2 State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing 100081;
3 Jiangsu Collaborative Innovation Center for Climate Change, Nanjing 210008;
4 Institute of Arid Meteorology, China Meteorological Administration, Lanzhou 730020;
5 Beijing Climate Center, Beijing 100081;
6 Yunnan Climate Center, Kunming 650034
Droughts are major meteorological events withlong duration and wide-ranging effects(Zhang, 2005). Increases in the frequency of droughts and disastersassociated with droughts under global warming havecaused great concern. Huang et al. (2006)analyzeddecadal variations in the characteristics of droughts and floods in China, and identified an increase in thefrequency of droughts in North China after the late1970s. Some studies have suggested that the area affectedby droughts has increased throughout Chinaduring the past half-century, although these increasesare only statistically significant in North, Northeast, and Northwest China(Zhang et al., 2003; Zhai and Zou, 2005; Ma and Fu, 2006; Zou and Zhang, 2008; Yang et al., 2013).
Droughts in Southwest China have attracted great concern. A number of studies have examined thecharacteristics of droughts in this region. Shi et al. (2002, 2003)showed that the climate in NorthwestChina was predominantly warm and dry from theend of the little ice age until the 1980s, but transitionedto warm and wet in 1987. Using the st and ardizedprecipitation index, Qi et al. (2006)foundthat droughts had become more frequent in SouthwestChina, the western part of Northwest China, and thenorthern part of North China over the last 60 years. Ma et al. (2006)analyzed the characteristics of climatechange in Southwest China. They identified asignificant warming trend over the Yungui Plateau, aswell as decreases in precipitation in the Sichuan basin and the hilly regions of Guizhou. Peng et al. (2009)reported that droughts in Yunnan are most intensebetween January and March. Two recent studies revealedthat precipitation has decreased and temperaturehas increased over Yunnan since the 1950s(Cheng and Xie, 2008; Liu et al., 2010). Liu et al. (2012)revealedthat the frequency of droughts increased significantlyin the hilly regions of Yunnan, Guizhou, and Guangxi between 1961 and 2009. Severe droughts havebeen particularly prevalent in Southwest China duringthe past 10 years, with a severe drought acrossSouthwest China in the summer of 2003, a rare earlyspringdrought in Yunnan in 2005, an extreme summerdrought in Sichuan and Chongqing in 2006(Li et al., 2011), and a long period of extreme drought in SouthwestChina from autumn of 2009 through spring of2010. These trends have been especially pronouncedsince 2000.
Almost all of the elements or drought indices usedin the above studies have been calculated at individualpoints(observing sites or gridded values). Droughtsare typically regional phenomena with distinct durations and areas of impact. A few recent studies havefocused on regional drought events. Andreadis et al. (2005)analyzed the characteristics of regional droughtevents in the United States during the 20th centuryusing monthly soil moisture data. Qian et al. (2011)studied regional drought events in China. Their resultsindicated that droughts in South and SouthwestChina generally occur during winter, while droughts innorthern China occur more frequently during spring. Ren et al. (2012)proposed an objective identificationtechnique for regional extreme event(OITREE). This method has been used to identify a number ofdifferent regional extreme events in China, such asdroughts, heavy precipitation events, and temperatureextreme events. Cui(2010)analyzed regionalmeteorological drought events in China over the past50 years. Regional drought events occur most frequently and with greatest intensity in North China, but Southwest China is also seriously influenced byregional droughts(see Fig. 1). Previous studiesof meteorological droughts in Southwest China havefocused mainly on individual drought events. Thisstudy employs the new method proposed by Ren etal. (2012)to study Southwest China regional meteorologicaldrought events between 1960 and 2010.2. Data and method2. 1 Data
According to the Grade-I China MeteorologicalGeographic Zoning(CMA Forecast and MitigationDivision, 2006), Southwest China includes Yunnan, Sichuan, and Guizhou provinces, and Chongqing Municipality.
The daily composite-drought index(CI)computedby the National Climate Center are adopted inthis study. The underlying measurements are taken at101 stations in Southwest China between 1 January1960 and 31 December 2010. Zhang et al. (2006)havedescribed the method for calculating CI in detail. 2. 2 Method
The OITREE method(Ren et al., 2012)is appliedin this study. The technique includes five steps. (1)Adaily index is selected for each individual station; (2)the region is partitioned into daily abnormality belts; (3)the temporal continuity of each event is verified; (4)a st and ard index for regional events is constructed; and (5)the extremity of each regional event is assessed.
