater than 0.4 are significant at 95% level of confidence. The regional mean is calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
88 J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95 Fig. 4. Tendency of the warm nights (Tn90) and warm days (Tx90) in the last quarter of the 20th century. Trend coefficients greater than 0.4 are significant at 95% level of confidence. The regional means are calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
More detailed quantitative information on the trend analysis is presented in Fig. 2, where the distribution of the trend coefficients, determined for each station, can be seen. The Whisker plot diagrams provide statistical characteristics of the decadal trends for each extreme temperature index, for the entire period (1961–2001) and the two subperiods (1961–1975, 1976–2001). Similarly to Table 1, warming and cooling tendencies are indicated by different background colors (grey and white, respectively). In case of the index ETR, hatched background is used since the sign of the trend coefficient is not directly connected to warming or cooling tendency. Due to the limited extent of the paper, tendency analysis of only three indices (HWDI, Tn90, Tx90) are presented in details for the 1976–2001 subperiod. Fig. 3 shows map with the trend coefficients of HWDI. Circles represent decadal trend coefficients of the meteorological stations (using the baseperiod 1961–1990). Black and grey circles indicate increasing and decreasing tendencies, respectively, while circle size depends on the intensity of these positive or negative trends. As it can be seen on the map, all the increasing tendencies of HWDI, detected in the last 26 years, are significant. Detailed tendency analysis of the indices Tn90 (warm nights) and Tx90 (warm days) are presented in Fig. 4. Similarly to Fig. 3, the upper maps provide the trend coefficients of the stations. The lower graphs show the regional mean index anomaly from the 1961–1990
89
average values. The fitted linear trend is clearly increasing in the last quarter of the 20th century in case of both indices. Also, no decreasing tendency can be identified in the upper maps. The positive trend coefficients are significant at 95% level of confidence. Similarly to the global and European trends (Frich et al., 2002; Klein Tank et al., 2002a), analysis of the extreme temperature indices suggests that the regional climate of the Carpathian Basin tended to get warmer in the last 41 years. Based on the results presented in this section, the frequency of extreme temperature values increased considerably in the Carpathian Basin. Although, our database covers the 1961–2001 period, the European summer heat wave of 2003 can be considered as a sign of this warming tendency. The analysis of Schär et al. (2004) demonstrated that year-to-year variability of summer temperature of Europe increased significantly as a response to the increasing greenhouse gas forcing. According to the modeling study of Schär et al. (2004), this extreme summer is likely to represent mean climate of 2071–2100. van Oldenborgh and van Ulden (2003) related the observed temperature increase to the circulation characterized very simply by the local wind. As their results suggest, detected changes in the distribution of wind direction explain most of the interannual variability of temperature. Another study by Domonkos et al. (2003) indicated that changes of the frequency of large-scale circulation patterns over
Table 2 Summary of tendency analyses of extreme precipitation indices for the Carpathian Basin for the 1946–2001 (based on 26 stations) and for the 1976– 2001 (based on 31 stations) periods Nr. Extreme index
Definition [unit]
1946–2001 1976–2001
1
CDD; Consecutive dry days
+; (4)
−; (3)
2 3 4
Rx1; Highest 1-day precipitation amount Rx5; Greatest 5-day rainfall total SDII; Simple daily intensity index
(− +); (4) −; (5) (+); (24)
−; (4) + −; (3) (+); (23)
5
R95T; Fraction of annual total rainfall due to events above the 95th percentile of the daily precipitation in the baseperiod 1961–1990 RR10; Number of heavy precipitation days RR20; Number of very heavy precipitation days R75; Number of moderate wet days
Maximum number of consecutive dry days when Rday b 1 mm [day] Maximum of Rday [mm] Maximum of Σ5Rday [mm] Total precipitation sum/total number of days when Rday ≥ 1 mm [mm/day] ΣRday / Rtotal, where ΣRday is the sum of daily precipitation exceeding R95% [%]
(+); (5)
+; (3)
(− +); (7) (− +); (17) (−); (11)
+; (8) +; (5) +; (6)
(− +); (22)
+; (11)
(−); (10) −; (1) −; (4)
(−); (8) −; (2) −; (1)
6 7 8 9
10 11 12
when Rday ≥ 10 mm [day] when Rday ≥ 20 mm [day] Rday > R75%, where R75% is the upper quartile of the daily precipitation in the baseperiod 1961–1990 [day] R95; Number of very wet days Rday > R95%, where R95% is the 95th percentile of the daily precipitation in the baseperiod 1961–1990 [day] RR5; Number of precipitation days exceeding 5 mm when Rday ≥ 5 mm [day] RR1; Number of precipitation days exceeding 1 mm when Rday ≥ 1 mm [day] RR0.1; Number of precipitation days exceeding 0.1 mm when Rday ≥ 0.1 mm [day]
Signs in parentheses indicate regional mean coefficients being not significant at 95% level. Numbers in parentheses indicate the stations with trend coefficients being not significant at 95% level. Regional means of the 75th and 95th percentile values are 56 and 178, respectively.
