From points to patterns - Transferring point scale groundwater measurements to catchment scale response patterns using time series modeling

Abstract:

Detailed groundwater measurements across a catchment can provide information on subsurface stormflow generation and hydrologic connectivity of hillslopes to the stream network. However, groundwater dynamics can be highly variable in space and time, especially in steep headwater catchments. Prediction of groundwater response patterns at non-monitored sites requires transferring point scale information to the catchment scale through analysis of continuous groundwater level time series and their relationships to covariates such as topographic indices or landscape position. We applied time series analysis to a 4 year dataset of continuous groundwater level data for 51 wells distributed across a 20 ha pre-alpine headwater catchment in Switzerland to address the following questions: 1) Is the similarity or difference between the groundwater time series related to landscape position? 2) How does the relationship between groundwater dynamics and landscape position change across long (seasonal) and shorter (event) time scales and varying antecedent wetness conditions? 3) How can time series modeling be used to predict groundwater responses at non-monitored sites? We employed hierarchical clustering of the observed groundwater time series using both dynamic time warping and correlation based distance matrices. Based on the common site characteristics of the members of each cluster, the time series models were transferred to all non-monitored sites. This categorical approach provided maps of spatio-temporal groundwater dynamics across the entire catchment. We further developed a continuous approach based on process-based hydrological modeling and water table dynamic similarity. We suggest that continuous measurements at representative points and subsequent time series analysis can shed light into groundwater dynamics at the landscape scale and provide new insights into space-time patterns of hydrologic connectivity and streamflow generation.