Comparison of digital terrain and field-based channel derivation methods in a subalpine catchment, Front Range, Colorado

2012 Summer.Includes bibliographical references.Understanding the reliability of digitally derived channel networks for mountainous headwater catchments is important to many water resource and land-use management applications. Digital elevation models (DEMs) have become an essential tool for an increasing array of mountain runoff analyses. The purpose of this study is to investigate the influence of digitally-derived topographic variables on channel network formation for a high-elevation glaciated watershed. To accomplish this, our objectives were to (1) test how differences in gridded DEM resolution affect spatially distributed topographic parameters of local slope (tan β), specific contributing area (αs), and topographic wetness index (TWI) derived from both eight and infinite directional flow algorithms, (2) map the actual stream channel network at Loch Vale and examine the influence of surface variables on channel initiation, and (3) evaluate the performance of common methods for deriving channel networks from gridded topographic data by comparing to the observed network. We found that coarser DEM resolution leads to a loss of detail in spatial patterns of topographic parameters and an increase in the calculated mean values of ln(αs) and TWI. Grid cell sizes above 1m result in a substantial shift in the overall cumulative frequency distributions of ln(αs) and TWI towards higher values. A field survey at Loch Vale revealed a complex and disjointed channel network, with 242 channelized points and 30 channel heads. We found no predictable relationships between channel head locations and geomorphic process domains. Analysis of variance (ANOVA) showed no statistically significant difference in mean ln(αs) and TWI for channel head locations grouped by elevation, aspect, slope, formation process or upslope land cover type. For most DEM resolutions and flow partitioning algorithms, deriving channel networks with spatially constant flow accumulation and TWI thresholds provides poor network representation. The publicly available National Hydrography Dataset (NHD) layer oversimplifies the channel network by neglecting almost all first and second order channels. Many of the DEM-derived channel networks that use spatially constant flow accumulation and TWI thresholds also do not reproduce the locations of low order channels in the observed channel network well. Assumptions of topographic control on channel initiation are not shown to be valid at Loch Vale, likely due to their inability to capture subsurface processes and geologic features important to channel formation. However, if using these topographically dependent threshold methods to delineate channel networks, we suggest the use of field-based survey data to identify appropriate thresholds. With appropriate thresholds, both 1m and 10m DEMs can produce channel networks with similar drainage densities to the observed network, even if locations of low order channels are not predicted accurately. Performance degrades for 30m DEMs, so we suggest that DEMs with resolutions coarser than 10m should be avoided for channel network delineation