90
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
Central Europe (characterized by the Hess and Brezowsky, 1977 classification system) in the 20th century can partly be responsible for the variability of seasonal temperature extreme events (winter extreme cold events and summer extreme warm events). Based on the results of Domonkos et al. (2003), similarly to the global warming tendency, a slight warming trend can be detected in the winter and summer extreme events in the 20th century, but only a few of the 11 stations, involved in their analysis, is statistically significant. Similarly, an overall warming tendency can be detected on the base of our results for the 1961–2001 period, although our analysis has a smaller spatial coverage, a larger station density, and a smaller temporal coverage than those of Domonkos et al. (2003). Furthermore, the trend coefficients, determined in case of the 13 extreme temperature indices, for the 13 stations included in the selected database of our analysis, are mostly significant at 95% level (except the index Tn20GT, number of hot nights). 4. Analysis of extreme precipitation indices Definition of the 12 precipitation extreme indices used in the present analysis can be found in Table 2. Two subperiods have been defined, namely, 1946–2001 and 1976–2001. Results of the tendency analysis are summarized in Table 2 for these two subperiods. When
similar changes are detected for all stations, only one “+” or “−” sign indicates the tendency. In case of different tendencies of the western and eastern parts of the selected region, “− +” or “+ −” signs can be found in the table. The detected regional mean trend coefficients for the 1976–2001 subperiod are significant at 95% level, except indices SDII and RR5. In case of the entire 1946–2001 period, only 4 precipitation indices exhibit significant regional mean tendency, namely, CDD, Rx5, RR1, and RR0.1. Similarly to Table 1, parentheses indicate not significant regional tendency. In addition, numbers in parentheses after the signs, represent the stations where the decadal trend coefficients are not significant at 95% level. Note that the numbers of stations used in the trend analysis are different in case of periods 1946–2001 and 1976–2001 (26 and 31, respectively). Based on the analysis of tendency maps (Bartholy and Pongrácz, 2005), only some of the extreme indices can be characterized by homogeneous positive or negative trends for both periods. Most of the extreme precipitation indices increased considerably in the Carpathian Basin by the end of the 20th century. Positive trends were detected mostly in the last 26 years. The strongest increasing tendencies appear in case of extreme indices indicating very intense or large precipitation (i.e., R95T, RR20, R75, R95). Detailed quantitative information on the trend analysis of precipitation
Fig. 5. Distribution of decadal trend coefficients of extreme precipitation indices for the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
extreme indices can be seen in Fig. 5. The presented Whisker plot diagrams indicate the distribution of the decadal trend coefficients of station data for each extreme index, for the two periods (1946–2001 and 1976–2001). In this paper, three parameters of Table 2 are presented in details. The Carpathian Basin's tendencies of annual number RR20 of heavy precipitation days (when daily precipitation is greater than 20 mm) are shown for the last quarter of the 20th century in Fig. 6. Based on the tendency analysis of the entire European continent (Klein Tank et al., 2002a), heavy precipitation days occurred more often in the last 2–3 decades in northern
91
stations, while they became less frequent in the Mediterranean region. The Carpathian Basin is located inbetween, however, our detailed regional analysis (Fig. 6) suggests that except for a few southern stations, the annual number of heavy precipitation days (RR20) increased during the last 26 years. The entire Carpathian Basin can be characterized by a strong positive trend. Considering only the Hungarian stations, the annual number of wet days exceeding 20 mm increased more in Transdanubia than in the Great Plains. Indices listed in Table 2, include a few precipitationrelated parameters which do not indicate extreme conditions. They belong to the index type annual number of
Fig. 6. Tendency of annual number of very heavy precipitation days exceeding 20 mm (RR20) in the last quarter of the 20th century. Trend coefficients greater than 0.3 in absolute value are significant at 95% level of confidence. The regional mean of the lower graph is calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
92
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
precipitation days exceeding a given threshold; for instance, RR1 is one of them. Decadal tendency of the annual number of wet days with daily precipitation exceeding 1 mm (RR1) is analyzed for the second half of the 20th century (Fig. 7). Similarly to Fig. 6, the upper map provides the spatial distribution of decadal trends for the Carpathian Basin, while the lower graph shows the regional mean time series of the RR1 anomalies (using the baseperiod 1961–1990) for the Carpathian Basin. The small graph in the upper right part illustrates the number of stations used for calculating the spatial average. Decadal tendency of RR1 is strongly
negative in the last 56 years in most of the stations, as well, as in case of the regional mean anomaly. Fig. 8 compares the tendency of annual rainfall fraction due to very wet days (R95T) during the second half and the last quarter of the 20th century. Slight decreasing tendencies can be detected in the Transdanubian stations during 1946–2001, while intermediate positive trends appear in other stations of the region on the left map of the figure. Furthermore, very strong positive trends were found during the last 26 years (shown on the right map) indicating that the annual fraction of total rainfall (Rtotal) due to events above the
Fig. 7. Tendency of annual number of precipitation days exceeding 1 mm (RR1) in the Carpathian Basin during the second half of the 20th century. Trend coefficients greater than 0.3 in absolute value are significant at 95% level of confidence. The regional mean of the lower graph is calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95 Fig. 8. Tendency of fraction of total annual rainfall due to very wet days (R95T). Trend coefficients greater than 0.3 and 0.4 in absolute value are significant at 95% level of confidence on the left and right map, respectively. The regional means of the lower graphs are calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
93
94
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
95th percentile (R95%) of daily precipitation in the baseperiod 1961–1990 (ΣRday / Rtotal, where ΣRday indicates the sum of daily precipitation exceeding R95%), increased significantly between 1976 and 2001. Summarizing the results presented in Figs. 6–8, although in general, precipitation occurred more rarely in the Carpathian Basin, the ratio of heavy or extreme precipitation days increased considerably by the end of the 20th century. Similar tendencies can be expected in many areas of Europe (including the Carpathian Basin) for the last three decades of the 21st century on the base of the modeling studies of Christensen and Christensen (2003). They analyzed the A2 global scenario (IPCC, 2001) for the European continent, and concluded that the number of very wet days (R95) is very likely to increase in many European regions, despite a possible reduction in the summer mean precipitation amount over a substantial part of the continent. Furthermore, the results of Christensen and Christensen (2003) indicated that severe flood episodes are likely to become more frequent by the 2071– 2100 period, despite the overall expected trend towards drier climate conditions in summer. Lloyd-Hughes and Saunders (2002) analyzed two indices (Palmer drought severity index, PDSI; and standardized precipitation index, SPI) characterizing dry and wet periods in Europe for the entire 20th century. According to their results, extreme and moderate drought conditions averaged for the entire continent have changed insignificantly during the 20th century. However, significant changes were detected in Northeastern Europe towards wetter conditions, while significant drying tendencies were observed in Central and Eastern Europe. The strongest and the weakest trends occurred in winter/spring and in summer/autumn, respectively (Lloyd-Hughes and Saunders, 2002). Haylock and Goodess (2004) focused on two of the extreme precipitation indices (CDD — number of consecutive dry days, and R90 — number of precipitation days exceeding the 90th percentile of the daily precipitation using the reference period 1961–1990) and the winter months (DJF) of the 1958–2000 period. According to their results, the North Atlantic Oscillation (NAO) has an important influence on extreme rainfall, and the changes in the NAO caused the observed trends in the selected indices (Haylock and Goodess, 2004). 