CTRE is an Iowa State University center, administered by the Institute for Transportation.

Address: 2711 S. Loop Drive, Suite 4700, Ames, IA 50010-8664

Phone: 515-294-8103
FAX: 515-294-0467

Website: www.ctre.iastate.edu/

Iowa State University--Becoming the Best

Stream and Watershed Delineations using LiDAR Data

Researcher(s)

Principal investigator:

Project status

In progress

Start date: 08/01/15
End date: 03/31/17

Sponsor(s)/partner(s)

Sponsor(s):

Partner(s): http://www.water.iastate.edu/waterspecialists/david-eash and https://profile.usgs.gov/daeash

About the research

Abstract:

With the availability of Light Detection and Ranging (LiDAR) data for Iowa (http://www.iowadnr.gov/Environment/GeologyMapping/MappingGIS/LiDAR.aspx) and the development of programs to enforce drainage networks on 3-meter LiDAR digital elevation models (DEMs) (Gelder, 2015), the delineation of accurate drainage networks needs to be determined for the appropriate enforcement of LiDAR DEMs and measurement of drainage-basin characteristics. The basin-characteristic measurements of stream-channel length, slope, density, and order have been identified as significant variables for the estimation of flood discharges (Eash and others, 2013; Eash, 2001), flow-duration discharges (Linhart and others, 2012), and low-flow discharges (Eash and Barnes, 2012) in Iowa. The constant of channel maintenance (CCM) basin characteristic was a significant variable in the development of flood-estimation equations for the Des Moines Lobe landform region (flood region 1; Eash and others, 2013). CCM is a measure of drainage density calculated as a ratio of drainage area divided by the total length of all mapped streams in the basin. However, the placement of channel initiation points has always been a matter of individual interpretation, leading to variances in stream definitions between analysts. Thus testing of different quantitative stream initiation methods on hydrologically enforced LiDAR DEMs, will provide different drainage-network delineations from which basin-characteristic measurements can be evaluated for the optimization of stream-channel delineations from LiDAR data. Side-by-side testing of basin-characteristic values measured for the total drainage area versus the "effective" drainage area of basins is needed to determine which watershed delineation provides the best predictive accuracy for flood estimation. The effective drainage area represents a subset of the total watershed area and is the area that actually contributes streamflow under "reasonable" flow conditions for a given storm event, such as a 20- or 2-percent annual exceedance-probability (AEP) 24-hour rainfall. Because the predictive accuracy of flood-estimation equations for watersheds located within the Des Moines Lobe landform region (Eash and others, 2013; Eash, 2001) is the poorest in the State, research is needed to improve the accuracy of stream-channel delineations and flood estimation within the Des Moines Lobe landform region. The proposed study will test at least four different quantitative methods to define stream initiation using 3-meter LiDAR data for 17 streamgages with drainage areas less than 50 square miles that are located within the Des Moines Lobe landform region in north-central Iowa. