EFFECT OF CATCHMENT-SCALE SUBSURFACE MIXING ON STREAM ISOTOPIC RESPONSE

A 3.8-ha watershed on the west coast of New Zealand was instrumented with suction lysimeters and automatic water samplers to determine the relationship between subsurface isotopic and chemical concentrations to those of rainfall and resulting streamflow. A t test showed that +/- 2 parts per thousand represented a significant difference between successive sample deuterium values. Eleven rainfall episodes were subdivided into two categories: (1) two events where stream isotopic composition did not deflect > 2 parts per thousand from prestorm values, and (2) four events which demonstrated new water flushing. Detailed analysis of one 47-mm rainfall (9.8-mm runoff) event showed that old water dominated streamwater exiting the watershed by 90% using a standard two-component hydrograph separation for deuterium (corroborated by Cl and electrical conductivity). Three-component hydrograph separation indicated that 12-16% was in the form of soil water, with < 5% as on-channel precipitation and 80% groundwater. Analysis of over 1000 water samples revealed systematic trends in soil water and groundwater isotopic composition both in a downslope and downprofile direction. Between-storm suction lysimeter deuterium data showed a systematic dampened response to temporally variable rainfall deuterium concentrations. Multivariate cluster analysis revealed three distinct soil water/groundwater groupings, with respect to soil depth and geographic position within the watershed. Within-storm suction lysimeter sampling preserved similar groupings, indicating that the subsurface reservoir is poorly mixed on short time scales. Understanding subsurface mixing response to rainfall should greatly improve models of episodic stream response and partitioning of storm flow into waters of different age.