5. Conclusions Global (Frich et al., 2002) and European (Klein Tank and Können, 2003) scale studies on extreme climate indices indicated that trends of the temperature related indices are consistent with the global warming. In this paper, a spatially more detailed analysis for the
Carpathian Basin illustrates similar warming tendencies for the 1961–2001 period. In case of extreme precipitation indices, trends are spatially less coherent than in case of temperature. Opposite tendencies can be observed in the northern (towards wetter conditions) and southern (towards drier conditions) parts of Europe (e.g., Klein Tank and Können, 2003; Haylock and Goodess, 2004). The Carpathian Basin is located in the boundary area of these two large regions, which justifies why it is necessary to accomplish a detailed regional analysis using data of more stations from this particular region. Based on the analysis of the extreme temperature and precipitation indices (according to the suggestions of the WMO-CCl/CLIVAR Working Group) for the second half of the 20th century, presented in this paper, the following conclusions can be drawn. 1. Results of the analysis of the extreme temperature indices determined for 13 stations located in the Carpathian Basin: (i) Significant warming tendencies are dominant during the entire 1961–2001 period. (ii) In case of most of the indices (e.g., HWDI — heat wave duration, Tx90 — number of warm days, SU — number of summer days, Tx30GE — number of hot days, Tx35GE — number of extremely hot days, Tn20GT — number of hot nights), the entire 41 years can be separated into a cooling period until the middle of the 1970's, and then a warming period in the last quarter of the 20th century. (iii) The largest trend coefficients (more than 6 days per decade) were detected in case of the following indices: Tn90 (number of warm nights), Tx90 (number of warm days), SU (number of summer days), Tx30GE (number of hot days), HWDI (heat wave duration index). 2. Results of the analysis of the extreme precipitation indices determined for 31 stations located in the Carpathian Basin: (i) Strong positive trends were detected in most of the extreme precipitation indices (e.g., R95T — annual fraction due to extreme precipitation events, RR20 — number of very heavy precipitation days, R75 — number of moderate wet days, R95 — number of very wet days) for the last quarter of the 20th century indicating increasing precipitation extremity in the Carpathian Basin. (ii) Significant negative trends dominate the region in case of the non-extreme parameters (i.e., RR5, RR1,
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
and RR0.1 — numbers of precipitation days exceeding 5 mm, 1 mm, and 0.1 mm, respectively) during the second half of the 20th century. (iii) In general, precipitation occurred less frequently in the Carpathian Basin, however, the ratio of heavy or extreme precipitation days increased considerably by the end of the 20th century. Acknowledgements Research leading to this paper has been supported by the Hungarian National Science Research Foundation (OTKA) under grants T-049824, T-034867, and T038423, also by the CHIOTTO project of the European Union Nr. 5 program under grant EVK2-CT-2002/0163, and the Hungarian National Research Development Program under grants NKFP-6/079/2005 and NKFP3A/082/2004, and VAHAVA project of the Hungarian Academy of Sciences and the Ministry of Environment and Water. ESRI software has been used to create maps. References Bárdossy, A., Kontur, I., Stehlik, J., Bálint, G., 2003. Could the global warming cause the last floods of Tisza river? Water Resources Systems — Hydrological Risk Management and Development. IAHS Publ., p. 281. Bartholy, J., Pongrácz, R., 2005. Tendencies of extreme climate indices based on daily precipitation in the Carpathian Basin for the 20th century. Időjárás 109, 1–20. Christensen, J.H., Christensen, O.B., 2003. Severe summertime flooding in Europe. Nature 421, 805–806. Dezső, Zs., Bartholy, J., Pongrácz, R., Barcza, Z., 2005. Analysis of land-use/land-cover change in the Carpathian region based on remote sensing techniques. Phys. Chem. Earth 30, 109–115. Domonkos, P., Kysely, J., Piotrowicz, K., Petrovic, P., Likso, T., 2003. Variability of extreme temperature events in south-central Europe during the 20th century and its relationship with large-scale circulation. Int. J. Climatol. 23, 987–1010. Frich, P., Alexander, L.V., Della-Marta, P., Gleason, B., Haylock, M., Klein Tank, A.M.G., Peterson, T., 2002. Observed coherent
95
changes in climatic extremes during the second half of the twentieth century. Clim. Res. 19, 193–212. Haylock, M.R., Goodess, C.M., 2004. Interannual variability of European extreme winter rainfall and links with mean largescale circulation. Int. J. Climatol. 24, 759–776. Hess, P., Brezowsky, H., 1977. Katalog der Grosswetterlagen. Bericht. Deutsch. Wetterd. Offenb. 113 Bd 15. IPCC, 2001. Climate Change 2001: Third Assessment Report. The Scientific Basis. Cambridge University Press, Cambridge, UK. Karl, T.R., Nicholls, N., Ghazi, A., 1999. Clivar/GCOS/WMO Workshop on Indices and Indicators for Climate Extremes Workshop Summary. Clim. Change 42, 3–7. Klein Tank, A.M.G., 2003. The European Climate Assessment and Dataset project. http://www.knmi.nl/samenw/eca/index.html. Klein Tank, A.M.G., Können, G.P., 2003. Trends in indices of daily temperature and precipitation extremes in Europe, 1946–99. J. Climate 16, 3665–3680. Klein Tank, A.M.G., et al., 2002a. Daily dataset of 20th-century surface air temperature and precipitation series for the European Climate Assessment. Int. J. Climatol. 22, 1441–1453. Klein Tank, A.M.G., Wijngaard, J.B., van Engelen, A., 2002b. Climate of Europe; Assessment of observed daily temperature and precipitation extremes. KNMI, De Bilt, Netherlands. Lloyd-Hughes, B., Saunders, M.A., 2002. A drought climatology for Europe. Int. J. Climatol. 22, 1571–1592. Peterson, T.C., Vose, R.S., 1997. An overview of the global historical climatology network database. Bull. Am. Meteorol. Soc. 78, 2837–2849. Pongrácz, R., Bartholy, J., 2000. Changing trends in climatic extremes in Hungary. In: Kircsi, A. (Ed.), Third Conference on Forest and Climate. Kossuth University Press, Debrecen, pp. 38–44 (in Hungarian with English Abstr.). Schär, C., Vidale, P.L., Lüthi, D., Frei, C., Häberli, C., Liniger, M.A., Appenzeller, C., 2004. The role of increasing temperature variability in European summer heatwaves. Nature 427, 332–336. van Oldenborgh, G.J., van Ulden, A., 2003. On the relationship between global warming, local warming in the Netherlands and changes in circulation in the 20th century. Int. J. Climatol. 23, 1711–1724. Wijngaard, J.B., Klein Tank, A.M.G., Können, G.P., 2003. Homogeneity of 20th century European daily temperature and precipitation series. Int. J. Climatol. 23, 679–692.
