Mann-Kendall 检验

        Mann-Kendall 检验是一种以非参数检验为基础,进行时间序列数据趋势分析的检验方法。世界气象组织(WMO)推荐使用这种方法来评估水文气象时间序列是否存在单调趋势[15]。该方法将非参数检验与非线性趋势检验相结合,使结果合理化,主要用于蒸散量时序数据趋势变化显著性分析及突变检验[16]。Mann-Kandall 方法通过检验统计量Z 值评估蒸散量时序数据显著趋势的存在:

Z = \begin{cases} \frac{​{S-1}}{\sqrt{\text{Var}(S)}} & , \text{if } S > 0 \\ 0 & , \text{if } S = 0 \\ \frac{​{S+1}}{\sqrt{\text{Var}(S)}} & , \text{if } S < 0 \end{cases}

S = \sum_{i=1}^{n-1} \sum_{j=i+1}^{n} \text{sgn}(x_j - x_i)

\text{sgn}(x_j - x_i) = \begin{cases} 1 & , \text{if } x_j > x_i \\ 0 & , \text{if } x_j = x_i \\ -1 & , \text{if } x_j < x_i \end{cases} \quad (4)

S——Kendall秩次统计量

xi,yj——第i,j年对应的蒸散量;

n——年对应的蒸散量;

        在特定 α 显著性水平下,采用双边趋势检验,根据正态分布表中Z_{1-\frac{\alpha}{2}}的值判断蒸散量趋势变化是否显著。若|Z| > |Z_{1-\frac{\alpha}{2}}|,该时序数据趋势变化通过了 α 显著性水平。本文选取 0.01、0.05 和 0.10 的 α 水平进行显著性检验,Z_{1-\frac{\alpha}{2}}的值分别为 2.576、1.960 和 1.645。根据三个断点将蒸散量变化趋势分为极显著、显著、弱显著以及不显著四种状态。

Mann-Kandall 可通过对蒸散量时序数据构造一组秩序列,用于总体趋势分析中的突变点检验。秩序列计算公式为:

S_k = \sum_{i=1}^{k} r_i

此时设定蒸散量时序数据是独立、互不干扰的,UF 检验统计量为:


r_j = \begin{cases} 1 & \text{if } x_j > y_j \\ 0 & \text{if } 0 \leq x_j \leq y_j \end{cases}

此时设定蒸散量时序数据是独立、互不干扰的,UF 检验统计量为:


UF_k = \frac{E(S_k) - S_k}{\sqrt{\text{Var}(S_k)}}


E(S_k) = \frac{n(n+1)}{4}


\text{Var}(S_i) = \frac{n(n - 1)(2n + 5)}{72}

通过计算,得到蒸散量曲线。通过UF_i曲线与α 显著性水平下对应的正态分布表临界值U_\alpha对比:若|UFi|>U_\alpha,意味着蒸散量时序数据存在显著的趋势变化。将逆序的蒸散量时序数据X(x_n,……,x_2,x_1) 重复以上计算步骤,并将计算结果乘以−1,得到统计量曲线UB_i。通过两条统计量曲线在α 置信区间内的交点个数及位置,可以判断该时序数据突变发生的次数及相应的时间。

[15] TIAN Q, WANG Q, ZHAN C, et al. Analysis of climate change in the coastal zone of Eastern China, against the background of global climate change over the last fifty years: case study of Shandong peninsula, China [J]. International Journal of Geosciences, 2012, 3(2): 379−390.

近50年全球气候变化背景下中国东部沿海地区气候变化分析——以山东半岛为例

摘要

        利用山东半岛近50年来的年平均、最高、最低气温和年降水、6 ~ 9月降水资料,采用非参数Mann-Kendall检验、累积差曲线和有序聚类分析等方法对气候变化进行了分析。结果表明:

        1)近50 a,山东半岛年平均气温呈显著上升趋势,在1990年前后有明显的突变点;近50年来,半岛变暖的主要原因是年最低气温的显著升高。年最高气温呈增减混合趋势,但在统计上不显著,未出现突变;