Steps(2) and (3)in this process are the most important. The partitioning of daily abnormality beltsstarts from a structural analysis of the daily distributionof abnormality, which can then be separated intoa set of distinct daily abnormality belts. The temporalcontinuity of each event is determined by analyzing thespatial overlap of the natural daily abnormality beltson adjacent dates. The index system(Table 1), whichis developed based on the features of regional events, includes five independent indices(extreme intensity, accumulated intensity, accumulated area, maximumimpacted area, and duration) and one integratedindex.
|Note: These indices are classified into three grades. Grade I indices are calculated for the entire event, grade II indices providedaily extreme values, and grade III indices provide station extreme values. K is the duration of the event in days and J is thenumber of stations impacted by the event, while Jk is the number of stations impacted on day k, Sk is the spatial distribution ofthe impacted stations and Area(Sk)is the area covered by Sk. Tki is the value of the daily index for station i on day k, and Tki|cis the threshold value of that index. Tkj and Tkj|c are similar to Tki and Tki|c, but are calculated for each of the stations impactedby the event. I1, I2, As, Am, and D are individual indices. After Ren et al. (2012).|
The values of the OITREE parameters are listedin Table 2. These parameters are determined by analyzinga set of Southwest China regional meteorologicaldrought events, and can be summarized as follows. CI is the daily drought index for individual stations. This parameter is required for step(1)as describedabove. Step(2)requires four parameters. The thresholdfor abnormality is set to the threshold CI for moderatedrought(–1. 2), the threshold distance for definingneighboring sites is set to 250 km, the thresholdfor the minimum number of sites in a daily droughtbelt is set to 11, and the threshold ratio for overlappingsites in a daily drought belt is set to 0. 3. Forthe third step, the threshold for the number of daysallowed for a gap in an event is set to zero, so that allevents are continuous.
|Note: I%1, I%2, A%s, A%m, and D% are st and ardized I1, I2, As, Am, and D, respectively, and e1, e2, e3, e4, and e5 are theircoefficients.|
The fourth step requires a number of parameters. The direction indicator is set to –1 because of the natureof CI. Ren et al. (2012)cautioned that the fiveweighting coefficients used to calculate the integratedindex Z must be determined objectively and carefully. Here, we use the following objective method to calculatethese five coefficients. First, the absolute valuesof the st and ardized drought indices are ranked in descendingorder. Second, the ratio of the sum of the top10% percentile of values to the sum of all values in eachseries is calculated. Third, the five coefficients are calculatedby multiplying the direction indicator for eachindex and the ratio for each series divided by the sumof the five ratios. The five coefficients obtained usingthis approach are –0. 06, –0. 32, 0. 28, 0. 10, and 0. 24, respectively. More than 200 events with durations of10 days or longer are identified during 1960–2010.
Step(5)involves a comparison of different regional drought events. Figure 2 shows the frequencydistribution of different values of the integrated indexZ. Small values of Z(Z < 0. 3, corresponding toweak events with small impacted areas)occur with thegreatest frequency. The threshold for defining a SouthwestChina regional meteorological drought event istherefore set to 0. 3. A total of 87 Southwest Chinaregional meteorological drought events are identified. Following Ren et al. (2012), these 87 events are dividedinto extreme droughts(9 events, about 10% ofthe total), severe droughts(17 events, about 20%), moderate droughts(35 events, about 40%) and slightdroughts(26 events, about 30%). The threshold valuesof the integrated index Z for distinguishing thesefour classes of drought events are 3. 63, 1. 72, and 0. 70, respectively.3. Examples of the identified drought events
According to the integrated intensity, the threestrongest drought events between 1960 and 2010 areidentified as follows. The first strongest event lastedfor 231 days from September 2009 through April 2010 and covered most of Southwest China. The secondstrongest event lasted for 217 days from November1962 through June 1963. The third strongest eventlasted for 134 days from January to June 1969, and influenced all of Southwest China.
Previous studies have shown that these three extremedrought events were associated with seriouseconomic losses(Ding, 2008; Zhang et al., 2009). The methodology also identifies other recent severedrought events, such as the serious autumn droughtin Yunnan during 2003(number 15) and the extremedrought that occurred in Sichuan and Chongqing during2006(number 14).
