时空CoKriging的变异函数建模

Variogram modeling in space-time CoKriging

在对地观测中,所研究的地学变量不仅具有时间、空间特征,还受其他变量的影响,采用多元时空相关数据,可以提高时空估值的精度.时空CoKriging是多元时空插值中一种常用的方法,建立时空变异函数和协变异函数是时空CoKriging插值的关键一步.以东北三省为试验区,利用该地区气象站观测数据的月平均空气相对湿度为主变量,引入同时间同位置的月平均空气温度作为协变量,对空气相对湿度和空气温度进行时空变异函数和时空协变异函数建模.实验结果表明,采用和度量时空模型的时空变异函数的随机性空间结构建模的实际拟合效果较好.

The variables used in various environmental studies not only have the spatial temporal dependence, but also are affected by other variables. However, the multivariate space-time correlated data can be used to improve the accuracy of space-time prediction and interpolation. The space-time CoKriging is a commonly used method of multivariate space-time interpolation, with modeling a space-time direct variogram and a cross variogram as a key step. Therefore, taking the three provinces in Northeastern China as study area. this paper modeled the space-time direct variograms and cross variograms for air relative humidity and air temperature by using the monthly mean relative air humidity observed in meteorological stations as a main variable and the air temperature in the same position and at the same time as a covariate variable. Inverse distance weighted interpolation was used to get the current meteorological observation data. The experi mental results show that the space-time variogram using sum-metric model is well fitted, compared to the deterministic structure modeling spatial distance inversely.