        (2)近50 a来,年降水量呈减少趋势,大部分台站在1980年前后出现突变;但早在1966年就有一个过渡点,在几个车站。6 - 9月降水量的减少是年降水量减少的主要原因。6 - 9月降水占全年降水的比重在近50年略有下降;

        ③山东半岛的气温演变与中国大部分地区基本一致,但同期增温速度较快;与中国其他气候带相比,降水减少趋势更为显著。在半岛内部,东南部的气温和降水突变要早于西北部;东南部降水减少幅度较大,西北部增温幅度较大。该研究对了解海岸带气候变化及其环境影响具有重要意义。

1. 介绍

        气候是人类赖以生存的自然环境的组成部分;全球气候变化已经成为一个热点问题,引起了各国政府和科学家的共同关注。近百年来,全球气候变化特别是全球变暖已成为不可避免的科学事实[1-7]。根据IPCC(2007), 20世纪(1906-2005)全球平均地表温度上升了0.74℃。海岸带是位于陆地和海洋之间的狭长地带。海岸带作为全球经济最发达、人口最密集的地区,其气候变化对人类社会的发展具有特殊的意义[8,9]。此外,本文所涉及的海岸带是陆地、海洋和大气的界面,海陆之间存在频繁的水分、热通量、沉积物交换以及C、N、S等重要营养元素交换;滨海湿地生态系统还会影响大气中CH4、CO2和水蒸气等温室气体的浓度[10]。因此,海岸带的气候变化将对全球气候变化产生深远的影响。然而,海岸带气候变化不仅受到全球气候变化的影响,也不可避免地有其独特的模式,海岸带气候受到来自大陆和海洋的大气环流的综合影响,海岸带存在着海洋和陆地两个下垫面,它们的热物性完全不同[11]。因此,海岸带的气候变化值得进一步研究。

        中国位于欧亚大陆东部,与太平洋西海岸接壤,拥有18000多公里的海岸线。中国东海岸面积300万平方公里,从南到北横跨热带、亚热带和温带[12]。中国沿海地区占全国人口的41%以上,占全国大中城市的50%以上,占国内生产总值(GDP)的60%以上[13],中国沿海发展在国民经济中起着主导作用。受最大大陆和海洋热力性质不同的影响,中国东海岸以典型的东亚季风气候为主[14,15]。本文以位于中国东部温带的山东半岛为例,分析了近50年来中国东部沿海地区的气候变化,以进一步揭示全球气候变化背景下的温带季风气候演变。因此,本研究不仅对了解气候变化及其对海岸带环境的影响具有重要意义,而且对我们应对全球变化,实现区域可持续发展,特别是对实施山东半岛蓝色经济区规划具有重要的实用价值。

2. 研究区域

        山东半岛位于连接莱州湾和胶州湾的胶莱河东岸,向东延伸至渤海和黄海之间,是中国最大的半岛(北纬35˚35′~ 38˚23′,东经119˚30′~ 122˚42′),面积为2.7 × 104 km2(图1)。半岛主要地貌以古老的变质花岗岩山丘为主,仅有少数中低洼山脉突出于山丘之上,地形起伏平缓。山区约占总面积的70%,海拔约200米。半岛内的山脉主要包括崖、昆嵛山、威德山、罗山等,呈近东西走向,靠近半岛北部。


        海拔约500 - 1000米,其中崂山最高(1130米)。


        山东半岛发育典型的温带基岩海岸,海岸线蜿蜒曲折[14]。

        山东半岛冬季受欧亚大陆高纬度西伯利亚-蒙古高压气团控制,气温、湿度较低,形成近北来的干冷冬季季风;来自太平洋低纬度气团,夏季高温高湿盛行,由此形成近南暖湿夏季风。在夏季风和冬季风的控制下,山东半岛形成了典型的暖温带湿润季风气候,四季分明,夏季炎热多雨,冬季寒冷干燥。


        受夏季风影响显著,年降水多发生在6 ~ 9月的夏季,降水强度大,常伴有暴雨。山东半岛年平均降水量600 ~ 850毫米;1月平均气温为-3˚C ~ -1˚C, 8月(最热月)为25˚C左右,极端最高气温为38˚C左右,年温差较大[14]。

