基于站点尺度通量数据的黑河流域蒸散发产品评价
网络出版日期: 2025-04-29
基金资助
中国科学院战略性先导科技专项 B 类课题项目(XDB0720401);甘肃省杰出青年基金项目(24JRRA076);山西省高等学校教学改革创新项目(J20240022);国家自然科学基金青年基金项目(42301030);阿拉善科技计划项目(AMYY2022-13);2023 年山西省研究生科研创新项目(2023KY111)
Assessment of Evapotranspiration Remote Sensing Products in the Heihe River Basin Based on Station Fluxes Dataset
Online published: 2025-04-29
蒸散发是陆地生态系统能量平衡的重要组成部分,也是大气水循环的关键环节,随着全球气候变暖其变化深刻影响着“地-气”系统的互馈过程,对区域乃至全球气候、生态系统、农业生产活动等具有重要影响。但是由于实际条件限制,目前缺乏对流域尺度不同下垫面类型长时序遥感蒸散发产品表现和适用性的评价,在流域生态系统需水估算、水资源评价与管理使用等方面存在诸多不确定性。因此,本文利用黑河流域不同生态系统地面定位站的长时序通量监测数据,采用随机森林算法插补缺失数据构建多站点、长时序、高精度的地面实际蒸散发数据集(15 个站点,113 个站点年,日尺度),选择 6种常用的遥感蒸散发产品(SSEBop、GLEAM、MOD16、PML_V2、GLASS 和 ETMonitor)并提取各生态系统类型站点所在栅格像元的蒸散发产品年值,通过 R2、RMSE、MAE、Bias 等指标评价了各遥感蒸散发产品在黑河流域的精度及其适用性。结果表明:(1)SSEBop 产品在黑河流域的整体精度最高(R²=0. 63,RMSE=251. 99 mm·a-1),其次是 ETMonitor 产品(R²=0. 26,RMSE=275. 47 mm·a-1),表现最差的是 GLEAM 产品(R²不显著),Bias 最小的是 GLASS 产品,为-22. 57 mm,最大的是 GLEAM 产品,为-317. 49 mm。(2)6 种遥感蒸散发产品在山地森林系统和农田系统的表现相对较好,在荒漠森林系统和荒漠系统的表现最差,而湿地系统的实际蒸散发普遍被低估。其中:SSEBop 产品在除农田系统以外的其他生态类型站点均表现为低估,而 GLASS 产品在荒漠森林系统表现优异但严重高估了荒漠系统的实际蒸散发。(3)站点尺度上,湿地生态系统 ET 最大,约为 1210 mm,荒漠生态系统 ET 最小,约为 180 mm,从下游至上游沿海拔梯度,呈先增加后减少的趋势;本研究通过对黑河流域不同遥感蒸散发产品的精度和适用性进行评价,为干旱区复杂地形、气候和生态系统条件下流域蒸散发模型选取等相关研究工作提供了科学依据,也为流域水资源科学管理使用及生态保护工作提供了参考。
武 博, 高冠龙, 鱼腾飞, 韩 拓, 王麒翔 . 基于站点尺度通量数据的黑河流域蒸散发产品评价[J]. 高原气象, 0 : 1 . DOI: 10.7522/j.issn.1000-0534.2025.00030
Evapotranspiration is an important part of the energy balance of terrestrial ecosystems and a key link in the atmospheric water cycle. However,due to the limitation of practical conditions,there is a lack of evaluation of the performance and applicability of long-time remote sensing evapotranspiration products of different underlying surface types at the basin scale,resulting in many uncertainties in the estimation of water demand of the basin ecosystem,the evaluation of water resources,and the management and use of water resources. Therefore, based on the long-term time-series flux monitoring data of the ground positioning stations of different ecosystems in the Heihe River Basin,this paper uses the random forest algorithm to construct a multi-site,long-time series, high-precision ground actual evapotranspiration dataset(15 stations,113 station years,daily scale),and selects six commonly used remote sensing evapotranspiration products (SSEBop,GLEAM,MOD16,PML_V2, GLASS and ETMonitor)and extracted the annual values of evapotranspiration products of raster pixels in each ecosystem type site,and evaluated the accuracy and applicability of each remote sensing evapotranspiration prod‐ uct in the Heihe River Basin by R²,RMSE,MAE,Bias and other indicators. The results showed that:(1)SSE‐ Bop products had the highest overall accuracy in the Heihe River Basin(R²=0. 63,RMSE=251. 99 mm·a-1),followed by ETMonitor products(R²=0. 26,RMSE=275. 47 mm·a-1),GLEAM products(R² was not significant), GLASS products with the smallest Bias were -22. 57 mm,and GLEAM products were the largest(-317. 49 mm).(2)The performance of the six remote sensing evapotranspiration products in mountain forest system and cropland system was relatively good,and the performance in desert forest system and desert system was the worst,while the actual evapotranspiration of wetland system was generally underestimated. Among them,the SSEBop product was underestimated in all ecological types except the cropland system,while the GLASS product performed well in the desert forest system but seriously overestimated the actual evapotranspiration of the desert system.(3)At the station scale,the ET of wetland ecosystem was the largest,about 1210 mm,and that of desert ecosystem was the smallest,about 180 mm. By evaluating the accuracy and applicability of different re‐ mote sensing evapotranspiration products in the Heihe River Basin,this study provides a scientific basis for the selection of evapotranspiration models in the basin under complex terrain,climate and ecosystem conditions in arid areas,and also provides a reference for the scientific management and use of water resources and ecological protection in the basin.
/
〈 |
|
〉 |