论文部分内容阅读
论文利用青海东部农业区内的12个气象站2003—2005年气象资料,应用Penman-Monteith公式计算得到各站逐旬的ET_0值,并进一步研究其与高程(DEM)、地表温度(LST)及归一化植被指数(NDVI)3个因子之间的关系,提取遥感数据并耦合到时间分辨率为旬,空间分辨率为1 km,将其与计算所得ET_0建立多元反演模型。由于3个自变量因子之间存在着很强的相关关系,LST与NDVI间判定系数R2平均在0.7左右,不能直接用最小二乘回归方法建立模型。为有效避免自变量间相关性对模型的影响,研究中采用岭估计方法建模。结果表明,通过岭估计建立2003年10~33旬区域二元模型反演最低精度达76.19%,区域三元模型反演最低精度也有83.54%。与传统方法所建模型相比,检查点均方根误差减小约1.1,反演最低精度提高11%左右,能满足实际应用需求。
Based on the meteorological data of 12 meteorological stations in the agricultural region of eastern Qinghai from 2003 to 2005, the ET-0 values of each station were calculated by Penman-Monteith formula and further compared with the DEM, LST and (NDVI), the remote sensing data was extracted and coupled to a temporal resolution of 10 days, with a spatial resolution of 1 km, to establish a multivariate inversion model with the calculated ET_0. Because there is a strong correlation between the three independent variables, the coefficient of determination R2 between LST and NDVI is on the average about 0.7, and the model can not be directly established by the least-squares regression method. In order to effectively avoid the influence of the correlation between independent variables on the model, the ridge estimation method was used in the study. The results show that the minimum accuracy of the regional binary model inversion from 2003 to 2007 is 76.19% by ridge estimation, and the minimum accuracy of regional ternary model inversion is 83.54%. Compared with the traditional method, the root-mean-square error of the checkpoint is reduced by about 1.1 and the minimum precision of inversion is increased by about 11%, which can meet the practical application requirements.