Method For Spatial-temporal Variability Analysis With Application In Precipitation

Spatial-temporal variability is common in hydrologic phenomena, such asprecipitation, evaporation, groundwater and salt, and so on. Traditional approach usedto study these hydrologic phenomena is to analyze the time variability and the spacevariability separately with time series analysis method and geostatistics, whichconsiders less the relation between spatial variability and temporal variation.Therefore, it is important to describe both their spatial and temporal variability.Existing methods can not solve the problem very well, so it is necessary to makefurther studies of the analytical methods for this spatial-temporal variability.We present a new approach to analyze the spatial-temporal variability, named astime-space analysis method. First, the temporal characteristic parameters ofhydrologic series in the study region are calculated. Then these parameters areregarded as regionalized variable and analyzed with geostatistics method and theirspatial variability is obtained. As an application, annual precipitation series from 1960to 2000 in the middle and east of China are studied.Four characteristic parameters, mean value, variation coefficient, skewnesscoefficient and first-order autocorrelation coefficient, of these precipitation series areestimated with classical statistics. These parameters are regarded as regionalizedvariables and their space distributions are analyzed. The stationarity of these fourparameters are studied. Considering isotropy with and without drift and anisotropyremoving drift, the spatial variability and distributions of the four parameters arestudied with geostatistics methods. The correlations of the four parameters areanalyzed. By using the method of co-kriging, the multivariate analysis of parameterswith significant correlations is made.The results show that annual precipitation in the study region decreases fromsoutheast to northwest. The variation coefficient is the highest in the northwest region,higher in the middle region and lowerest in the south region. The skewnesscoefficient is positive in most part of the study region. The first-order autocorrelationcoefficient is generally from -0.25 to 0.3.The results of the application show that the proposed time-space analysismethod can describe the spatial-temporal variability of hydrologic variables. Themain characteristic parameters are used to represent the temporal variability ofhydrologic phenomena and can be analyzed with geostatistics method. Thus the maincharacteristics of the hydrologic series in the study region can be obtained. Thismethod can be used to solve the hydrological problem of areas lacking data.Moreover, the spatial distribution of temporal characteristic parameters can beobtained, which is helpful to know the main characteristic of the hydrologic series ofthe whole study region.