JIA-2018-09

2109 WANG Qian-feng et al. Journal of Integrative Agriculture 2018, 17(9): 2107–2117 ( P <0.001 and slope=1.02). Furtherly, we calculated the correlation coefficient ( R ) between evaporation pan-PET and MOD16-PET for each station at 11 stations (Table 1), R was high from 0.82 to 0.92 for each station and R was close to 1 for the whole study area, the consistency was better between evaporation pan-PET and MOD16-PET, therefore, the result showed the accuracy. MOD16 PET were potential evaporation data, the data employed in our study were calculated by including transpiration function from plants in land coverage. The ET product is verified by comparing observed ET in North China Plain, and the accuracy of MOD16 ET data is higher (Sun Z et al. 2010), MOD16 ET product based on remote sensing images can provide the ET data with reasonable accuracy (Kim et al. 2012). Fourteen types of land coverage were included with the corresponding land surface variables of fraction of absorbed photosynthetically active radiation (FPAR), leaf area index (LAI), and albedo, or how much sunlight is reflected from the earth at the top of the atmosphere under all-sky condition (both clear and cloudy situations) (Mu et al. 2011). The data provide accurate precision of estimation that serves the requirement of this research (Alemu et al. 2014). To reduce the noise in time and the impact of mixed pixels on the quality of images in space, we used the MOD16 data with monthly temporal resolution and a spatial resolution of 0.05 degrees. Given the significant impact of different types of vegetation coverage on ET, we used the GLC2000 global land cover data set provided by the European Commission’s Joint Research Centre to differentiate the characteristics of spatial-temporal evolution of ET under different vegetation coverages. This dataset was completed collaboratively by more than 30 international research institutions, providing a higher degree of accuracy for interpretation on a global scale (Bartholome et al. 2005). In order to match the spatial resolution of the ET data, the GLC2000 data were re- sampled as a gridded data set at the same spatial resolution of 0.05 degrees, and was masked by a boundary map of the study area. We extracted four types of vegetation with the largest coverage area in the study area, farmland, broadleaved deciduous forest, meadow and needleleaved evergreen forest, which together account for 89.6% of the whole study area (other types of land coverage, representing the remaining 10.4%, are not discussed further here). We used the MODIS data of 2001 and 2014 to analyze the change of vegetation coverage in the study area, and found that there are only 5.3% regions where the types of vegetation coverage changed in our study area. Therefore, there would be only a slight impact of ET caused by the change of vegetation and this was not considered further. 2.2. Study area Located in the northern part of the North China Plain, Hebei Province, 113.45°–119.83°E and 36.08°–42.67°N, borders Inner Mongolia and Liaoning to the north, and Henan and Shandong to the south, and consists of 11 municipal prefectures a total coverage area of 1.89×10 5 km 2 . As one of China’s major producers of agricultural crops and vegetables, Hebei has a humid, semi-arid, and temperate continental monsoon climate, cold and dry in winter and spring and hot and rainy in summer, with typical annual precipitation of averaging between 300–800 mm and notably uneven spatial and temporal distribution. The vegetation coverage in the study area is mainly comprised of farmland, broadleaved deciduous forest, meadow and needleleaved evergreen forest (Fig. 2), and the eleven meteorological stations is distributed in the whole study area (Table 2). 3. Methods 3.1. Time-series variability characteristics method With the time-series variability characteristics method, we can effectively extract the evolution of environmental factors 0 100 200 300 400 0 100 200 300 400 y =1.02 x +39.04 ( P <0.001) y = x (1:1 line) MOD16-PET (mm) Evaporation pan-PET (mm) Fig. 1 The scatter plot between evaporation pan-PET and MOD16-PET. Table 1 Correlation coefficient ( R ) between evaporation pan- PET and MOD16-PET at 11 stations in Hebei Province, China Station name R Station name R Weichang 0.85 Langfang 0.85 Fengning 0.90 Raoyang 0.87 Chengde 0.82 Botou 0.90 Zunhua 0.92 Shijiazhuang 0.83 Yuxian 0.86 Nangong 0.82 Laoting 0.85 All stations 0.86