A Spatially Explicit Model for Mapping Headwater Streams

Abstract

Headwater streams are the primary sources of water in a drainage network and serve as a critical hydrologic link between the surrounding landscape and larger, downstream surface waters. Many states, including North Carolina, regulate activity in and near headwater streams for the protection of water quality and aquatic resources. A fundamental tool for regulatory management is an accurate representation of streams on a map. Limited resources preclude field mapping every headwater stream and its origin across a large region. It is more practical to develop a model for headwater streams based on a sample of field data that can then be extrapolated to a larger area of interest. The North Carolina Division of Water Quality has developed a costeffective method for modeling and mapping the location, length, and flow classification (intermittent and perennial) of headwater streams. We used a multiple logistic regression approach that combined field data and terrain derivatives for watersheds located in the Triassic Basins ecoregion. Field data were collected using a standard methodology for identifying headwater streams and origins. Terrain derivatives were generated from digital elevation models interpolated from bare-earth Light Detection and Range data. Model accuracies greater than 80% were achieved in classifying stream presence and absence, stream length and perennial stream length, but were not as consistent in predicting intermittent stream length.(KEY TERMS: hydrology; hydrologic model; headwater streams; stream mapping; logistic regression model; LiDAR; GIS; stream regulation.) Russell, Periann P., Susan M. Gale, Breda Muñoz, John R. Dorney, and Matthew J. Rubino, 2015. A Spatially Explicit Model for Mapping Headwater Streams.

Modeling wet headwater stream networks across multiple flow conditions in the Appalachian Highlands

“…The overall agreement between the modeled probability of wet stream pixels and field maps indicates logistic regression is an effective way to characterize headwaters across flow conditions. Similar studies successfully apply logistic regression to delineate headwater channels (Russell et al ., ), and our results demonstrate that this fairly simple approach is also suitable for modeling network expansion, contraction, and disconnection at a finer reach scale by varying the probability threshold values of model outputs. Traditionally, physically mapping wet stream length multiple times is the only method for measuring wet network variability.…”

“…() found that curvature variables can help differentiate perennial and intermittent streams. Longitudinal and cross‐sectional curvature (often known as ‘profile’ and ‘planform’ curvature, respectively, in ArcGIS) commonly correlate with the location of channel heads (Tarolli and Dalla Fontana, ; Julian et al ., ) and flow origins (Whiting and Godsey, ) and are often of secondary importance in stream network models following upslope area and slope (Sun et al ., ; Elmore et al ., ; Russell et al ., ). However, TWI and TPI seem to largely capture the topographic information that curvature metrics provide while producing higher model accuracy.…”

“…Using variables like upslope area, slope, and curvature, logistic regression determines the probability of the presence of a stream channel at each catchment pixel. Despite the success of logistic regression for delineating headwaters, model accuracy remains lower for the temporary channel network (Russell et al ., ). Alternatively, because flow duration varies widely both along and among non‐perennial streams, we believe modeling efforts should shift to continuous predictions of the wet, active network through time rather than a single, stable configuration of perennial or temporary channel length.…”

Classification of Ephemeral, Intermittent, and Perennial Stream Reaches Using aTOPMODEL‐Based Approach

“…() used channel geometry (width:depth ratio and channel slope) and watershed size to distinguish among the three stream types in eastern Kentucky. In North Carolina, hydrology, channel geomorphology, and the presence or absence of biological species associated with flow permanence were used to differentiate among stream types (Russell et al ., ). Fritz et al .…”

Measuring function and structure of urban headwater streams with citizen scientists

“…Understanding individual headwater structural and functional variability could provide valuable insight into these cumulative effects, yet considering that headwater streams can represent 70–80% of the total stream length in unimpaired stream networks, monitoring their presence (i.e., in the case of ephemeral streams; Russell et al. , Williamson et al. ) and properties requires considerable effort.…”