地表通量均衡法对中国陆面蒸散发估算的适用性评价

为验证和评估地表通量均衡法（surface flux equilibrium, SFE）在大尺度上的估算精度，基于SFE方法估算中国区域2001—2015年日尺度陆面蒸散发量，并开展适用性评价。结果表明：在站点尺度上，SFE估算值与观测值具有较好的一致性（均方根误差为1.03 mm/d，相关系数为0.70），优于GLEAM产品精度（均方根误差为1.19 mm/d，相关系数为0.62）；在流域尺度上，SFE方法与水量平衡蒸散发较为接近，对多数流域的模拟效果较好，优于GLEAM产品和CR产品；在时空变化趋势上，SFE产品与CR产品的变化趋势一致性较高，与GLEAM产品一致性较差。综合SFE估算结果的精度验证和趋势分析，可以发现SFE方法基本达到现有产品的精度，因其输入参数少、物理机制明确、参数化方案简单，有潜力为蒸散发的大尺度准确估算提供一套行之有效的替代方案。

Evaluation of surface flux equilibrium in land evapotranspiration estimation in China

Evapotranspiration (ET) plays a key role in the surface energy balance and water cycle, and accurate estimation of ET is crucial for agricultural irrigation, climate change prediction, and water resources management and utilization. However, it is still challenging to accurately estimate ET due to its dependence on heterogeneous land surface features, which can rise the complexity of the model parameterization scheme and substantial disagreement between models. Recently, surface flux equilibrium (SFE) method was proposed based on a strong physical mechanism to estimate ET with only standard meteorological station atmospheric observations. Therefore, it has superiority in large-scale and long-time ET estimation for requiring fewer input parameters compared with other methods. Considering the lack of related research work at present, the applicability of SFE method still needs to be validated and evaluated across a broad range of land conditions.CMFD (China Meteorological Forcing Dataset) and GLASS (The Global Land Surface Satellite) net radiation data from 2001 to 2015 were used to estimate daily ET in China based on SFE method. The ET estimations were validated with eddy covariance measurements at 8 ChinaFLUX sites and ET data of water balance at the basin scale, respectively. Moreover, the results were also compared with two types of ET products, including GLEAM (Global Land-surface Evaporation: the Amsterdam Methodology) and CR (Complementary Relationship). Based on the validation of estimation accuracy and the analysis of temporal and spatial variation trends, the applicability evaluation of SFE method was developed for ET estimation in China. The results showed that at a site scale, the daily ET estimated by SFE was in good agreement with the ET observations ( r=0.70, ERMS=1.03 mm/d), which was a little better than the accuracy of GLEAM product ( r=0.62, ERMS=1.19 mm/d). Moreover, SFE method had better performances at the sites covered with forest and grass than those covered with the crop. At the basin scale, the ET estimated by SFE was generally close to the ET of water balance, with a correlation coefficient of 0.96. Specifically, SFE method performed best in the Pearl River basin with a relative deviation of 1.63%. In addition, the accuracy of SFE was better than GLEAM product and CR product in most basins. Based on the spatial and temporal variation trend analysis, it could be found that the interannual variation of SFE and CR products presented a slight decreasing trend from 2001 to 2015 and their variation trend had well consistency on spatial and temporal scales. In contrast, GLEAM product showed a significant increasing trend and the spatial distribution of its variation trend was quite different from the other two ET results. In conclusion, SFE method performed well in the estimation of daily ET in China at the site and basin scale. Based on the verification and trend analysis of SFE results, it could be found that SFE method can achieve the accuracy of existing products. However, SFE method has systematical overestimation when ET is low and underestimation when it is high, especially in the cropland. Apart from the applicability problem in extreme water conditions of SFE method, net radiation data is another source of errors in ET estimation accuracy. Although the estimation accuracy varied across different surface conditions, the prediction errors of SFE method were comparable to errors of ET products. Besides, SFE method has the potential to provide a set of effective alternatives for the accurate estimation of ET because of its few input parameters, strong physical mechanism and simple parameterization scheme.