Regional analysis of extreme temperature and precipitation indices for the Carpathian Basin from 1946 to 2001 Judit Bartholy ⁎, Rita Pongrácz Department of Meteorology, Eötvös Loránd University, Pázmány P. st. 1/a, H-1117, Budapest, Hungary Available online 3 January 2007
Abstract Since human and natural systems may be especially affected by changes of extreme climate events, the main objective of our research is to detect the possible changes of intensity and frequency of these extreme events. Several climate extreme indices are analyzed and compared for Central/Eastern Europe (focusing on Hungary) for the 20th century based on the guidelines suggested by the joint WMO-CCl/CLIVAR Working Group on climate change detection. These climate extreme indices include the numbers of severe cold days, winter days, frost days, cold days, warm days, summer days, hot days, extremely hot days, cold nights, warm nights, hot nights, the intra-annual extreme temperature range, the heat wave duration, the number of wet days (using several threshold values defining extremes, i.e., 10 mm and 20 mm), the maximum number of consecutive dry days, the highest 1-day precipitation amount, the greatest 5-day rainfall total, the simple daily precipitation intensity, the numbers of moderate wet days and very wet days, the annual fraction due to extreme precipitation events, etc. Therefore, daily maximum, minimum and mean temperature observations (from 13 stations) and daily precipitation amounts (from 31 stations) are used in the present statistical analysis. Our results suggest that similarly to the global and continental trends, regional temperature of Central/Eastern Europe has become warmer during the second half of the 20th century. Specifically, the strongest increasing tendency is detected in case of the annual numbers of hot days, summer days, warm days, warm nights, and the heat wave duration index. Before this warming period in the last quarter of the 20th century, most of the indices exhibit a cooling period until the middle of the 1970's. Furthermore, regional intensity and frequency of extreme precipitation has increased (as shown by the annual fraction due to extreme precipitation events, and the numbers of moderate wet days, very wet days, and very heavy precipitation days) between 1976 and 2001, while the total precipitation has decreased (as shown by numbers of precipitation days exceeding 0.1 mm, 1 mm, and 5 mm) in the region. © 2006 Elsevier B.V. All rights reserved. Keywords: extreme climate index; daily precipitation; daily temperature; Carpathian Basin; tendency analysis
1. Introduction In the 1998–2002 period, more flood events occurred in the Carpathian Basin (mainly on rivers Tisza and
⁎ Corresponding author. Tel.: +36 1 3722945; fax: +36 1 3722904. E-mail addresses: [email protected] (J. Bartholy), [email protected] (R. Pongrácz). 0921-8181/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.gloplacha.2006.11.002
Danube) than in the previous 30–35 years. For instance, water level at many stations exceeded the previous historical record values during these recent floods (e.g., Autumn, 1998; Spring, 1999, 2000, 2001). Annual maximum water level exceeded the 3rd order flood warning level in 6 years out of the last decade on the headwaters of river Tisza (Bárdossy et al., 2003; Dezső et al., 2005). Because of the severe social and economical consequences of these recent floods, enhanced
84
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
public interest appeared to demand scientific analysis. In order to objectively study the recent climate change, especially, the change of the frequency of the extreme events, extreme precipitation and temperature indices can be used. In 1998, a joint WMO-CCl/CLIVAR Working Group (Karl et al., 1999) formed on climate change detection. One of its task groups aimed to identify the climate extreme indices, and completed a climate extreme analysis on all part of the world where
appropriate data was available (Frich et al., 2002). Some results of this working group also appeared in the third assessment report of IPCC (2001). For the European continent, Klein Tank and Können (2003) accomplished and summarized the extreme climate index analysis. In this paper, similar analyses for 13 temperature indices and 12 precipitation indices are presented for a finer spatial scale, focusing on the Carpathian Basin, located in Central/Eastern Europe. First, the database is briefly
Fig. 1. Geographical locations of meteorological stations in the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
described, then, results of the tendency analysis of extreme climate indices (suggested by the WMOCCl/CLIVAR Working Group), based on daily temperature and daily precipitation, are discussed in Sections 3 and 4, respectively. In the analysis, decadal trend coefficients are determined (using linear regression modeling), which represent the increasing or decreasing rate of the given parameter in 10 years. Hereafter, trend coefficient and tendency are used as equivalent terms. 2. Database For the evaluation of recent tendency of temperature and precipitation extremes in the Carpathian Basin, 13 and 31 meteorological stations have been
85
used (Fig. 1), respectively. Datasets for the 21 Hungarian stations were available from the Hungarian Meteorological Service (HMS) while datasets for the 11 stations in the neighbouring countries are freely available via Internet (Klein Tank, 2003) from the European Climate Assessment Dataset (ECAD). Daily precipitation, maximum, minimum, and mean temperature time series have been collected for the period 1946–2001. Stations have been selected on the base of considering the following general criteria used by the global NOAA NCDC datasets (Peterson and Vose, 1997), or the ECAD (Klein Tank et al., 2002b): (i) from the entire 1946–2001 period, data must be available for at least 40 years, (ii) missing data cannot be more than 10%, (iii) missing data from each year cannot exceed 20%, (iv) more than 3 months consecutive
Table 1 Summary of the trend analysis of extreme temperature indices for the Carpathian Basin based on 13 stations (warming and cooling trends are indicated by black and light grey color of the box, respectively)
Signs in parentheses indicate regional mean coefficients being not significant at 95% level. Numbers in parentheses indicate the stations with trend coefficients being not significant at 95% level. Regional means of the percentile values are as follows: 1.4 (10th percentile of Tmax), 27.8 (90th percentile of Tmax), − 3.6 (10th percentile of Tmin), 16.5 (90th percentile of Tmin).
