JIA-2018-09

2110 WANG Qian-feng et al. Journal of Integrative Agriculture 2018, 17(9): 2107–2117 in temporal and identify inter-annual and intra-annual ET variations and abnormalities. In this research, we used the monthly ET data from January 2001 to December 2014 as the data sequence, and the mean value and standard deviations of the monthly grid-based ET rates in statistical units as the statistical index. The statistical units were divided over the whole study area and the areas with the four types of vegetation coverage. In this way, we studied the time-series variability characteristics of each sub-region. 3.2. Trend analysis approach Mann-Kendall (MK) trend test is a nonparametric test method (Mann 1945; Kendall 1948). This method does not require a normal distribution of data, and is widely used in the field of meteorology, ecology, and agriculture (Burn et al. 2002; Xu et al. 2004; Wang et al. 2015). The null hypothesis H0 is that the data series x k ( k =1, 2, 3... n ) are independent from one another and have the same distribution, and the alternative hypothesis H1 is that there is a monotonic trend in the data series. The MK trend test is calculated as follows: 1 sgn( ) n n j i i =1 i = j +1 S = x x − − ∑ ∑ (1) Where, x j is the sequential data value, n is the size of the dataset, and sgn is calculated as follows: 1 sgn( )= 0 1 j i j i j i j i if x x x x if x x if x x > − = − < (2) According to Mann (1945) and Kendall (1948), when n ≥8, the test statistics S is approximately normally distributed with the mean and variance as follows: E ( S )=0 (3) ( m −1)(2 m +5) 18 n m m =1 n ( n −1)(2 n +5)− t m V ( S )= ∑ (4) Where, t m is the number of extent m . The standardized test statistics Z is calculated using the following formula: 0 S −1 S >0 V ( S ) Z = S =0 S +1 S <0 V ( S ) (5) | Z α |=1.65, 1.96, and 2.58, which correspond to the critical values at the significance level P =0.1, 0.05, and 0.01, respectively. If | Z |>| Z α |, the null hypothesis H0 is rejected and values of P =0.05 and 0.01 were considered in our research. To analyze the trend of the time series variable, we used 140°E 115°E 116°E 117°E 118°E 119°E 140°E 115°E 116°E 117°E 118°E 119°E 140°E 42°N 41°N 40°N 39°N 38°N 37°N 42°N 41°N 40°N 39°N 38°N 37°N 42°N 41°N 40°N 39°N 38°N 37°N 42°N 41°N 40°N 39°N 38°N 37°N 115°E 116°E 117°E 118°E 119°E 140°E 115°E 116°E 117°E 118°E 119°E Station Province boundary City boundary Needleleaved evergreen forest Broadleaved deciduous forest Meadow Farmland Other land types Fig. 2 Study area of Hebei Province, China and types of vegetation coverage in 2000. Table 2 Basic information table of 11 meteorological stations in Hebei Province, China Station name Longitude (°E) Latitude (°N) Station name Longitude (°E) Latitude (°N) Yuxian 114.57 39.83 Langfang 116.40 39.17 Shijiazhuang 114.35 38.07 Laoting 118.88 39.43 Fengning 116.63 41.20 Raoyang 115.73 38.23 Weichang 117.77 41.97 Botou 116.55 38.08 Chengde 117.92 40.97 Nangong 115.38 37.37 Zunhua 117.95 40.20