论文部分内容阅读
干球温度(气温)是地面气象观测中所要测定的常规要素之一。目前基于遥感数据获取该量的方法多采用线性拟合或直接应用遥感反演的温度近似代替干球温度,但是由于下垫面复杂,导致误差较大。本文提出用支持向量机(SVM)模型进行干球温度推算。选择广西省南宁市为研究区域,首先通过遥感反演温度与气象实测温度的对比,证明了利用遥感数据推算干球温度的可能性。然后,构建了针对干球温度的SVM推算模型。最后,尝试了分别使用表观亮温和遥感反演地温作为SVM模型的输入进行干球温度的推算。结果表明,SVM模型推算的干球温度与实测值更为接近,和传统方法相比,精度得到明显提高;且用表观亮温进行推算更为简单,更适合业务化的应用。
Dry bulb temperature (air temperature) is one of the common elements to be measured in surface meteorological observations. At present, the method of obtaining this amount based on remote sensing data mostly adopts the linear fitting or directly applies the temperature of the remote sensing inversion instead of the dry-bulb temperature, but due to the complicated underlying surface, the error is large. In this paper, the support vector machine (SVM) model is proposed to calculate the dry-bulb temperature. Choosing Nanning City, Guangxi Province as the research area, the possibility of using the remote sensing data to predict the dry-bulb temperature is first demonstrated by comparing the retrieved temperature with the measured temperature. Then, the SVM model for the dry-bulb temperature is constructed. Finally, we try to estimate the dry-bulb temperature by using apparent bright-temperature and remote-sensing ground temperature respectively as the input of SVM model. The results show that the dry bulb temperature predicted by the SVM model is closer to the measured value, and the precision is obviously improved compared with the traditional method. The calculation using the apparent brightness temperature is simpler and more suitable for the application of business.