Figure 3 shows a variety of indices calculated foran extreme meteorological drought event that persistedfrom 4 September 2009 to 22 April 2010. Themaximum area impacted by this event is 118. 4 × 104km2. Most regions of Yunnan, western Guizhou, and southern Sichuan are seriously affected(Fig. 3a). Thedaily variations of integrated intensity, extreme intensity, accumulated intensity, and impacted area arelargely consistent(Figs. 3b–e). The drought eventbegan in September 2009 and strengthened quicklyduring October before reaching its first peak in intensityaround the mid November. The drought thenentered a relatively weak phase until January 2010, beforestrengthening again and reaching a second peakin February and March 2010. The drought eased and ended in April 2010. These results are well consistentwith the drought monitorings.4. Analysis of results4. 1 Temporal characteristics
Figure 4 shows frequency distributions of the duration and maximum impacted area of the 87 SouthwestChina regional meteorological drought events. These events generally last for 10–80 days, with a peakat 30–40 days. The frequency of the events that lastlonger than 80 days is very small. Only three eventslast for more than 200 days, with the longest being231 days. All of these long-duration events are also extremedrought events, ranking first, second, and fourthout of the 87 events. The impacted area of these eventsvaries between 30×104 and 130×104 km2, with a peakat 70×104–100×104 km2.
Regional drought events generally start betweenlate autumn and early spring and end before mid summer(figure omitted). Figure 5 shows seasonal variationsof the regional meteorological drought frequency. Droughts are most frequent during November–April, and relatively infrequent within May–October. Wetherefore define November to April as the droughtseason and May to October as the non-droughtseason.
Figure 6 shows interannual variations of SouthwestChina regional meteorological drought event indices. The frequency of droughts has increased at arate of 0. 19 per decade(statistically significant at the0. 05 level according to Student’s t test). The largestnumber of droughts occurs in 1994(5 times), 1992(4times), and 1988(4 times). No droughts occur in 1965or 1967. Although the frequencies in 2010 and 1963 areboth listed as 0, both years include the drought eventsthat started the year before. The frequency statisticsonly account for the start time of drought events. Although accumulated integrated intensity(Fig. 6b) and accumulated maximum impacted area(Fig. 6c)both increase during 1960–2010, these increases arenot statistically significant. The top three values ofaccumulated integrated intensity occur in 2009, 1978, and 1962. The top three values of accumulated maximumimpacted area occur in 1994, 1992, and 1988. The top three years in terms of frequency are thesame as those for accumulated maximum impactedarea, but are completely different from those for accumulatedintegrated intensity. Severe drought eventsoften last for a long time, so the frequency is low duringthe years within which they occur(such as 2009). Weak droughts generally last for shorter periods and are therefore more likely to repeat during a single year(such as 1994).
The interannual variations of precipitation and mean temperature between 1960 and 2010(Fig. 7)are analyzed to explore the reasons why regional meteorologicaldrought events have become stronger and more frequent in Southwest China. Annual precipitationhas decreased significantly at a rate of 12. 63 mmper decade, while annual mean temperature has increasedat a rate of 0. 15℃ per decade.
The correlations between the time series of annualprecipitation(Fig. 7a) and the time series of droughtfrequency and accumulated integrated intensity(Figs. 6a and 6b)are –0. 44 and –0. 53, respectively. Both valuesare statistically significant at the 0. 01 level basedon Student’s t test. The correlations between the timeseries of mean temperature(Fig. 7b) and the time seriesof drought frequency and accumulated integratedintensity(Figs. 6a and 6b)are 0. 19 and 0. 25, respectively. This result suggests that the main reason whydroughts have become stronger and more frequent liesin the significant decrease in precipitation, althoughthe increase in temperature also contributes. Analysisof the accumulated integrated intensity on decadaltimescales(figure omitted)indicates that the intensityof droughts in the early 21st century has been significantlystronger than the intensity of droughts in otherperiods. The average annual precipitation is smallerafter 2000 than during other periods in the past 50years(Fig. 7), while the temperature is much higher. These may be important factors in droughts becomingmore serious in the first decade of the 21st century.
