五种干旱指数在淮河流域的适用性研究
摘要
我国地处东亚,季风气候明显,年际间季风的不稳定性造成了我国干旱的频繁发生,而淮河流域又地处我国南北气候过渡带,流域南部和北部的气候差异明显,包括干流在内的流域南部夏季处于江淮梅雨的北缘,降水年际变化大,旱、涝灾害频繁,因此,对各种气象干旱指标如CI指数、降水距平百分率、标准化降水指数、降水Z指数、相对湿润指数等进行对比研究,给出它们在我国淮河流域不同时段对干旱监测的优缺点和适应性(持续时间、突变性等),并提出改进和完善方法尤为必要。本文选用1960—2010年淮河流域34个地面气象观测台站的逐日降水、日平均气温、日最高气温、日最低气温、日照时数、风速、相对湿度等气象要素的实测资料,分别计算了34测站各自的Z指数、降水距平百分率、相对湿润指数、标准化降水指数以及CI指数,利用《中国气象灾害大典》安徽卷、河南卷、江苏卷以及山东卷的实况资料进行了详细比对。结果表明：
(1)在34个代表站中,Z指数在五种干旱指数应用中效果最好,空间上在安徽、江苏、河南、山东四省中都达到了70%以上,最好的是山东站达到75%；另外五种干旱指数中除CI指数外其余四种效果较好的都是位于山东；
(2)时间上,Z指数仅在4-6月吻合率维持在67%,其余月份也都在70%以上,最高的是在11月达到78%；SPI指数在1-2月吻合率较差其余同Z指数总体相当；MI指数效果最差,应用最好的山东仅为49.6%；Pa指数在11月到次年2月还有7,8月份伏夏这段时间,效果较差。
(3)CI指数应用效果最好的是在河南,最差的是在山东,且从CI指数在淮河流域四省的应用结果对比来看,CI指数在河南、安徽两个内陆省份比江苏和山东两个沿海省份要好,CI指数在夏季效果最好,吻合率达到了80.3%,其次是春季,吻合率为74.2%,效果较差的为秋季和冬季,吻合率分别为58.1%和51.5%。通过比对整个淮河流域的干旱实况同CI指数,结果表明2065个干旱事件中只有1290个干旱事件是同实况吻合的,吻合率达到62.5%。
(4)从淮河流域的五种干旱指数的应用效果来看：夏季,CI指数效果最好,其次是Z指数和SPI指数；秋季Z指数效果最好,其次是SPI指数和Pa指数；冬季效果最好的是Z指数,其次是SPI指数和Pa指数；Pa指数和SPI指数在春季的效果较好。
(5)五种干旱指数中Z指数是最优的,但是统计也发现,Z指数先前的阈值对淮河流域有偏宽,根据累计频率法进行了调整。
(6)对等权和非等权的CI、MI指数亦作了比较,得出非等权CI、MI指数与修正前的等权CI、MI指数相比无论是实际旱情还是干旱总日数在空间分布上都有很好的一致性,但非等权的指数对干旱的敏感度更高。对于实际干旱监测业务来说,非等权的指数明显优于等权的指数。
最后基于Z指数分析了淮河流域及淮河以南地区的干旱时空分布特征。淮河流域冬春夏秋四个季节的EOF向量场前三个特征向量表明淮河流域有三种空间型分布,其中第一特征向量皆为唯一的正值,表明全区旱涝趋势变化具有很好的一致性。淮河以南地区是通过REOF分析将降水距平的空问结构分为5个具有较高相关性的异常区,在5个异常区对应的载荷向量场最大值的岳阳站、广州站、丽水站、信阳站、巢湖站,分别作了趋势分析、M-K突变分析以及小波分析。
China is located in the East Asian, monsoon climate is significant, the inter-annual monsoon instability caused frequent drought in China. And The Huaihe River Basin is located in the north and south China climate transition zone, the southern and northern basin climate differences are significant, including the River Valley, and the southern summer in the northern edge of Meiyu, the inter-annual variation in rainfall, drought, floods and natural disasters. Therefore, a variety of meteorological drought indicators such as the CI index, precipitation anomaly percentage, standardized precipitation index, precipitation Z index, relative humidity index and other comparative studies, given their different times of the Huai River Basin in China's drought monitoring and adapting the strengths and weaknesses nature (duration, mutation, etc.), and made particularly necessary to improve and perfect the method. Based on daily precipitation, average temperature, daily maximum temperature, daily minimum temperatures, sunshine hours, wind speed, relative humidity and other weather elements measured data of1960-2010about34surface weather observation stations in Huaihe River Basin, the Z-index, precipitation anomaly percentage, relative humidity index, standardized precipitation index, and CI index were calculated. Then gave these index a detailed comparison with the Recorded live data in the China Meteorological Disasters about Anhui volume, Henan volume, Shandong volume and Jiangsu volume.The results showed that:
The Z index is the best in five kinds of drought index of34representative stations. In the spatial scale, The Z-index percentage of Anhui, Jiangsu, Henan and Shandong four provinces have reached more than70%, the best is the Shandong station, it reached75%; In addition, the better effect of the remaining four besides the CI is located in Shandong;
In the scale of time, The coincidence rate of the Z index maintain at67%Only from April to June, all the rest are in70%above. The highest have reached78%in November;The SPI index and Z index are roughtly the same in effect but in January and February.The MI index is the worst, the best application of the best in Shandong is only49.6%, the coincidence rate of Pa index is poor from November to next February and fromJuly to August.
The best effect application of the CI index is in Henan, on the contrary,the worst is in Shandong. The application effect of the CI index in Henan, Anhui two provinces are better than in Jiangsu and Shandong two coastal provinces, The application effect of the CI index is best in summer and its coincidence rate have reached80.3%, followed by the spring, its coincidence rate is74.2%, The effect of the CI index in autumn and winter is poorer, Their coincidence rate respectively is only58.1%and51.5%.In the whole of Huaihe Basin, only1290drought have reflected the actual situation events in2065drought events, the coincidence rate is62.5%.
In the summer, the CI index effect is best, followed by the Z index and the SPI index.In Autumn, the Z index effect is best,followed by the SPI index and the Pa index. In Winter, the Z index effect is best, followed by the SPI index and the Pa index. The effect of the Pa and SPI index is good in Spring
Of five kinds of drought index, the Z index is the best, but the statistics also found, the Z index previous threshold on Huaihe basin have partial width, and according to the cumulative frequency method, the Z index previous threshold were adjusted.
Equal weight and non equal CI, MI index also be made a comparison, they have a very good consistency in the space distributing, but non weight index sensitivity to drought were higher. For the actual drought monitoring service, non weight index were significantly better than equal weight index.
Finally, we gave some analysis to the drought spatial distribution characteristics of Huaihe basin and the area south based on the Z index. The first three feature vectors in the EOF vector field to Huaihe basin in winter, spring, summer and autumn four seasons reveal that Huaihe basin has three kinds of spatial distribution, In which the first feature vector is the only positive, indicating the region drought trend has good consistency. Through the REOF analysis, the precipitation anomaly spatial structure in the south is divided into5anomaly area that having higher correlation, these5abnormal area to the corresponding maximum load vector field are Yueyang station, Guangzhou station, Lishui station, Xinyang station, Chaohu station, respectively made a trend analysis, M-K mutation analysis and wavelet analysis.
(1)在34个代表站中,Z指数在五种干旱指数应用中效果最好,空间上在安徽、江苏、河南、山东四省中都达到了70%以上,最好的是山东站达到75%；另外五种干旱指数中除CI指数外其余四种效果较好的都是位于山东；
(2)时间上,Z指数仅在4-6月吻合率维持在67%,其余月份也都在70%以上,最高的是在11月达到78%；SPI指数在1-2月吻合率较差其余同Z指数总体相当；MI指数效果最差,应用最好的山东仅为49.6%；Pa指数在11月到次年2月还有7,8月份伏夏这段时间,效果较差。
(3)CI指数应用效果最好的是在河南,最差的是在山东,且从CI指数在淮河流域四省的应用结果对比来看,CI指数在河南、安徽两个内陆省份比江苏和山东两个沿海省份要好,CI指数在夏季效果最好,吻合率达到了80.3%,其次是春季,吻合率为74.2%,效果较差的为秋季和冬季,吻合率分别为58.1%和51.5%。通过比对整个淮河流域的干旱实况同CI指数,结果表明2065个干旱事件中只有1290个干旱事件是同实况吻合的,吻合率达到62.5%。
(4)从淮河流域的五种干旱指数的应用效果来看：夏季,CI指数效果最好,其次是Z指数和SPI指数；秋季Z指数效果最好,其次是SPI指数和Pa指数；冬季效果最好的是Z指数,其次是SPI指数和Pa指数；Pa指数和SPI指数在春季的效果较好。
(5)五种干旱指数中Z指数是最优的,但是统计也发现,Z指数先前的阈值对淮河流域有偏宽,根据累计频率法进行了调整。
(6)对等权和非等权的CI、MI指数亦作了比较,得出非等权CI、MI指数与修正前的等权CI、MI指数相比无论是实际旱情还是干旱总日数在空间分布上都有很好的一致性,但非等权的指数对干旱的敏感度更高。对于实际干旱监测业务来说,非等权的指数明显优于等权的指数。
最后基于Z指数分析了淮河流域及淮河以南地区的干旱时空分布特征。淮河流域冬春夏秋四个季节的EOF向量场前三个特征向量表明淮河流域有三种空间型分布,其中第一特征向量皆为唯一的正值,表明全区旱涝趋势变化具有很好的一致性。淮河以南地区是通过REOF分析将降水距平的空问结构分为5个具有较高相关性的异常区,在5个异常区对应的载荷向量场最大值的岳阳站、广州站、丽水站、信阳站、巢湖站,分别作了趋势分析、M-K突变分析以及小波分析。
China is located in the East Asian, monsoon climate is significant, the inter-annual monsoon instability caused frequent drought in China. And The Huaihe River Basin is located in the north and south China climate transition zone, the southern and northern basin climate differences are significant, including the River Valley, and the southern summer in the northern edge of Meiyu, the inter-annual variation in rainfall, drought, floods and natural disasters. Therefore, a variety of meteorological drought indicators such as the CI index, precipitation anomaly percentage, standardized precipitation index, precipitation Z index, relative humidity index and other comparative studies, given their different times of the Huai River Basin in China's drought monitoring and adapting the strengths and weaknesses nature (duration, mutation, etc.), and made particularly necessary to improve and perfect the method. Based on daily precipitation, average temperature, daily maximum temperature, daily minimum temperatures, sunshine hours, wind speed, relative humidity and other weather elements measured data of1960-2010about34surface weather observation stations in Huaihe River Basin, the Z-index, precipitation anomaly percentage, relative humidity index, standardized precipitation index, and CI index were calculated. Then gave these index a detailed comparison with the Recorded live data in the China Meteorological Disasters about Anhui volume, Henan volume, Shandong volume and Jiangsu volume.The results showed that:
The Z index is the best in five kinds of drought index of34representative stations. In the spatial scale, The Z-index percentage of Anhui, Jiangsu, Henan and Shandong four provinces have reached more than70%, the best is the Shandong station, it reached75%; In addition, the better effect of the remaining four besides the CI is located in Shandong;
In the scale of time, The coincidence rate of the Z index maintain at67%Only from April to June, all the rest are in70%above. The highest have reached78%in November;The SPI index and Z index are roughtly the same in effect but in January and February.The MI index is the worst, the best application of the best in Shandong is only49.6%, the coincidence rate of Pa index is poor from November to next February and fromJuly to August.