86
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
missing values are not allowed. Originally, data series from 43 meteorological stations from HMS have been tested for including to our database. Based on the metadata containing the station history, and on the ratio of missing data, the number of Hungarian stations had to be reduced to 21. Inhomogeneity of the daily temperature and precipitation time series originating from the ECAD was analyzed by Wijngaard et al. (2003). In case of temperature, two characteristics of the diurnal temperature range (annual means and day-to-day absolute differences) were evaluated. The results demonstrated that among the 11 meteorological stations, used in our analysis, only station no. 5 (Cluj Napoca, Romania) is ‘suspect’ in both characteristics, while stations no. 1 (Arad, Romania) and no. 10 (Zagreb, Croatia) are ‘suspect’ in one characteristic, and ‘useful’ in the other. In case of precipitation, all the stations selected for our analysis were confirmed ‘useful’, which means that no inhomogeneity could be found. Extreme events (e.g., local floods and droughts, heat waves, local cold spells) often occur on local scale, but on the other hand, they are all important part of largescale climate patterns. However, local extremes could disappear in case of a spatial data interpolation. Therefore, maps presented in this paper use station data instead of gridded database.
3. Tendency analysis of extreme temperature indices On the base of our previous study of time series of mean temperature and extreme temperature parameters, a strong warming tendency was detected from the middle of the 1970's (Pongrácz and Bartholy, 2000). Therefore, the entire 1961–2001 period has been separated into two subperiods, namely, 1961–1975 and 1976–2001. The tendency analysis has been accomplished for these subperiods. The extreme temperature indices used in our analysis are listed and defined in Table 1. Furthermore, Table 1 summarizes the increasing (+) and decreasing (−) tendencies of the indices for the entire 41 years and for the two subperiods (15 and 26 years). The sign of the trend coefficients is not directly indicative of a warming or cooling tendency. For instance, negative coefficients of the number of cold days (Tx10) and positive coefficients of the number of hot days (Tx30GE) both indicate warming climate. Therefore, warming tendencies are shown in black boxes, while cooling tendencies in light grey. The trend coefficients of the index ETR (intra-annual extreme temperature range) are in white since they do not imply either warming or cooling tendency by themselves. The detected regional mean trend coefficients are significant at 95% level, except indices Tn90 (in 1961–1975), FD (in 1976–2001), and Tn20GT (in 1961–2001). These
Fig. 2. Distribution of decadal trend coefficients of extreme temperature indices for the Carpathian Basin. For each index, warming and cooling tendencies are indicated by grey and white background, respectively. In case of ETR, hatched background indicates that the sign of the trend coefficient is not directly indicative of a warming or cooling tendency.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
exceptions are indicated by using parentheses around the sign of the overall regional tendency. Furthermore, numbers in parentheses below the signs, represent the stations where the decadal trend coefficients are not significant at 95% level. In general, all the 13 stations exhibit significant tendency values for each extreme index, more than 2 stations with not significant trend can be detected only in a few cases (ETR in 1961–2001 and 1976–2001, Tx35GE in 1961–2001 and 1961–1975, Tn20GT in 1961–2001 and 1961–1975). Warming tendencies (in black) are dominant in the table. The regional climate of the Carpathian Basin show a warming tendency for most indices when the entire 41 years are considered (except HWDI — heat wave duration index), however cooling tendencies in
87
some indices are found in one of the other subperiod. Note that four extreme indices (Tn10 — number of cold nights, Tn-10LT — number of severe cold days, FD — number of frost days, Tn90 — number of warm nights) indicate warming tendencies both for the 1961–1975 and 1976–2001 subperiods. Based on the trend coefficients of HWDI, Tx90, SU, Tx30GE, Tx35GE, Tn20GT, the cooling tendencies until the middle of the 1970's is followed by a warming climate in the last quarter of the 20th century. Opposite tendency can be detected in case of two indices (Tx10, Tx0LT) using regional scale average. However, these cooling trend coefficients of the last decades are small (the regional means of the decadal trends are 0.39 and 0.36, respectively).
