Identifying critical source areas for water quality: 1. Mapping and validating transport areas in three headwater catchments in Otago, New Zealand

Abstract

Summary Validity of five empirical to process-based, hydrological models described by Srinivasan and McDowell (2007) in mapping transport areas was tested in the Invermay and Glenomaru headwater catchments in Otago, New Zealand. These transport areas together with contaminant source areas form critical source areas (CSAs), where the majority of contaminant loss occurs and therefore represent areas where mitigation potential would be most efficient. Rainfall and 15-min instantaneous surface flows at the catchment outlets and the shallow water table ( 10 m from the stream during storm events. There appeared to be an active subsurface (shallow) flow system transferring flows from land to streams. However, during many rainfall events, semi-pervious areas like fence lines, animal tracks and gateways were connected to the stream via infiltration-excess surface runoff, as measured by surface runoff samplers. This may be a significant for contaminant transfer given the amount of time spent by animals on these areas and deposition of contaminants (e.g., in dung) and is explored in a companion paper. Of the approaches used to delineate transport areas during storms, generally the empirically-based curve number model and Phosphorus Index over-predicted stormflow areas. The drainage density approach predicted stormflow areas well, but under-predicted flow volumes by 55–82%. The physically-based topographic index (TI) model and a model combining TI model and surface runoff from semi-pervious areas predicted flow volumes within 6% of that observed, but the extent of predicted transport areas greatly differed from those observed. Among the five approaches, the process-based approaches were identified to be more applicable and expandable for future prediction of CSAs for contaminant loss. In contrast, the other approaches offered little flexibility in representing the dynamic hydrology of these catchments. However, improved input data (elevation, soil properties) may improve the applicability these models.