目前水文序列趋势研究主要集中于判断一个水文序列是否存在着显著趋势特征和该序列趋势是上升还是下降,但由于水文序列可分解为趋势、周期、随机等成分,论文提出了趋势成分在水文时间序列中的比重问题,并基于Mann-Kendall法,得到了趋势成分比重的计算公式。结果表明,趋势的显著程度与对偶数p和p0的差值有直接关系,差值越大,显著程度越明显;趋势成分比重取决于该差值及序列长度;显著趋势条件下趋势成分比重分析的结果,初步揭示了序列长度对趋势成分比重的影响,证明了用Mann-Kendall方法分析时间序列趋势特征时至少需要10个及以上的样本,且随着序列长度的增加,比重相应减小,因此当序列长度超过一定的范围时,Mann-Kendall方法是否适用于分析水文时间序列的显著趋势特征有待进一步探讨。应用于闽江流域竹岐站近72 a来径流系列的分析表明,序列呈下降趋势,但趋势不显著;趋势成分占整个时间序列近1/4。
The researches of trend in a hydrological series usually focus on detecting whether the trend characteristic is significant and judging whether the trend characteristic is increasing or decreasing. The paper proposes a concept about the weight of trend component in a hydrological time series which is usually decomposed as trend, period, random component and so on. The calculation formula of the weight of trend component is derived based on Mann-Kendall method. The results show that the difference of dual number p and p0 affects directly the significant level of the trend, and the weight of trend component depends on the value of (p-p0) and the sequence length of series. It also reveals the effects of the sequence length on the weight of trend component for a series with significant trending level. Firstly, it verifies the conclusion that at least 10 samples are necessary to conduct the trend analysis using Mann-Kendall method. Secondly, the weight decreases accordingly with the increase of the sequence length, which also indicates that whether the Mann-Kendall method is applicable for the analysis of trend characteristic should be further studied when the sequence length exceeds a certain range in a hydrological time series. The annual runoff series of Zhuqi Gauging Station in the Minjiang River Watershed from 1936 to 2007, about 72 years, has been taken as an example. The results show that the trend of the series is decreasing, but not significantly. And the weight of the trend component for the runoff series is 24.73%, which is nearly 1/4.
[1] 全国统计专业技术资格考用书编写委员会.统计基础知识和实务[M].北京:中国统计出版社,2005.
[2] Jain S, Lall U. Magnitude and timing of annual maximum floods: Trends and large-scale elimatic associations for the Blacksmith Fork River[J]. Water Resources,2000,36(12):3641-3651.
[3] Kimpel Ichiyangl, Manaby D Yamanaka, Toshitaka Muraji, et al. Precipitation in Nepell between 1987 and 1996 [J]. International Journal of Climatology,2007,2(4):609-623.
[4] Yan Luo, Shen Liu, Sheng Leifu, et al. Trends of precipitation in Beijiang River Basin, Guangdong Province, China [J]. Hydrological Processes,2008,22(13):2377-2386.
[5] 徐宗学,张楠.黄河流域近50年降水变化趋势分析[J].地理研究,2006,25(1):27-35.
[6] 徐淑升,冯彦,罗艳.大盈江流域降雨径流变化趋势及原因分析[J].长江流域资源与环境,2007,16(增2):59-64.
[7] 赵少华,杨永辉,邱国玉,等.河北平原34年来气候变化趋势分析[J].资源科学,2007,29(4):109-112.
[8] LAN Yong-chao, KANG Er-si, ZHANG Ji-shi, et al. Trends and characteristics on mountainous runoff in Hexi Inland Arid region [J]. Journal of Desert Research,2002,22(2):135-141.
[9] 冯平,李建柱,徐仙.潘家口水库入库水资源变化趋势及影响因素[J].地理研究,2008,27(1):213-220.
[10] 曾燕.黄河流域实际蒸散发分布式模型研究.北京:中国科学院,2004.
[11] Khaled H Hamed. Trend detection in hydrologic data: The mann-kendall trend test under the scaling hypothesis [J]. Journal of Hydrology,2008,349:350-363.
[12] Donald H Burn, Mohamed A, Hag Elnur. Detection of hydrologic trends and variability [J]. Journal of Hydrology, 2005,255:107-122.
[13] 张建云,章四龙,王金星,等.近50年来中国六大流域年际径流变化趋势研究[J].水科学进展,2007,18(2):230-234.
[14] 徐宗学,和宛琳.近40年黄河源区气候要素分布特征及变化趋势分析[J].高原气象,2006,25(5): 906-913.
[15] 姜逢清,朱诚,胡汝骥.1960—1997年新疆北部降水序列的趋势探测[J].地理科学,2002,22(6):669-672.[JIANG Feng-qing, ZHU Cheng, HU Ru-ji. Trend analysis of precipitation over northern Xinjiang for the period 1961-1997. Scientia Geographica Sinica,2002,22(6):669-672.]
[16] 张文纲,李述训,庞强强.青藏高原40年来降水量时空变化趋势[J].水科学进展,2009,20(2):168-176.
[17] 曹建廷,秦大河,罗勇,等.长江源区1956—2000年径流量变化分析[J].水科学进展,2007,18(1):29-33.[CAO Jian-ting, QIN Da-he, LUO Yong, et al. Variation of annual runoff from 1956-2000 in the source area of the Yangtze River Basin. Advances in Water Science,2007,18(1):29-33.]
[18] 丁晶,邓育仁.随机水文学[M].成都:成都科技大学出版社,1988.
[19] 同济大学数学系.线性代数[M].第五版.北京:高等教育出版社, 2007.
[20] 梅长林,周家良.实用统计方法[M].北京:科学出版社,2002.
[21] Khaled H Hamed. Trend detection in hydrologic data: The Mann-Kendall trend test under the scaling hypothesis[J]. Journal of Hydrology,2008,349:350-363.
[22] Mann-Kendall test for detecting trends . http://www.minitab.com.au/support/macros/default.aspx?Action=display&cal=timeseries#92.