The best effect application of the CI index is in Henan, on the contrary,the worst is in Shandong. The application effect of the CI index in Henan, Anhui two provinces are better than in Jiangsu and Shandong two coastal provinces, The application effect of the CI index is best in summer and its coincidence rate have reached80.3%, followed by the spring, its coincidence rate is74.2%, The effect of the CI index in autumn and winter is poorer, Their coincidence rate respectively is only58.1%and51.5%.In the whole of Huaihe Basin, only1290drought have reflected the actual situation events in2065drought events, the coincidence rate is62.5%.
In the summer, the CI index effect is best, followed by the Z index and the SPI index.In Autumn, the Z index effect is best,followed by the SPI index and the Pa index. In Winter, the Z index effect is best, followed by the SPI index and the Pa index. The effect of the Pa and SPI index is good in Spring
Of five kinds of drought index, the Z index is the best, but the statistics also found, the Z index previous threshold on Huaihe basin have partial width, and according to the cumulative frequency method, the Z index previous threshold were adjusted.
Equal weight and non equal CI, MI index also be made a comparison, they have a very good consistency in the space distributing, but non weight index sensitivity to drought were higher. For the actual drought monitoring service, non weight index were significantly better than equal weight index.
Finally, we gave some analysis to the drought spatial distribution characteristics of Huaihe basin and the area south based on the Z index. The first three feature vectors in the EOF vector field to Huaihe basin in winter, spring, summer and autumn four seasons reveal that Huaihe basin has three kinds of spatial distribution, In which the first feature vector is the only positive, indicating the region drought trend has good consistency. Through the REOF analysis, the precipitation anomaly spatial structure in the south is divided into5anomaly area that having higher correlation, these5abnormal area to the corresponding maximum load vector field are Yueyang station, Guangzhou station, Lishui station, Xinyang station, Chaohu station, respectively made a trend analysis, M-K mutation analysis and wavelet analysis.
引文
[1]叶笃正,黄荣辉.长江黄河流域旱涝规律和成因研究[M].济南：山东科学技术出版社,1996：9-10.
[2]袁文平,周广胜.标准化降水指标与Z指数在我国应用的对比分析[J].植物生态学报,2004,28(4)：523-529.
[3]卫捷,马柱国.Palmer干旱指数、地表湿润指数与降水距平的比较[J].地理学报,2003,58(3)：117-118.
[4]American Meteorological Society. Meteorological drought:policy statement [R]. [s..1] American Meteorological Society,1997.
[5]谢应齐.关于干旱指标的研究[J].自然灾害学报,1993,2(2)：55-62.
[6]孙荣强.干旱定义及其指标评述[J].灾害学,1994,9(1)：17-21.
[7]王延禄.我国建立、引用和验证气象干旱指标综述[J].干旱区地理,1990,13(3)：80-86.
[8]冯平,李绍飞,王仲珏,等.干旱识别与分析指标综述[J].中国农村水利水电,2002,1(7)：13-15.
[9]姚玉璧,张存杰,邓振镛,等.气象、农业干旱指标综述[J].干旱地区农业研究,2007,25(1)：185-189.
[10]王劲松,郭江勇,周跃武,等.干旱指标研究的进展与展望[J].干旱区地理,2007,30(1)：60-65.
[11]王英,迟道才.干旱指标研究与进展[J],科技创新导报,2009,35(1)：72-74.
[12]鞠笑生,杨贤为,陈丽娟,等.我国单站旱涝指标确定和区域旱涝级别划分的研究[J].应用气象学报,1997,8(1)：26-33.
[13]李有宏,董立新,白彦芳,等.三种干旱指标应用对比分析[J].青海气象,2003,2(1)：35-38.
[14]程炳岩,钱晓燕,朱业玉,等.河南干旱指标的客观性研究[J].河南气象, 2000,1(1)：32-33.
[15]韩海涛,胡文超,陈学君,等.三种气象干旱指标应用比较研究[J].干旱地区农业研究,2009,27(1)：237-241.
[16]刘招,吴新,乔长录,等.三种干旱指数在渭北旱塬应用的对比分析[J].人民黄河,2010,32(7)：71-75.