Southwest China regional meteorological droughtevents vary on seasonal timescales, with drought seasonbetween November and next April. Correlation coefficientshave been calculated for the drought seasonto further examine the relationships between precipitation, temperature, and drought intensity(Table 3). Accumulated integrated intensity is negatively correlatedwith precipitation during all analyzed time periods, with statistically significant relationships for annualprecipitation, drought season precipitation, and September–April precipitation. The strongest correlation(–0. 68)is for September–April precipitation. Accumulated integrated intensity is positively correlatedwith temperature during all analyzed time periods, but none of the correlations is statistically significant. To summarize, the intensity of Southwest Chinaregional meteorological drought events between 1960 and 2010 is most closely related to precipitation duringthe September–April season.4. 2 Spatial distribution characteristics
Figure 8 shows spatial distributions of the frequency and accumulated intensity of Southwest Chinaregional meteorological drought events between 1960 and 2010. Drought frequency is generally greater than20 across Southwest China. The highest frequenciesoccur in Yunnan and southern Sichuan, with peak valuesof greater than 70 near the Yunnan–Sichuan border. The spatial distribution of accumulated intensityis similar, with peak values in eastern-northern Yunnan and southern Sichuan. The spatial distribution ofdrought in Southwest China over the past 50 years ischaracterized by two typical patterns of “strong south and weak north” and “strong west and weak east”. To summarize, Yunnan and southern Sichuan are themost heavily impacted areas by droughts in SouthwestChina.
Figure 9 shows the spatial distribution of trendsin drought event frequency in Southwest China during1960–2010. The frequency of droughts has increasedacross Southwest China, with significant increasingtrends in central and northern Yunnan and northern and western Guizhou.
Among the 87 identified drought events, 26 canbe categorized as extreme and severe drought events(9 extreme events and 17 severe events). These 26events can then be divided into five types based onthe spatial distribution of their accumulated intensity:eastern type, southern type, western type, northerntype, and whole-area type. Figure 10 shows spatialdistributions of the accumulated intensity for representativecases of the five distribution types. Easterntype events mainly cover Guizhou and Chongqing(Fig. 10a). Southern type events are concentratedin Yunnan(Fig. 10b). Western type events are generallylocated in western Sichuan and Yunnan(Fig. 10c). Northern type events mainly occur in northernSichuan and Chongqing(Fig. 10d). Whole-area typeevents extend across Southwest China(Fig. 10e). Figure 11 shows frequency of the five-type extreme and severe droughts. Thirteen of the total 26 events and 6 of the 9 extreme events are southern type events. The second most frequent type is the whole-area type(8, including the remaining 3 extreme events). Twoof the events are western type, two are northern type, and one is eastern type. Yunnan is the area most frequentlyimpacted by the extreme and severe droughtevents in Southwest China.
5. Conclusions and discussion
The results of the above analysis can be summarizedas follows.
(1)The OITREE method shows a good capabilityfor identifying Southwest China regional meteorologicaldrought events. A total of 87 events are identifiedbetween 1960 and 2010, including 9 extreme events and 17 severe events. The 2009–2010 extreme droughtis the most serious meteorological drought during thepast 50 years.
(2)Southwest China regional meteorologicaldrought events generally last for 10–80 days, withthe longest being 231 days. The maximum impactedarea varies between 70 × 104 and 100 × 104 km2. The drought season in Southwest China runs fromNovember to next April and the non-drought seasonruns from May to October.
(3)Droughts are most frequent and most intensein Yunnan and southern Sichuan. Extreme and severedrought events can be divided into five types basedon their spatial distributions: southern type, westerntype, northern type, eastern type, and whole-areatype. The southern type occurs the most frequent. Yunnan is the area most frequently stricken by extreme and severe drought events in Southwest China.
(4)Southwest China regional meteorologicaldrought events have become more frequent and stronger over the past 50 years. The main reasonfor these increases appears to be a significant decreasein precipitation in Southwest China, although an increasein temperature also contributes.
This study has analyzed the characteristics ofregional meteorological drought events in SouthwestChina. Further studies are needed to reveal the causesof and specific mechanisms that lead to droughts inSouthwest China.