Fig. 3. Tendency of the heat wave duration index (HWDI) in the Carpathian Basin during the last quarter of the 20th century. Trend coefficients greater than 0.4 are significant at 95% level of confidence. The regional mean is calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
88 J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95 Fig. 4. Tendency of the warm nights (Tn90) and warm days (Tx90) in the last quarter of the 20th century. Trend coefficients greater than 0.4 are significant at 95% level of confidence. The regional means are calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
More detailed quantitative information on the trend analysis is presented in Fig. 2, where the distribution of the trend coefficients, determined for each station, can be seen. The Whisker plot diagrams provide statistical characteristics of the decadal trends for each extreme temperature index, for the entire period (1961–2001) and the two subperiods (1961–1975, 1976–2001). Similarly to Table 1, warming and cooling tendencies are indicated by different background colors (grey and white, respectively). In case of the index ETR, hatched background is used since the sign of the trend coefficient is not directly connected to warming or cooling tendency. Due to the limited extent of the paper, tendency analysis of only three indices (HWDI, Tn90, Tx90) are presented in details for the 1976–2001 subperiod. Fig. 3 shows map with the trend coefficients of HWDI. Circles represent decadal trend coefficients of the meteorological stations (using the baseperiod 1961–1990). Black and grey circles indicate increasing and decreasing tendencies, respectively, while circle size depends on the intensity of these positive or negative trends. As it can be seen on the map, all the increasing tendencies of HWDI, detected in the last 26 years, are significant. Detailed tendency analysis of the indices Tn90 (warm nights) and Tx90 (warm days) are presented in Fig. 4. Similarly to Fig. 3, the upper maps provide the trend coefficients of the stations. The lower graphs show the regional mean index anomaly from the 1961–1990
89
average values. The fitted linear trend is clearly increasing in the last quarter of the 20th century in case of both indices. Also, no decreasing tendency can be identified in the upper maps. The positive trend coefficients are significant at 95% level of confidence. Similarly to the global and European trends (Frich et al., 2002; Klein Tank et al., 2002a), analysis of the extreme temperature indices suggests that the regional climate of the Carpathian Basin tended to get warmer in the last 41 years. Based on the results presented in this section, the frequency of extreme temperature values increased considerably in the Carpathian Basin. Although, our database covers the 1961–2001 period, the European summer heat wave of 2003 can be considered as a sign of this warming tendency. The analysis of Schär et al. (2004) demonstrated that year-to-year variability of summer temperature of Europe increased significantly as a response to the increasing greenhouse gas forcing. According to the modeling study of Schär et al. (2004), this extreme summer is likely to represent mean climate of 2071–2100. van Oldenborgh and van Ulden (2003) related the observed temperature increase to the circulation characterized very simply by the local wind. As their results suggest, detected changes in the distribution of wind direction explain most of the interannual variability of temperature. Another study by Domonkos et al. (2003) indicated that changes of the frequency of large-scale circulation patterns over
Table 2 Summary of tendency analyses of extreme precipitation indices for the Carpathian Basin for the 1946–2001 (based on 26 stations) and for the 1976– 2001 (based on 31 stations) periods Nr. Extreme index
Definition [unit]
1946–2001 1976–2001
1
CDD; Consecutive dry days
+; (4)
−; (3)
2 3 4
Rx1; Highest 1-day precipitation amount Rx5; Greatest 5-day rainfall total SDII; Simple daily intensity index
(− +); (4) −; (5) (+); (24)
−; (4) + −; (3) (+); (23)
5
R95T; Fraction of annual total rainfall due to events above the 95th percentile of the daily precipitation in the baseperiod 1961–1990 RR10; Number of heavy precipitation days RR20; Number of very heavy precipitation days R75; Number of moderate wet days
Maximum number of consecutive dry days when Rday b 1 mm [day] Maximum of Rday [mm] Maximum of Σ5Rday [mm] Total precipitation sum/total number of days when Rday ≥ 1 mm [mm/day] ΣRday / Rtotal, where ΣRday is the sum of daily precipitation exceeding R95% [%]
(+); (5)
+; (3)
(− +); (7) (− +); (17) (−); (11)
+; (8) +; (5) +; (6)
(− +); (22)
+; (11)
(−); (10) −; (1) −; (4)
(−); (8) −; (2) −; (1)
6 7 8 9
10 11 12
when Rday ≥ 10 mm [day] when Rday ≥ 20 mm [day] Rday > R75%, where R75% is the upper quartile of the daily precipitation in the baseperiod 1961–1990 [day] R95; Number of very wet days Rday > R95%, where R95% is the 95th percentile of the daily precipitation in the baseperiod 1961–1990 [day] RR5; Number of precipitation days exceeding 5 mm when Rday ≥ 5 mm [day] RR1; Number of precipitation days exceeding 1 mm when Rday ≥ 1 mm [day] RR0.1; Number of precipitation days exceeding 0.1 mm when Rday ≥ 0.1 mm [day]
Signs in parentheses indicate regional mean coefficients being not significant at 95% level. Numbers in parentheses indicate the stations with trend coefficients being not significant at 95% level. Regional means of the 75th and 95th percentile values are 56 and 178, respectively.
90
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
Central Europe (characterized by the Hess and Brezowsky, 1977 classification system) in the 20th century can partly be responsible for the variability of seasonal temperature extreme events (winter extreme cold events and summer extreme warm events). Based on the results of Domonkos et al. (2003), similarly to the global warming tendency, a slight warming trend can be detected in the winter and summer extreme events in the 20th century, but only a few of the 11 stations, involved in their analysis, is statistically significant. Similarly, an overall warming tendency can be detected on the base of our results for the 1961–2001 period, although our analysis has a smaller spatial coverage, a larger station density, and a smaller temporal coverage than those of Domonkos et al. (2003). Furthermore, the trend coefficients, determined in case of the 13 extreme temperature indices, for the 13 stations included in the selected database of our analysis, are mostly significant at 95% level (except the index Tn20GT, number of hot nights). 4. Analysis of extreme precipitation indices Definition of the 12 precipitation extreme indices used in the present analysis can be found in Table 2. Two subperiods have been defined, namely, 1946–2001 and 1976–2001. Results of the tendency analysis are summarized in Table 2 for these two subperiods. When
