本文将随机水文学和多变量地质统计学相结合 ,提出了多变量时间剩余协克立格模型。该模型将水文变量的周期性成分和趋势性成分视为确定性分量从原始各测站水文时间序列中剔除 ,剔除剩下的剩余值属于时空随机函数 ,从而以这种剩余值作为研究变量去刻划原水文变量的时空相关特征。该模型可用于水文变量的插补延展和短期预报 ,具有较高的精度 In this paper, we combine the random hydrology and multivariate geostatistics, and propose a multivariate co-Lattice model. In this model, the periodic and trend components of hydrological variables are treated as deterministic components and removed from the hydrological time series of original stations. The remaining residual values ​​are classified as spatiotemporal random functions, and the residual values ​​are used as research variables Scoring the Temporal and Spatial Features of Original Hydrological Variables. The model can be used for interpolation extension and short-term forecast of hydrological variables with high accuracy