Hydrography - the building block of IP - Session Guide Lead(s): Mindi Sheer (NOAA), Dan Miller (ESI), Bill Kaiser (USFS), Group Goal: discuss technical hydro considerations independent of source, review source hydro data and collect feedback on concerns specific to IP analyses, briefly brainstorm coordinated hydro efforts and requesting IP variables from hydro source agencies to improve application of IP 1. NHD and DEM-derived hydro 1. What experiences / concerns are there with using NHD linework (24k and 100k)? NHD+ Attributes? 2. General questions on NHD? 3. Perspectives on NHD+ methods of calculating mean annual discharge (Vogel, Unit Area Runoff method) 4. What methods have you used for linking hydro of different scales or sources together for IP or habitat potential work (to get attributes)? 5. Discuss concerns for DEM-derived streams that affect IP analyses (missed channels, incorrect channel locations, especially in flat areas, unresolved topographic features) 6. Is it feasible to generate your own streams to calculate IP? Is there interest in using existing NHD* (*see note in strawman from P. Eldred), state data sets, or NetMap sources? 7. LiDAR-derived streams. Dan Miller may be able to answer some questions on using LiDAR-derived DEMs/streams for IP analysis. Please refer to this section of the strawman. 2. Discrepancies in hydro that affect IP analyses 1. Discuss methods and sources for discerning dry and ephemeral to improve accuracy and consistency of thresholds cutoffs in IP analyses streams (and consequent removal?) 1
2. What is being done about regional differences in stream density in NHD stream (<100k)? Is there a minimum and maximum scale? 3. Developing coordinated hydro data sets 1. What are good methods for linking DEM-generated streams to other stream layers 2. Most fish data and supportive aquatic data sets are created and aligned with state stream sources (Davies et al 2007; 100k SalmonScape (WA); NRIMP / Streamnet (OR);???? (CA)). What are the best solutions for dealing with this when generating DEM-derived streams examples? 3. Should there be a regional DEM-derived stream layer appropriate for IP analyses create? Is this feasible? 4. Reach-level Hydrographic Details 1. Discuss methods of determining reach breaks 2. DEM-derived streams calculate gradient on the fly, using various methods. Reach length variations can result in different gradient estimates. This, in turn, indicate variable species cutoffs. Experience with this? 3. Suggestions for dealing with double banked streams (lakes, ponds, reservoirs) in IP analyses? Should there be a way to include these areas more effectively? 2
Hydrography - the building block of IP Lead(s): Mindi Sheer (NOAA), Dan Miller (ESI), Bill Kaiser (USFS), Group Goal: discuss technical hydro considerations independent of source, review source hydro data and collect feedback on concerns specific to IP analyses, briefly brainstorm coordinated hydro efforts and requesting IP variables from hydro source agencies to improve application of IP. Source of some NHD info is from NHD training session (Sept 2008) 1. NHD and DEM-derived hydro a. NHD and NHD+ stream networks o Attributes Perennial/ephemeral codes (100k?). Based on the original source maps from the 70s. DNR,BLM,and USFS data may have better value-added info on this, locally..this information may have changed Were state data on ephemeral / perennial added to this for WA/OR/CA/ID? NHD+ provides a flow attributes table that includes variables derived from 10m(?) DEMs These link to NHD streams (NHD 100k only???) NHD+ also provides flow accumulation grids and flow direction grids for users. NHD+ stream attributes are problematic in very flat areas Drainage Enforced Dems used to create the NHD+ variables NHD+ variables that may be pertinent to this workshop mean annual flow estimates (cfs) velocity (fps).?? o Availability of 24k (nationwide, except AK) and 100k (nationwide) NHD data in WA, OR, and CA o Reach length 100k - NHD 100k reaches are 1km in length (broken at 1km lengths), and catchments avg 1km2 in size. o Reach length 24k -?????? o Benefits for IP analyses D D D o Caveats for use in IP analyses D D 3
b. DEM-derived stream networks HYDROGRAPHY Session Guide o Attributes available (using certain methods/tools) (Table 1) o Availability 10m DEMs widely available 10m Drainage Enforced (DE) available where? Note that drainage enforced DEMs have lowered elevations for cells traversed by mapped channels. Alteration of DEM elevation values affects subsequent calculation of attributes such as gradient and valley width. Drainage enforcement during flow routing, without any alteration of the DEM, can also be done (e.g., Netstream). Need to use these DEMs to generate your own stream network ArcHydro NetStream (Miller 2003) Arc GRID TauDEM (http://hydrology.neng.usu.edu/taudem/) NetMap and various projects have generated stream networks for different purposes all over the region these datasets may be available (see Map Appendix B) o Benefits for IP analyses A DEM-derived stream used by all provides meaningful comparisons of watershed attributes between watersheds, e.g., road-stream crossings. Stream layer at same resolution (e.g., 1:24,000) as topographic data. Channel density will be consistent across administrative units (with some exceptions crossing 24k quad boundaries, i.e., crenulation changes) Methods of calculating attributes the same across admin boundaries DEM-derived streams simplify delineating catchment basins catchments sometimes are needed for determining relationship of upslope and in-channel attributes. Ancillary data can be linked to these streams, and routed to the streams where necessary. This is especially useful for IP analyses where researchers want to include extrinsic factors (land use, riparian vegetation, runoff, roads (vector sources), erosional indices etc.) Streams are routed. o Caveats for use in IP analyses Inaccuracies in elevation values result in: Missed stream channels Incorrect channel locations Unresolved topographic features, e.g., channel-confining terraces Horizontal resolution may be insufficient to resolve pertinent topographic features (e.g., incised channels) DEM-traced channels may not line up with other georeferenced channel data, e.g., with stream layers derived with other techniques c. LIDAR-derived stream networks and IP Considerations 4
LiDAR can potentially provide elevation data of sufficient resolution and accuracy to directly resolve most or all stream channels (Mouton 2005) and topographic attributes pertinent for calculations of IP. The data resolution is often sufficient to resolve road prisms, but not culverts and cross drains. Hence, flow routing from LiDAR-derived DEMs can result in rerouting of flow paths by roads. Specialized algorithms Availability of LiDAR data is limited, and even where available, the data may not span entire watersheds. Algorithms are needed for combining DEMs of different resolution and accuracy to provide complete watershed coverage while maintaining data accuracy and resolution. To use these high-resolution data sets over contiguous watersheds requires development of algorithms to deal with large data sets that can provide reasonable processing time. For example, channel delineation may be done with data subsampled to a lower resolution (from 1 m to 5 m, say), and then attributes for each reach (e.g., valley width) can be calculated using the original high-resolution DEM. To by-pass road prisms. Data smoothing works, but reduces resolution. New algorithms may be developed to deal with flow routing by roads, e.g., use of smoothed data to identify channel flow paths, then forcing flow directions along these paths in higher-resolution analysis. Require algorithms to combine multiply-resolved flow paths within a single channel. The horizontal point spacing of many LiDAR-derived DEMs is less than the width of many channels; hence, multiple parallel flow paths may be traced within a single channel. These need to be combined to obtain appropriate estimates of contributing area to the channel. It may be possible to measure channel width and terrace extent directly from LIDAR DEMs, rather than relying on regional regressions to drainage area (and precipitation, etc.) for estimates of channel width. It may also be possible, in some cases, to obtain measures of floodplain attributes, e.g., number and area of side channels. File size. High point density (e.g., 1-m horizontal grid spacing) results in large data files, e.g., 5 th -field HUCs require Gbytes. Availability of LiDAR data is limited and data quality is variable. LiDAR coverage is rapidly expanding, but it will be many years before complete regional coverage is available. 2. Discrepancies in hydro that affect IP analyses d. Stream density and percent of dry and ephemeral streams included in the IP base hydrography is dependent on channel initiation parameters o drainage area used 100k NHD = 1km 2, is it the same for NHD 24k?) o DEM-derived networks can set criterion for channel initiation when delineating (NetStream, TauDEM). Channel-initiation criteria can be based on landscape attributes, e.g., the typical drainage area and slope combination at which the transition from divergent to 5
convergent topography occurs (McNamara et al. 2006). Using such topographically based criteria, the thresholds (in drainage area, slope, and topographic convergence) used to identify channel initiation points can be calibrated to individual DEMs. Channel-initiation criteria calibrated in this fashion vary regionally and vary between low- and high-gradient terrain. Because this method relies solely on DEM-represented topography, the flow regime of delineated channels is unknown: many of the channels may be ephemeral or even represent past flow regimes and never carry surface flow in the current climate. e. Stream channel density dependent on channel initiation (CI) and characteristics of elevation data (crenulations? ) o CLAMS project had large variation in channel density across adjacent 7.5-minute quadrangles (with the same initiation criterion), reflecting variation in the degree to which contour lines were crenulated. o CLAMS attempted to use BLM surveyed channel initiation points to calibrate an initiation criterion, but these data provided no consistent relationship with slope and drainage area (see also Jaeger et al. 2007). o One strategy (Clarke et al. 2008) is to generate the densest coverage that can be resolved from DEMs and use other data sources to delineate dry from intermittent from perennial channels (and remove those if desired). f. NHD regional differences in stream density*(see note from P. Eldred at end of doc) i. What is being done about that? ii. Is there a max and minimum scale? 