[17]张成才,周勇.3种旱涝指标在郑州市干旱分析中的对比研究[J].安徽农业科学,2010,39(4)：2005-2007.
[18]谢五三,田红.五种干旱指标在安徽省应用研究[J].气象,2011,37(4)：503-507.
[19]方茸,周后福,屈雅,等.基于江淮分水岭地区的Z指数修正[J].气象,2010,36(10)：110-113.
[20]温克刚,翟武全.中国气象灾害大典(安徽卷)[M].北京：气象出版社,2007：181-211.
[21]温克刚,庞天荷.中国气象灾害大典(河南卷)[M].北京：气象出版社,2005：98-127.
[22]温克刚,卞光辉.中国气象灾害大典(江苏卷)[M].北京：气象出版社,2008：191-201.
[23]温克刚,王建国.中国气象灾害大典(山东卷)[M].北京：气象出版社,2006： 91-125.
[24]马开玉,丁裕国.气候统计原理与方法[M].北京：气象出版社,1993：61-63.
[25]陆桂荣,郑美琴,周秀君,等.山东日照市2种干旱指标的应用对比[J].干旱气象,2010,28(1)：102-106.
[26]张强,高歌.我国近50年旱涝灾害时空变化及监测预警服务[J].科技导报,2004,1(7)：21-24.
[27]中国气象局.GB／T20481—2006[S].北京：中国标准出版社,2006.
[28]赵俊芳,郭建平,徐精文,等。基于湿润指数的中国干湿状况变化趋势[J].农业工程学报,2010,26(8)：18-23.
[29]施能.气象科研与预报中的多元统计方法[M].北京：气象出版社,2002(第二版)：2-3.
[30]李海东,沈渭寿,赵卫,等.1957—2007年雅鲁藏布江中游河谷降水变化的小波分析[J].气象与环境学报,2010,26(4)：1-7.
[31]张楠,苗春生,邵海燕,等.1951—2007年华北地区夏季气温变化特征[J].气象与环境学报,2009,25(6)：23-28.
[32]尤卫红.气候变化的多尺度诊断分析和预测的多种技术方法研究[M].北京：气象出版社,1998：9-20.
[33]梁树献,杨亚群,徐珉,等.淮河流域6—8月旱涝分布特征[J].水文,2001,25(2)：54-56.
[34]温克刚,曾庆华.中国气象灾害大典(湖南卷)[M].北京：气象出版社,2006：181-211.
[35]温克刚,曾庆华。中国气象灾害大典(湖南卷)[M].北京：气象出版社,2006：91-127.
[36]孙林海,陈兴芳.南涝北旱的年代气候特点和形成条件[J].应用气象学报,2003,14(6)：641-647.
[37]王遵娅,丁一汇,何金海,等.近50年来中国气候变化特征的再分析[J].气象学报,2004,62(2)：228-236.
[38]茅海祥,王文.1960—2010年中国南方地区夏季旱涝时空分布特征[J].气象与环境学报,2012,28(1)：90-95.
[2]袁文平,周广胜.标准化降水指标与Z指数在我国应用的对比分析[J].植物生态学报,2004,28(4)：523-529.
[3]卫捷,马柱国.Palmer干旱指数、地表湿润指数与降水距平的比较[J].地理学报,2003,58(3)：117-118.
[4]American Meteorological Society. Meteorological drought:policy statement [R]. [s..1] American Meteorological Society,1997.
[5]谢应齐.关于干旱指标的研究[J].自然灾害学报,1993,2(2)：55-62.
[6]孙荣强.干旱定义及其指标评述[J].灾害学,1994,9(1)：17-21.
[7]王延禄.我国建立、引用和验证气象干旱指标综述[J].干旱区地理,1990,13(3)：80-86.
[8]冯平,李绍飞,王仲珏,等.干旱识别与分析指标综述[J].中国农村水利水电,2002,1(7)：13-15.
[9]姚玉璧,张存杰,邓振镛,等.气象、农业干旱指标综述[J].干旱地区农业研究,2007,25(1)：185-189.