Acknowledgments. The language editor ofthis manuscript is Dr. Jonathon S. Wright.
|||Andreadis, K. M., E. A. Clark, A. W. Wood, et al., 2005: Twentieth-century drought in the contermi-nous United States. J. Hydrometeor., 6, 985-1001.|
|||Cheng Jiangang and Xie Mingen, 2008: The analysis of regional climate change features over Yunnan in recent 50 years. Progress in Geography, 27, 19-26.|
|||CMA Forecast and Mitigation Division, 2006: China Meteorological Geographic Division. China Meteo-rological Press, Beijing, 5.|
|||Cui Donglin, 2010: A study on objective identification of persistent regional meteorological drought events and their changes in China during the recent 50 years. Master thesis, Lanzhou University, Lanzhou, 33.|
|||Ding Yihui, 2008: The Chinese Meteorological Disasters. China Meteorological Press, Beijing, 948 pp.|
|||Huang Ronghui, Cai Rongshuo, Chen Jilong, et al., 2006: Interdecadal variations of drought and flooding dis-asters in China and their association with the East Asian climate system. Chinese J. Atmos. Sci., 30, 730-743. (in Chinese)|
|||Li Yonghua, Xu Haiming, and Liu De, 2011: Features of the extremely severe drought in the east of South-west China and anomalies of atmospheric circulation in summer 2006. Acta Meteor. Sinica, 25, 176-187.|
|||Liu Xiaoyun, Li Dongliang, and Wang Jingsong, 2012: Spatiotemporal characteristics of droughts over China during 1961-2009. J. Desert Res., 32, 473-483. (in Chinese)|
|||Liu Yu, Zhao Erxu, Huang Wei, et al., 2010: Charac-teristic analysis of precipitation and temperature trends in Yunnan Province in recent 46 years. J. Catastrophol., 25, 39-44.|
|||Ma Zhenfeng, Peng Jun, Gao Wenliang, et al., 2006: Climate variation of Southwest China in recent 40 years. Plateau Meteor., 25, 633-642. (in Chinese)|
|||Ma Zhuguo and Fu Zongbin, 2006: Facts on the drying trend of northern China during 1951-2004. Chinese Sci. Bull., 51, 2429-2439. (in Chinese)|
|||Peng Guifen, Liu Yu, and Zhang Yiping, 2009: Research on characteristics of drought and climatic trend in Yunnan Province. J. Catastrophol., 24, 40-44.|
|||Qi Haixia, Zhi Xiefei, and Bai Yongqing, 2006: Inter-decadal variation and trend analysis of the drought occurrence frequency in China. Trans. Atmos. Sci., 34, 447-455. (in Chinese)|
|||Qian, W. H., X. Shan, and Y. Zhu, 2011: Ranking re-gional drought events in China for 1960-2009. Adv. Atmos. Sci., 28, 310-321.|
|||Ren, F. M., D. L. Cui, Z. Q. Gong, et al., 2012: An ob-jective identification technique for regional extreme events. J. Climate, 25, 7015-7027.|
|||Shi Yafeng, Shen Yongping, and Hu Ruji, 2002: Pre-liminary study on signal, impact and foreground of climatic shift from warm-dry to warm-humid in Northwest China. J. Glaciol. Geocryol., 24, 219-226.|
|||—-, —-, and Li Dongliang, 2003: Discussion on the present climate change from warm-dry to warm-wet in Northwest China. Quaternary Sciences, 23, 152-164.|
|||Yang Ping, Xiao Ziniu, Yang Jie, et al., 2013: Char-acteristics of clustering extreme drought events in China during 1961-2010. Acta Meteor. Sinica, 27, 186-198.|
|||Zhai Panmao and Zou Xukai, 2005: Changes in tempera-ture and precipitation and their impacts on drought in China during 1951-2003. Adv. Climate Change Res., 1, 16-18. (in Chinese)|
|||Zhang De0 er, 2005: Severe drought events as revealed in the climate records of China and their temperature situations over the last 1000 years. Acta Meteor. Sinica, 19, 485-491.|
|||Zhang Qiang, Zou Xukai, Xiao Fengjing, et al., 2006: Classification of Meteorological Droughts. GB/T20481-2006. Standards Press of China, Bei-jing, 1-17.|
|||—-, Fan Xuebiao, Ma Zhuguo, et al., 2009: Droughts. China Meteorological Press, Beijing, 178-191.|
|||Zhang Qingyun, Wei Jie, and Tao Shiyan, 2003: The decadal and interannual variations of drought in northern China and association with the circulation. Adv. Climate Change Res., 8, 307-318. (in Chinese)|
|||Zou Xukai and Zhang Qiang, 2008: Preliminary studies on variations in droughts over China during past 50 years. J. Appl. Meteor. Sci., 19, 679-687. (in Chi-nese)|