3. Developing coordinated hydro data sets g. Which methods are used for DEM-derived streams in the PNW and by whom? o Algorithms Netstream and Netmap: DEM-preprocessing: removal of closed depressions by cutting (Rieger 1998) and filling (Jenson and Domingue 1988). Flow directions (described in Clarke et al. 2008) Dinfinity, which allows dispersion (Tarboton 1997) to channel initiation, then D8 (O'Callaghan and Mark 1984) to ensure single-channel streams. Channels directed to downslope DEM cell with greatest topographic convergence, rather than steepest descent, to better follows contour crenulations. Channel initiation based on slope-area product, with separate criteria for low- and high-gradient areas, with topographic convergence threshold. Drainage enforcement utilizes entrenchment, but affects only flow directions. a. What are good methods for linking DEM-generated streams to other stream layers Spatial joining to NHD+ (flow attributes) 6
o Use NHD or 100k source to burn into the DEM (or use the DE-DEMs) for easier spatial joining Spatial joining to other sources? NHD tools to help with this link? Do tools work with DEM-derived streams (ie, no LLID) Spatial joins between stream layers can result in many miscorrelated channel segments (reaches). Better algorithms employ a combination of routing, topology, proximity, and stream attributes (e.g., drainage area, if available) to match stream segments between data sets. A tool to do this is being added to NetMap, but is not yet finished. h. Should there be a regional DEM-derived stream layer appropriate for IP analyses Is this necessary or even feasible? If so, what attributes and features should it include? benefits, caveats, use, extent, applicability 4. Reach-level Hydrographic Details b. Scale and sinuosity Traced channel length varies with horizontal resolution of the elevation data (Clarke and Burnett 2003) c. Reach length and scale The IP score is calculated by reach, so the length of the reach and method used for determining reach breaks is important Gradient uses reach length for calculation. Variable reach length can result in variability in channel gradient estimates. Reach length is designated during stream delineation or through routing and dynamic segmentation. Methods used are either a.) setting parameters prior to generating a stream network (ie reach segments created on the fly), or b.) obtaining a routed stream layer and designating reach segments using dynamic segmentation (ESRI ref). If you are interested in comparing scores in another region from another study, consideration should be given to the reach-break strategy employed. Different strategies of delineating reach breaks changes the linear scale of the information, so requires generalization if comparing results. The equal reach length method is the easiest for summarizing results, as it allows a relatively easy conflation to different lengths if necessary. Common methods of delineating hydrographic stream reaches: o at tributary junctions o at natural geomorphic breaks or transitions in valley width or channel gradient (Miller 2003). This method is meant to simulate natural reach breaks that might be visually identified by field survey. o reaches of equal length (e.g., 100 or 200m segments) for simplicity of analysis. Equal-length reaches can be easily summarized up or grouped. d. double-banked streams, polygon data 7
Are there ways that double-banked streams have been incorporated or could be incorporated more effectively into IP analyses? Typically, IP models for anadromous species do not incorporate a special classification for lakes, reservoirs, and ponds (GIS terminology: double-banked streams). In run of the river lakes or reservoirs, stream reaches that go through the midline of a lake or reservoir are considered the same as other stream reaches, with the exception that they have a large BFW. Curves for fish typically incorporate BFW or wetted width. Perhaps reservoirs or lakes should be considered separately for some species? Typically, IP curves for BFW are developed for lotic systems and are linked to river size. Curves have not been developed to effectively represent habitat potential in lentic parts of the system. These systems are more effectively represented by lateral rather than linear representation. Netstream and NetMap include an optional field in the channel coverage to flag channel reaches traversing lakes or reservoirs (which must be identified from another data source, e.g., a polygon file of water bodies). These reaches are then excluded from summary statistics of channel attributes or treated as sinks (e.g., for sediment routing). Table 1. Examples of some hydrogeomorphic and climatic variables related to habitat quality from hydrologic stream network and digital elevation models (DEM). Variable Source Channel gradient 1,2 From DEM 3,4 Mean annual flow 1,2 Regression of gage data to drainage area (DEM) and mean annual precipitation (PRISM) 3 Channel constraint 1,2 Valley-width index (ratio of valley to channel width, with channel width based on regional regression to mean annual flow) correlated with