        山东半岛受海洋影响,年降水量比同纬度大陆内陆地区多200 ~ 300毫米,夏季较凉爽。由于山东半岛靠近西北的小渤海


        东南与广袤的黄海接壤,同时半岛中部的山脉是阻挡黄海暖湿气团进入半岛的屏障,海洋对半岛气候的影响从东南向西北减弱。例如,山东半岛内部年降水量由东南向西北递减,东部在850 mm以上,而位于半岛西北部的滨海平原在600 mm以下[16]。


        山东半岛是中国改革开放以来的边疆地区,是近50年来中国经济最发达的地区之一,人口众多,城市众多。主要的工业企业和城市,如青岛、威海、烟台均位于沿海地区(图1)。由于这些城市设立的气象站更容易受到城市气候条件的影响,因此本文选择半岛中部的水文和气象资料,研究半岛沿海气候与全球气候变化的关系。

 3. Methodology 


3.1. Data 

It was generally acknowledged that human activities began to have a notable impact on climate since 1950’s [2], and taking into account the availability of the data, the last 50 years was selected as the study period.

In this paper, the observation data of temperature and precipitation from hydrological and meteorological stations in the central Peninsula were used to analyze the climate change in Shandong Peninsula for the last 50 years. In which, the precipitation data employed were from 18 hydrological stations, including two indicators of the annual precipitation and precipitation from June to September, and the observation period was from 49 to 58 years (52 years in average); the temperature data including 3 indicators, the annual mean, maximum and minimum temperature respectively, were from 8 meteorological stations. The observation period was from 44 to 58 years (53 years in average). All of the stations mentioned above are the National Basic Weather Stations, and the observation data are continuous data records for the period of the last 50 years, of good quality. As the temperate monsoon climate in the Peninsula is characterized by the synchronization of high temperature and ample precipitation, the 5 climate indicators selected in this paper were the best to reflect its climatic characteristics. The location of the stations was shown in Figure 1.

3.2. Methods


In this study, the non-parametric Mann-Kendall test [17- 20], Accumulated Difference Curve [18] and Order Cluster Analysis (OCA) [21] were applied, to identify the long-term trend and the abrupt change of both precipitation and temperature series in Shandong Peninsula during the last 50 years. The Mann-Kendall test was first used to describe the general change trend for the climatic indicators. An Accumulated Difference Curve method was employed next to analyze the periodical fluctuation of the climate element series. Finally, the transition point of the series was detected by OCA, a method to find the optimal dividing point of the time-series. All of the tests were under the 95% confidence level in this paper, and were calculated by Matlab software.

3.2.1. Mann-Kendall Test

Mann-Kendall test is a non-parametric method, it has the advantage of not assuming any distribution form for the data series, the test capacity is not affected by the breakpoint, and is more powerful than its parametric competitors. Therefore, it is highly recommended by the World Meteorological Organization to assess the monotonic trend in hydro-meteorological time-series. This test is based on the null hypothesis Ho, which supposes that the analyzed series are independent and randomly ordered, and there is no obvious change trend. The test statistics: Zc and β are stated simply as follows.

1) Parameter Zc reflects the general change trend of the series. For the analyzed series (x1, x2 xn)

In which,

Zc follows the standard normal distribution. Therefore, Ho is rejected if |Zc| ≤ Z1−α/2, which means there is a significant change trend for the series (abbreviated as R), while accepting Ho suggests no obvious trend for the series (abbreviated as A). Besides, Zc > 0 represents an upward trend of the series, while Zc < 0 denotes a negative trend. In which, α is the significance level for the test; ±Z1−α/2 are the standard normal deviates. In this paper, α= 95% was applied, ±Z1−α/2 = ±1.96.

2) Parameter β is the Kendall gradient, and it is used to estimate of average change rate of the series. It is based on the assumption that the change trend of the series is monotonic, which means the trend is a linear function of time.

3.2.2. Accumulated Difference Curve

For the series (x1, x2…xn), the accumulat every point was given as:

In which, 

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

___Y1

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值