Geostatistical investigation into the temporal evolution of spatial structure in a shallow water table

Abstract. Shallow water tables near-streams often lead to saturated, overland flow generating areas in catchments in humid climates. While these saturated areas are assumed to be principal biogeochemical hot-spots and important for issues such as non-point pollution sources, the spatial and temporal behavior of shallow water tables, and associated saturated areas, is not completely understood. This study demonstrates how geostatistical methods can be used to characterize the spatial and temporal variation of the shallow water table for the near-stream region. Event-based and seasonal changes in the spatial structure of the shallow water table, which influences the spatial pattern of surface saturation and related runoff generation, can be identified and used in conjunction to characterize the hydrology of an area. This is accomplished through semivariogram analysis and indicator kriging to produce maps combining soft data (i.e., proxy information to the variable of interest) representing general shallow water table patterns with hard data (i.e., actual measurements) that represent variation in the spatial structure of the shallow water table per rainfall event. The area used was a hillslope in the Catskill Mountains region of New York State. The shallow water table was monitored for a 120 m×180 m near-stream region at 44 sampling locations on 15-min intervals. Outflow of the area was measured at the same time interval. These data were analyzed at a short time interval (15 min) and at a long time interval (months) to characterize the changes in the hydrologic behavior of the hillslope. Indicator semivariograms based on binary-transformed ground water table data (i.e., 1 if exceeding the time-variable median depth to water table and 0 if not) were created for both short and long time intervals. For the short time interval, the indicator semivariograms showed a high degree of spatial structure in the shallow water table for the spring, with increased range during many rain events. During the summer, when evaporation exceeds precipitation, the ranges of the indicator semivariograms decreased during rainfall events due to isolated responses in the water table. For the longer, monthly time interval, semivariograms exhibited higher sills and shorter ranges during spring and lower sills and longer ranges during the summer. For this long time interval, there was a good correlation between probability of exceeding the time-variable median water table and the soil topographical wetness index during the spring. Indicator kriging incorporating both the short and long time interval structure of the shallow water table (hard and soft data, respectively) provided more realistic maps that agreed better with actual observations than the hard data alone. This technique to represent both event-based and seasonal trends incorporates the hillslope-scale hydrological processes to capture significant patterns in the shallow water table. Geostatistical analysis of the spatial and temporal evolution of the shallow water table gives information about the formation of saturated areas important in the understanding hydrological processes working at this and other hillslopes.