JIA-2018-09

2108 WANG Qian-feng et al. Journal of Integrative Agriculture 2018, 17(9): 2107–2117 important component of hydrologic and climate changes in the biosphere (Seevers et al. 1994), accounting for the movement of surface or subsurface water through soil evaporation and plant transpiration (Rey 1999) due to internal (biological) and external (physical and atmospheric) factors (Stoy et al. 2006). These factors contribute to spatial-temporal heterogeneity and the evolution of ET due to combined effects of climate change and ecosystem management (Wear and Greis 2002; Foley et al. 2003). Therefore, in order to response to climate change, analyzing the spatial-temporal evolution of ET in different complex ecosystems is essential to understand and reshape water resource cycles in region scale. Currently, evapotranspiration (ET) can be measured by several methods. ET at the global scale can be measured by using land-surface parameters, such as net land surface radiation, temperature, vegetation index, and soil moisture (Wang and Liang 2008), and the ET of natural vegetation and crop can be simulated using water balance or crop growth models (Consoli and Vanella 2014; Wang et al. 2017). Satellite remote sensing is widely adopted to estimate ET using vegetation index (VI) alone (Parmar and Gontia 2016; Zhang et al. 2016; Li et al. 2017), and together with other land-surface parameters (e.g., vegetation coefficient, crop coefficient) (Ghilain et al. 2014; Lei and Yang 2014; Howes et al. 2015; Liu 2016). Remote sensing-based methods are cost-effective, efficient, and provide up-to-date information, and these methods are used to estimate ET in this research. Different types of vegetation (forest, grass or cropland) have diverse ET rates and structures of canopy covers (Suzuki et al. 1998). For different categories of vegetation, there may be changes in ET rate in response to the surrounding environment. For example, increasing carbon dioxide in grassland ecosystems can lead to a decrease in ET (Polley et al. 2011). Against the background of climate change, the conversion from forests, shrublands, to grasslands in southern China during 1967 to 1993 resulted in an increase in ET (Sun G et al. 2010). Climate change has an even stronger impact on ET, as climate and vegetation coverage were shown to respectively account for 85.18 and 14.82% of the impact on ET in Yunnan of China (Yang et al. 2015). Consequently, there is an urgent need to explore the spatial-temporal evolution of ET driven by environmental factors in agricultural areas to allow effective implementation of regional climate change adaptation and vegetation coverage projects at the beginning of the 21st century. The shortage of fresh water resource is one of the major factors adversely affecting the sustainability of the ecological system, and social and economic development in the study area (Maas et al. 2017; Zhou et al . 2017). Due to exploding population growth and global warming, soil salinization associated with irrigation is worsening (Yeo 1998). According to the statistical research of the Hebei Academy of Agriculture and Forestry, China, the area of saline and alkaline soil in Hebei is 7.8×10 3 km 2 , accounting for 10.4% of the cultivated land of the province. The regions with soil salinization are mainly in the low and coastal plains in the eastern part of Hebei as a result of farm irrigation, overgrazing of existing meadows, and the overexploitation of groundwater in superficial zones. In this study, the main land coverage types in Hebei are considered as vegetation, vegetation ET is a main driver of water consumption and salinization. In this research, we used the time-series data set from the MODIS remote sensing product MOD16 for the heavily salinized agricultural areas in Hebei to analyze the variation of ET for different types of vegetation for the years 2000–2014. We explored the characteristics of spatial heterogeneity, patterns, and variability of ET using spatial pattern method. The research provides a certain basis of reference for the reasonable allocation of water resources of effective regional climate change strategies. 2. Data source and study area 2.1. Data source The ET data used in this research were the land surface ET data with spatial resolution of 1 km under the MODIS sensor covering Hebei Province (MOD16), as provided by Numberical Terradynamic Simulation Group of the University of Montana, USA (http://www.ntsg.umt.edu/ project/mod16#data-product). The temporal resolution of MOD16 data were categorized at 8-day, monthly and annual intervals, using the Penman-Monteith estimation model of ET. In addition, we also applied time series of meteorological data including observed ET factor from the China Meteorological Data Network, covering 11 weather stations across the whole study area, we converted the daily ET data into monthly by summing the daily data in each month at each station from 2000 to 2014. The observed ET data were obtained from evaporation pan, we called observed ET as evaporation pan-PET (potential evaporation). The evaporation pan-PET data only include evaporates from water body without transpiration from the plant, thus the pan-PET data can be used as auxiliary data. In order to investigate the accuracy of MOD16, we compared the consistency between evaporation pan-PET and MOD16-PET (Fig. 1), we also conducted the scatter plot and the linear fitting between evaporation pan-PET and MOD16-PET, the performance of the linear fitting is better