similar changes are detected for all stations, only one “+” or “−” sign indicates the tendency. In case of different tendencies of the western and eastern parts of the selected region, “− +” or “+ −” signs can be found in the table. The detected regional mean trend coefficients for the 1976–2001 subperiod are significant at 95% level, except indices SDII and RR5. In case of the entire 1946–2001 period, only 4 precipitation indices exhibit significant regional mean tendency, namely, CDD, Rx5, RR1, and RR0.1. Similarly to Table 1, parentheses indicate not significant regional tendency. In addition, numbers in parentheses after the signs, represent the stations where the decadal trend coefficients are not significant at 95% level. Note that the numbers of stations used in the trend analysis are different in case of periods 1946–2001 and 1976–2001 (26 and 31, respectively). Based on the analysis of tendency maps (Bartholy and Pongrácz, 2005), only some of the extreme indices can be characterized by homogeneous positive or negative trends for both periods. Most of the extreme precipitation indices increased considerably in the Carpathian Basin by the end of the 20th century. Positive trends were detected mostly in the last 26 years. The strongest increasing tendencies appear in case of extreme indices indicating very intense or large precipitation (i.e., R95T, RR20, R75, R95). Detailed quantitative information on the trend analysis of precipitation
Fig. 5. Distribution of decadal trend coefficients of extreme precipitation indices for the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
extreme indices can be seen in Fig. 5. The presented Whisker plot diagrams indicate the distribution of the decadal trend coefficients of station data for each extreme index, for the two periods (1946–2001 and 1976–2001). In this paper, three parameters of Table 2 are presented in details. The Carpathian Basin's tendencies of annual number RR20 of heavy precipitation days (when daily precipitation is greater than 20 mm) are shown for the last quarter of the 20th century in Fig. 6. Based on the tendency analysis of the entire European continent (Klein Tank et al., 2002a), heavy precipitation days occurred more often in the last 2–3 decades in northern
91
stations, while they became less frequent in the Mediterranean region. The Carpathian Basin is located inbetween, however, our detailed regional analysis (Fig. 6) suggests that except for a few southern stations, the annual number of heavy precipitation days (RR20) increased during the last 26 years. The entire Carpathian Basin can be characterized by a strong positive trend. Considering only the Hungarian stations, the annual number of wet days exceeding 20 mm increased more in Transdanubia than in the Great Plains. Indices listed in Table 2, include a few precipitationrelated parameters which do not indicate extreme conditions. They belong to the index type annual number of
Fig. 6. Tendency of annual number of very heavy precipitation days exceeding 20 mm (RR20) in the last quarter of the 20th century. Trend coefficients greater than 0.3 in absolute value are significant at 95% level of confidence. The regional mean of the lower graph is calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
92
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
precipitation days exceeding a given threshold; for instance, RR1 is one of them. Decadal tendency of the annual number of wet days with daily precipitation exceeding 1 mm (RR1) is analyzed for the second half of the 20th century (Fig. 7). Similarly to Fig. 6, the upper map provides the spatial distribution of decadal trends for the Carpathian Basin, while the lower graph shows the regional mean time series of the RR1 anomalies (using the baseperiod 1961–1990) for the Carpathian Basin. The small graph in the upper right part illustrates the number of stations used for calculating the spatial average. Decadal tendency of RR1 is strongly
negative in the last 56 years in most of the stations, as well, as in case of the regional mean anomaly. Fig. 8 compares the tendency of annual rainfall fraction due to very wet days (R95T) during the second half and the last quarter of the 20th century. Slight decreasing tendencies can be detected in the Transdanubian stations during 1946–2001, while intermediate positive trends appear in other stations of the region on the left map of the figure. Furthermore, very strong positive trends were found during the last 26 years (shown on the right map) indicating that the annual fraction of total rainfall (Rtotal) due to events above the
Fig. 7. Tendency of annual number of precipitation days exceeding 1 mm (RR1) in the Carpathian Basin during the second half of the 20th century. Trend coefficients greater than 0.3 in absolute value are significant at 95% level of confidence. The regional mean of the lower graph is calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95 Fig. 8. Tendency of fraction of total annual rainfall due to very wet days (R95T). Trend coefficients greater than 0.3 and 0.4 in absolute value are significant at 95% level of confidence on the left and right map, respectively. The regional means of the lower graphs are calculated as the mean trend coefficient using all stations located in the Carpathian Basin.
93
94
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
95th percentile (R95%) of daily precipitation in the baseperiod 1961–1990 (ΣRday / Rtotal, where ΣRday indicates the sum of daily precipitation exceeding R95%), increased significantly between 1976 and 2001. Summarizing the results presented in Figs. 6–8, although in general, precipitation occurred more rarely in the Carpathian Basin, the ratio of heavy or extreme precipitation days increased considerably by the end of the 20th century. Similar tendencies can be expected in many areas of Europe (including the Carpathian Basin) for the last three decades of the 21st century on the base of the modeling studies of Christensen and Christensen (2003). They analyzed the A2 global scenario (IPCC, 2001) for the European continent, and concluded that the number of very wet days (R95) is very likely to increase in many European regions, despite a possible reduction in the summer mean precipitation amount over a substantial part of the continent. Furthermore, the results of Christensen and Christensen (2003) indicated that severe flood episodes are likely to become more frequent by the 2071– 2100 period, despite the overall expected trend towards drier climate conditions in summer. Lloyd-Hughes and Saunders (2002) analyzed two indices (Palmer drought severity index, PDSI; and standardized precipitation index, SPI) characterizing dry and wet periods in Europe for the entire 20th century. According to their results, extreme and moderate drought conditions averaged for the entire continent have changed insignificantly during the 20th century. However, significant changes were detected in Northeastern Europe towards wetter conditions, while significant drying tendencies were observed in Central and Eastern Europe. The strongest and the weakest trends occurred in winter/spring and in summer/autumn, respectively (Lloyd-Hughes and Saunders, 2002). Haylock and Goodess (2004) focused on two of the extreme precipitation indices (CDD — number of consecutive dry days, and R90 — number of precipitation days exceeding the 90th percentile of the daily precipitation using the reference period 1961–1990) and the winter months (DJF) of the 1958–2000 period. According to their results, the North Atlantic Oscillation (NAO) has an important influence on extreme rainfall, and the changes in the NAO caused the observed trends in the selected indices (Haylock and Goodess, 2004). 5. Conclusions Global (Frich et al., 2002) and European (Klein Tank and Können, 2003) scale studies on extreme climate indices indicated that trends of the temperature related indices are consistent with the global warming. In this paper, a spatially more detailed analysis for the