[10]王劲松,郭江勇,周跃武,等.干旱指标研究的进展与展望[J].干旱区地理,2007,30(1)：60-65.
[11]王英,迟道才.干旱指标研究与进展[J],科技创新导报,2009,35(1)：72-74.
[12]鞠笑生,杨贤为,陈丽娟,等.我国单站旱涝指标确定和区域旱涝级别划分的研究[J].应用气象学报,1997,8(1)：26-33.
[13]李有宏,董立新,白彦芳,等.三种干旱指标应用对比分析[J].青海气象,2003,2(1)：35-38.
[14]程炳岩,钱晓燕,朱业玉,等.河南干旱指标的客观性研究[J].河南气象, 2000,1(1)：32-33.
[15]韩海涛,胡文超,陈学君,等.三种气象干旱指标应用比较研究[J].干旱地区农业研究,2009,27(1)：237-241.
[16]刘招,吴新,乔长录,等.三种干旱指数在渭北旱塬应用的对比分析[J].人民黄河,2010,32(7)：71-75.
[17]张成才,周勇.3种旱涝指标在郑州市干旱分析中的对比研究[J].安徽农业科学,2010,39(4)：2005-2007.
[18]谢五三,田红.五种干旱指标在安徽省应用研究[J].气象,2011,37(4)：503-507.
[19]方茸,周后福,屈雅,等.基于江淮分水岭地区的Z指数修正[J].气象,2010,36(10)：110-113.
[20]温克刚,翟武全.中国气象灾害大典(安徽卷)[M].北京：气象出版社,2007：181-211.
[21]温克刚,庞天荷.中国气象灾害大典(河南卷)[M].北京：气象出版社,2005：98-127.
[22]温克刚,卞光辉.中国气象灾害大典(江苏卷)[M].北京：气象出版社,2008：191-201.
[23]温克刚,王建国.中国气象灾害大典(山东卷)[M].北京：气象出版社,2006： 91-125.
[24]马开玉,丁裕国.气候统计原理与方法[M].北京：气象出版社,1993：61-63.
[25]陆桂荣,郑美琴,周秀君,等.山东日照市2种干旱指标的应用对比[J].干旱气象,2010,28(1)：102-106.
[26]张强,高歌.我国近50年旱涝灾害时空变化及监测预警服务[J].科技导报,2004,1(7)：21-24.
[27]中国气象局.GB／T20481—2006[S].北京：中国标准出版社,2006.
[28]赵俊芳,郭建平,徐精文,等。基于湿润指数的中国干湿状况变化趋势[J].农业工程学报,2010,26(8)：18-23.
[29]施能.气象科研与预报中的多元统计方法[M].北京：气象出版社,2002(第二版)：2-3.
[30]李海东,沈渭寿,赵卫,等.1957—2007年雅鲁藏布江中游河谷降水变化的小波分析[J].气象与环境学报,2010,26(4)：1-7.
[31]张楠,苗春生,邵海燕,等.1951—2007年华北地区夏季气温变化特征[J].气象与环境学报,2009,25(6)：23-28.
[32]尤卫红.气候变化的多尺度诊断分析和预测的多种技术方法研究[M].北京：气象出版社,1998：9-20.
[33]梁树献,杨亚群,徐珉,等.淮河流域6—8月旱涝分布特征[J].水文,2001,25(2)：54-56.
[34]温克刚,曾庆华.中国气象灾害大典(湖南卷)[M].北京：气象出版社,2006：181-211.
[35]温克刚,曾庆华。中国气象灾害大典(湖南卷)[M].北京：气象出版社,2006：91-127.
[36]孙林海,陈兴芳.南涝北旱的年代气候特点和形成条件[J].应用气象学报,2003,14(6)：641-647.
[37]王遵娅,丁一汇,何金海,等.近50年来中国气候变化特征的再分析[J].气象学报,2004,62(2)：228-236.
[38]茅海祥,王文.1960—2010年中国南方地区夏季旱涝时空分布特征[J].气象与环境学报,2012,28(1)：90-95.