青海省地表净辐射通量的遥感估算方法及时空特征分析

Estimation of Land Surface Net Radiation Flux based on Remote Sensing and Analysis of its Spatial-temporal Characteristics in Qinghai Province

地表净辐射通量是地气系统的关键参数,对其准确估算并进行时空特征分析是地表水热平衡研究的重要内容。本文基于2011-2019年的MODIS产品分别估算了青海省晴天和有云条件下卫星过境时刻的瞬时地表净辐射通量,并依据地表净辐射通量的日变化模型计算得到日间地表净辐射通量。在精度评价方面,利用青海省西宁和格尔木两站的辐射通量数据和50个站的气象观测数据,分别验证了瞬时和日间地表净辐射通量的遥感估算精度。结果表明,晴天条件下瞬时和日间地表净辐射通量在西宁站估算精度较高,RMSE分别达到72.35 W·m^(-2)和2.29 MJ·m^(-2),在格尔木站存在明显的高估,Bias分别达到133 W·m^(-2)和3.25 MJ·m^(-2)。在各气象站点处,遥感估算的日间地表净辐射通量有较高的精度,RMSE为2.04 MJ·m^(-2)。青海省晴天条件下地表净辐射通量的年均值为4176.62 MJ·m^(-2),年内变化呈单峰分布,最大值和最小值分别出现在6月(559.23 MJ·m^(-2))和12月(134.68 MJ·m^(-2)),其空间分布总体呈现东南高西北低的格局。

Land surface net radiation flux(Rn)is an important parameter of energy and water exchange processes in land-atmosphere interactions.Accurate estimation of spatial-temporal characteristics of Rn is useful in a wide range of applications,such as hydrology,meteorology,ecology and agriculture.In this paper,we presented a framework to estimate instantaneous and daytime Rn under all sky conditions by using MODIS products from 2011 to 2019 in Qinghai Province.The accuracy of instantaneous and daytime Rn estimation was verified by using the radiation flux observations in Xining and Golmud and FAO-56-based estimation across 50 meteorological stations,respectively.The results showed that the root mean square error(RMSE)of instantaneous and daytime Rn estimated under clear sky conditions in Xining station,in comparison of ground-based measurements was 72.35 W·m^(-2) and 2.29 MJ·m^(-2),respectively.However,significant overestimates were observed in Golmud station with Bias of 133.19 W·m^(-2) and 3.25 MJ·m^(-2).Judging from daytime Rn,there was a good agreement between FAO-56-based estimation and our estimation with a RMSE of 2.04 MJ·m^(-2) across 50 meteorological stations.In terms of the monthly and seasonal variations of Rn in Qinghai Province,the annual average value was 4176.62 MJ·m^(-2),and its monthly variation showed a single peak distribution,with the maximum and minimum values in June(559.23 MJ·m^(-2))and December(134.68 MJ·m^(-2)),respectively.The annual variation of Rn presented a fluctuant decreasing trend from 2011 to 2019,and significantly decreased about 1.16%from 2012 to 2013,probably due to climate change and human activity.For example,Rn decreased with increasing land surface temperature and decreasing fractional vegetation cover from 2012 to 2016.In Qinghai Province,the regions with larger Rn were mainly located in the southeast,while regions with lower Rn were mainly located in the northwest.The spatial distribution characteristics of Rn had significant relation to various factors,including altitude,surface albedo and fractional vegetation cover.Specifically,there was a good positive correlation between the spatial distribution characteristics of altitude and that of Rn in Qinghai Province.In consideration of the similar overestimation in FAO-56-based Rn,the overestimation of instantaneous and daytime Rn in Golmud station may result from the original error of radiation flux observations.Compared with results in clear sky,the accuracy of daytime Rn estimation was lower under cloudy sky conditions(RMSE=4.37 MJ·m^(-2)).The accuracy of monthly and annual mean Rn,however,had better accuracy with the RMSE of 42.59 MJ/m2 and 266.35 MJ·m^(-2).There are mainly three possible reasons of low accuracy in cloudy sky,including the error of applying data and model under cloudy conditions,the mismatch of remote sensing data and ground observation data in temporal and spatial scales,the effect of clouds on estimation accuracy of daytime Rn in the time scale expansion of instantaneous Rn.