基于GLEAM遥感模型的中国1980－2011年地表蒸散发时空变化

对中国地表蒸散发时空变化的分析，有助于了解气候变化对水循环的影响及对中国水资源的合理配置。该文基于GLEAM（global land-surface evaporation: the Amsterdam methodology）遥感蒸散发模型，通过对GLEAM产品在站点尺度和流域尺度的精度验证以及中国地表蒸散发时空变化特征的研究，得出以下结论：1）GLEAM产品在中国区域满足精度要求，在站点尺度上，GLEAM产品在草原半干旱区的模拟程度最好，海北、内蒙古、当雄3个草原站皮尔逊相关系数（pearson correlation coefficient，CC）均值为0.77（0.65～0.85）；森林站的CC相关系数均值为0.66（0.40～0.85），禹城农田站CC值为0.68；在流域尺度上，海河（相对偏差（relative bias，RB）16.2%）、黄河（RB，15.2%）、西北诸河流域（RB，9.2%）的验证结果精度较好。在绿洲或农灌区降水较少的年份，GLEAM产品符合地表实际蒸散发可能大于降水的规律；2）1980－2011年中国的多年平均蒸散发为18～1 400 mm，空间分布呈从西北向东南方向递增，西北地区多年平均蒸散发最少，海南岛与台湾岛是多年平均蒸散发的极大值区；3）1980－2011年中国平均的年地表蒸散发变化范围为349.7～436.0 mm，多年平均年地表蒸散量为397.5 mm。近32 a中国区域平均地表蒸散发呈显著的上升趋势，上升速率为12.3 mm/(10 a)；4）1980－2011年中国各栅格地表蒸散量变化速率为?86.5～108.7 mm/(10 a)，地表蒸散发减少的面积占28.4%，9.45%的区域地表蒸散发呈明显减少、显著减少及急剧减少趋势，主要位于内蒙古东部、青藏高原西部（新疆西部及东北部、西藏西北部）、甘肃南部等地。地表蒸散发增加的面积占71.6%，18.2%的区域地表蒸散发呈显著增加、急剧增加的趋势，主要位于海河区的河北南部及山东西北部、淮河流域的山东半岛、黄河区的青海东部、长江中下游区的四川东部、山西南部、湖北、湖南、安徽、江西等地、东南诸河区、珠江区及云南南部等；5）各栅格年蒸散发的变化趋势主要由夏季蒸散发变化趋势主导，春季、秋季、冬季对年蒸散发变化趋势的影响较弱。该研究对理解中国气候变化与水资源之间的相互影响具有重要作用，可为中国水资源评价与管理提供参考及决策依据。

Spatio-temporal variability of terrestrial evapotranspiration in China from 1980 to 2011 based on GLEAM data

Abstract: Analysis of spatio-temporal variability of terrestrial evapotranspiration (ET) over China facilitates understanding the response of water cycle to global climate change and water resources rational allocation. However, large-scale ET estimation methods based on satellite-based observations have some uncertainty. In this study, we firstly validated the accuracy and applicability of ET data from Global Land-surface Evaporation: the Amsterdam Methodology (GLEAM) using in-situ observations from China FLUX at point scale. Secondly, based on water balance model, we also validated the GLEAM ET data over China and Chinese 10 sub-basins using hydrological data at basin scale. In addition, we obtained the 32-year spatio-temporal datasets of terrestrial ET over China from 1980 to 2011 using GLEAM ET data. The research results indicated that: 1) GLEAM ET data were basically reasonable over China after the validation at both point and basin scales. Point scale validation using daily data revealed that GLEAM ET data showed overall comparable daily accuracies. GLEAM ET data showed the highest performance for three Grassland sites, which had the highest CC (correlation coefficient) mean value 0.77 (0.65-0.85) and lower relative bias (RB), root mean squared error (RMSE), mean absolute error (MAE) value. GLEAM ET data showed high performance for forest sites except XSBN site, which had CC mean value 0.66 (0.40-0.85) and had a low performance at XSBN (CC, 0.4, RMSE, 1.14 mm). GLEAM ET data also showed high performance for cropland (CC, 0.68; RB, 11.73%). Validation results at basin scale over 2003-2011 using water balance ET data as a reference indicated reasonable accuracies for GLEAM ET data, especially in Haihe River basin (RB, 16.2%), Yellow River basin (RB, 15.2%) and the Northwest River basin (RB, 9.2%). GLEAM ET value may be greater than precipitation value in oasis or agricultural irrigation area due to the artificial irrigation increased the soil moisture content in relative drought years. 2) GLEAM ET data showed that the climatological annual mean ET value varied spatially from 18 to 1400 mm during the 1980-2011 period over China. Spatial patterns of mean annual ET showed an increase from the northwest to the southeast part with the smallest ET value over Northwest region and the largest ET value over Hainan Island and Taiwan Island. 3) The domain-averaged annual ET over China showed an inter-annual variability ranging from 349.7 to 436.0 mm with the climatological value of 397.5 mm and a significant increasing trend of 12.3 mm per 10 years. 4) The trend in grid-based GLEAM ET over China displayed a prominent spatial variability ranging from -86.5 to 108.7 mm/(10a). The percentage of areas with decreased trends was 28.4%, and the significantly decreased areas was 9.45% of the total, which were mainly distributed in east Nei menggu, west Tibetan plateau (west and northeast Xinjiang, northwest Xizang), south Gansu. The areas with increased trends were 71.6% of the total and the significantly increased regions were 18.2%, which were mainly distributed in south Hebei and northwest Shandong (Haihe River basin), Shandong peninsula in Huai River basin, east Qinghai in the Yellow River basin, east Sichuan, south Shanxi, Hubei, Hunan, Anhui, and Jiangxi (Middle-lower Yangze River), Southeast River basin, Pearl River basin and south Yunnan et al. 5) The trends in grid-based GLEAM ET over China showed a remarkable seasonal cycle with the largest value in summer. The trend in the spring, autumn and winter time ET was not significant. Results from this research probably played an important role in understanding the interaction between climate change and water resources and provided reference and decision making base for China's water resources evaluation and management.