Carpathian Basin illustrates similar warming tendencies for the 1961–2001 period. In case of extreme precipitation indices, trends are spatially less coherent than in case of temperature. Opposite tendencies can be observed in the northern (towards wetter conditions) and southern (towards drier conditions) parts of Europe (e.g., Klein Tank and Können, 2003; Haylock and Goodess, 2004). The Carpathian Basin is located in the boundary area of these two large regions, which justifies why it is necessary to accomplish a detailed regional analysis using data of more stations from this particular region. Based on the analysis of the extreme temperature and precipitation indices (according to the suggestions of the WMO-CCl/CLIVAR Working Group) for the second half of the 20th century, presented in this paper, the following conclusions can be drawn. 1. Results of the analysis of the extreme temperature indices determined for 13 stations located in the Carpathian Basin: (i) Significant warming tendencies are dominant during the entire 1961–2001 period. (ii) In case of most of the indices (e.g., HWDI — heat wave duration, Tx90 — number of warm days, SU — number of summer days, Tx30GE — number of hot days, Tx35GE — number of extremely hot days, Tn20GT — number of hot nights), the entire 41 years can be separated into a cooling period until the middle of the 1970's, and then a warming period in the last quarter of the 20th century. (iii) The largest trend coefficients (more than 6 days per decade) were detected in case of the following indices: Tn90 (number of warm nights), Tx90 (number of warm days), SU (number of summer days), Tx30GE (number of hot days), HWDI (heat wave duration index). 2. Results of the analysis of the extreme precipitation indices determined for 31 stations located in the Carpathian Basin: (i) Strong positive trends were detected in most of the extreme precipitation indices (e.g., R95T — annual fraction due to extreme precipitation events, RR20 — number of very heavy precipitation days, R75 — number of moderate wet days, R95 — number of very wet days) for the last quarter of the 20th century indicating increasing precipitation extremity in the Carpathian Basin. (ii) Significant negative trends dominate the region in case of the non-extreme parameters (i.e., RR5, RR1,
J. Bartholy, R. Pongrácz / Global and Planetary Change 57 (2007) 83–95
and RR0.1 — numbers of precipitation days exceeding 5 mm, 1 mm, and 0.1 mm, respectively) during the second half of the 20th century. (iii) In general, precipitation occurred less frequently in the Carpathian Basin, however, the ratio of heavy or extreme precipitation days increased considerably by the end of the 20th century. Acknowledgements Research leading to this paper has been supported by the Hungarian National Science Research Foundation (OTKA) under grants T-049824, T-034867, and T038423, also by the CHIOTTO project of the European Union Nr. 5 program under grant EVK2-CT-2002/0163, and the Hungarian National Research Development Program under grants NKFP-6/079/2005 and NKFP3A/082/2004, and VAHAVA project of the Hungarian Academy of Sciences and the Ministry of Environment and Water. ESRI software has been used to create maps. References Bárdossy, A., Kontur, I., Stehlik, J., Bálint, G., 2003. Could the global warming cause the last floods of Tisza river? Water Resources Systems — Hydrological Risk Management and Development. IAHS Publ., p. 281. Bartholy, J., Pongrácz, R., 2005. Tendencies of extreme climate indices based on daily precipitation in the Carpathian Basin for the 20th century. Időjárás 109, 1–20. Christensen, J.H., Christensen, O.B., 2003. Severe summertime flooding in Europe. Nature 421, 805–806. Dezső, Zs., Bartholy, J., Pongrácz, R., Barcza, Z., 2005. Analysis of land-use/land-cover change in the Carpathian region based on remote sensing techniques. Phys. Chem. Earth 30, 109–115. Domonkos, P., Kysely, J., Piotrowicz, K., Petrovic, P., Likso, T., 2003. Variability of extreme temperature events in south-central Europe during the 20th century and its relationship with large-scale circulation. Int. J. Climatol. 23, 987–1010. Frich, P., Alexander, L.V., Della-Marta, P., Gleason, B., Haylock, M., Klein Tank, A.M.G., Peterson, T., 2002. Observed coherent
95
changes in climatic extremes during the second half of the twentieth century. Clim. Res. 19, 193–212. Haylock, M.R., Goodess, C.M., 2004. Interannual variability of European extreme winter rainfall and links with mean largescale circulation. Int. J. Climatol. 24, 759–776. Hess, P., Brezowsky, H., 1977. Katalog der Grosswetterlagen. Bericht. Deutsch. Wetterd. Offenb. 113 Bd 15. IPCC, 2001. Climate Change 2001: Third Assessment Report. The Scientific Basis. Cambridge University Press, Cambridge, UK. Karl, T.R., Nicholls, N., Ghazi, A., 1999. Clivar/GCOS/WMO Workshop on Indices and Indicators for Climate Extremes Workshop Summary. Clim. Change 42, 3–7. Klein Tank, A.M.G., 2003. The European Climate Assessment and Dataset project. http://www.knmi.nl/samenw/eca/index.html. Klein Tank, A.M.G., Können, G.P., 2003. Trends in indices of daily temperature and precipitation extremes in Europe, 1946–99. J. Climate 16, 3665–3680. Klein Tank, A.M.G., et al., 2002a. Daily dataset of 20th-century surface air temperature and precipitation series for the European Climate Assessment. Int. J. Climatol. 22, 1441–1453. Klein Tank, A.M.G., Wijngaard, J.B., van Engelen, A., 2002b. Climate of Europe; Assessment of observed daily temperature and precipitation extremes. KNMI, De Bilt, Netherlands. Lloyd-Hughes, B., Saunders, M.A., 2002. A drought climatology for Europe. Int. J. Climatol. 22, 1571–1592. Peterson, T.C., Vose, R.S., 1997. An overview of the global historical climatology network database. Bull. Am. Meteorol. Soc. 78, 2837–2849. Pongrácz, R., Bartholy, J., 2000. Changing trends in climatic extremes in Hungary. In: Kircsi, A. (Ed.), Third Conference on Forest and Climate. Kossuth University Press, Debrecen, pp. 38–44 (in Hungarian with English Abstr.). Schär, C., Vidale, P.L., Lüthi, D., Frei, C., Häberli, C., Liniger, M.A., Appenzeller, C., 2004. The role of increasing temperature variability in European summer heatwaves. Nature 427, 332–336. van Oldenborgh, G.J., van Ulden, A., 2003. On the relationship between global warming, local warming in the Netherlands and changes in circulation in the 20th century. Int. J. Climatol. 23, 1711–1724. Wijngaard, J.B., Klein Tank, A.M.G., Können, G.P., 2003. Homogeneity of 20th century European daily temperature and precipitation series. Int. J. Climatol. 23, 679–692.