field inventoried constraint categories. Valley width estimated from DEM 3,6 Mean Summer (August) Low PRISM Temperature 1 Valley-width transitions (e.g., From DEM 5 from confined to unconfined channels) 5 Tributary confluences 5 From DEM 5 1 Agrawal et al. (2005) 2 Burnett et al. (2003, 2007) 3 Clarke et al. (2008) 4 Davies et al. (2007) 5 Benda et al. (2004, 2007) 6 Hall et al. (2007) *Notes on Hydrography: (from Peter Eldred Aquatic & Riparian Effectiveness Monitoring Program (AREMP)) The 100k National Hydrography dataset (NHD) does not have enough map accuracy and detail for our stream analysis purposes. The current 24K Pacific Northwest Hydrography Framework layer is being migrated into the 24K NHD data layer. I believe this migration process to be mostly complete for Oregon and Washington. This stream layer is a collection of the best available data from multiple agencies pieced 8
together. It has been mapped by a wide variety of methods with very variable densification. Mapping tends to be much more detailed on public lands, with a sparse network of steams on private lands. There is no current effort that I m aware of to make the stream mapping more consistent. The only potential way I know of to make the mapping somewhat more consistent is to select out and use only the perennial streams, but this attribute is not currently fully or consistently attributed. Even if the perennial streams were completely attributed, there would still be problems with differences in stream density due to different mapping intensities, and the problem of lack of stream mapping on private land would persist. References Agrawal, A., R. S. Schick, E. P. Bjorkstedt, R. G. Szerlong, M. N. Goslin, B. C. Spence, T. H. Williams, and K. M. Burnett. 2005. Predicting the potential for historical coho, chinook and steelhead habitat in northern California. National Oceanic and Atmospheric Administration. Benda, L., D. J. Miller, K. Andras, P. Bigelow, G. H. Reeves, and D. Michael. 2007. NetMap: A new tool in support of watershed science and resource management. Forest Science 53:206-219. Benda, L. E., K. Andras, D. J. Miller, and P. Bigelow. 2004. Confluence effects in rivers: Interactions of basin scale, network geometry, and disturbance regimes. Water Resources Research 40:W05402. Burnett, K. M., G. H. Reeves, D. J. Miller, S. Clarke, K. Christiansen, and K. Vance-Borland. 2003. A first step toward broad-scale identification of freshwater protected areas for Pacific Salmon and Trout in Oregon, USA. Pages 144-154 in A. Grant and D. C. Smith, editors. Aquatic Protected Areas: what works best and how do we know? Proceedings of the World Congress on Aquatic Protected Areas, Cairns, Australia, August 2002. Australian Society for Fish Biology, North Beach, WA, Australia. Burnett, K. M., G. H. Reeves, D. J. Miller, S. Clarke, K. Vance-Borland, and K. Christiansen. 2007. Distribution of salmon-habitat potential relative to landscape characteristics and implications for conservation. Ecological Applications 17:66-80. Clarke, S. and K. Burnett. 2003. Comparison of digital elevation models for aquatic data development. Photogrammetric Engineering and Remote Sensing 69:1367-1375. Clarke, S. E., K. M. Burnett, and D. J. Miller. 2008. Modeling streams and hydrogeomorphic attributes in Oregon from digital and field data. Journal of the American Water Resources Association 44:20. Davies, J. R., K. M. Lagueux, B. Sanderson, and T. J. Beechie. 2007. Modeling stream channel characteristics from drainage-enforced DEMs in Puget Sound, Washington, USA. Journal of the American Water Resources Association 43:414-426. Hall, J. E., D. M. Holzer, and T. J. Beechie. 2007. Predicting river floodplain and lateral channel migration for salmon habitat conservation. Journal of the American Water Resources Association 43:786-797. Jaeger, K. L., D. R. Montgomery, and S. M. Bolton. 2007. Channel and perennial flow initiation in headwater streams: management implications of variability in source-area size. Environmental Management 40:775-786. Jenson, S. K. and J. O. Domingue. 1988. Extracting topographic structure from digital elevation data for geographic information system analysis. Photogrammetric Engineering and Remote Sensing 54:1593-1600. McNamara, J. P., A. D. Ziegler, S. H. Wood, and J. B. Vogler. 2006. Channel head locations with respect to geomorphologic thresholds derived from a digital elevation model: a case study in northern Thailand. Forest Ecology and Management 224:147-156. Miller, D. J. 2003. Programs for DEM Analysis. Landscape Dynamics and Forest Management, General Technical Report RMRS-GTR-101CD. USDA Forest Service, Rocky Mountain Research Station, Fort Collins, CO, USA. Mouton, A. 2005. Generating Stream Maps Using LiDAR Derived Digital Elevation Models and 10-m USGS DEM. University of Washington, Seattle, WA. 9
O'Callaghan, J. F. and D. M. Mark. 1984. The extraction of drainage networks from digital elevation data. Computer Vision, Graphics, and Image Processing 28. Rieger, W. 1998. A phenomenon-based approach to upslope contributing area and depressions in DEMs. Hydrological Processes 12:85-72. Tarboton, D. G. 1997. A new method for the determination of flow directions and upslope areas in grid digital elevation models. Water Resources Research 33:309-319. 10