AMSR-E遥感土壤湿度产品在青藏高原地区的适用性

卫星遥感土壤湿度产品有广泛的应用前景,但其精度已引起学术界的高度关注。针对被动微波遥感反演土壤湿度的准确性问题,该文对比分析了3种国际上比较广泛关注的高级微波扫描辐射计/地球观测系统(AMSR-E/EOS)土壤湿度产品(JAXA,NASA及VUA)。首先,利用试验观测数据评价了3种土壤湿度产品的精度,分析了不同植被覆盖和降水对被动微波遥感反演土壤湿度精度的影响。结果表明:在平坦裸露地表,被动微波遥感反演土壤湿度具有较高精度,卫星降轨数据估算土壤湿度与实测值相关系数大于0.7,均方根误差小于0.16,但在高密度植被区域误差较大,相关系数小于0.7,均方根误差最大可达到0.2。然后,分析了降水发生时刻土壤湿度产品的精度,结果表明:3种产品精度均有不同程度下降,但NASA产品的相关系数仍然能够达到0.69。在此基础上,基于青藏高原土壤湿度观测网,制作了青藏高原地区土壤湿度时空分布图,对比了3种产品对青藏高原地区土壤湿度时空分布特征描述的准确性,分析了其适应性,发现NASA与VUA产品在空间分布上符合青藏高原地区土壤湿度的空间分布特征,但在土壤湿度值的变化范围上仍均存在较大误差,在具体的应用中,可以利用实测数据对产品进行线性回归校正,以消除系统性误差。该研究可为基于AMSR-E土壤湿度产品的应用与研究提供参考。

Applicability evaluation of AMSR-E remote sensing soil moisture products in Qinghai-Tibet plateau

Abstract: Soil moisture products generated from satellite remote sensing have wide applications, but their accuracy has drawn great attention in the relevant research studies. In this paper, to evaluate the accuracy of soil moisture estimated by passive microwave remote sensing, three kinds of Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) soil moisture products have been inter-compared and analyzed. First, the ground measured soil moisture data was used for evaluating the precision of three kinds of soil moisture products and analyzing the effects of different vegetation coverage and precipitation on passive microwave soil moisture estimates. For this purpose, four soil moisture observation networks, which were displayed in the Qinghai-Tibet Plateau, were taken into account to ensure that this experiment considered most land cover types. In each network, at least six field measurement points were used to calculate the mean value of the corresponding pixels to reduce the deviation caused by scale effect. The results showed that the estimated soil moisture in a flat and bare soil area was more accurate with its correlation coefficient above 0.7 and RMSE below 0.16, while in the high vegetation density area, the correlation coefficient was less than 0.7 and RMSE could reach to 0.2. In general, NASA's soil moisture product performs well in all land cover types with stable relativity. VUA's soil moisture product has less RMSE in high soil moisture areas. Otherwise, JAXA's soil moisture product's relativity rises in low soil moisture areas. Especially in the area where the soil moisture was less than 0.12m3/m3, JAXA's soil moisture product had the highest correlation coefficient of three kinds of soil moisture products. It showed potential for drought monitoring. When precipitation occurs, the accurate of the three kinds of soil moisture products all decreased to different degrees. However, NASA's product sustained preferable stability with a 0.69 correlation coefficient. Based on this, the soil moisture maps of the Qinghai-Tibet Plateau were generated to analyze the space-temporal distributed characteristics of the three kinds of soil moisture products. It proved that NASA's and VUA's soil moisture products were in solid agreement with the actual situation. In addition, the monthly variations of NASA's and VUA's soil moisture were basically the same with the monitoring results of the China Meteorological Administration. But the magnitudes of these two kinds of soil moisture products had a great difference with the actual situation, while JAXA's soil moisture product was not in accordance with the reality in spatial distribution in the northwest and southeast of Qinghai-Tibet Plateau.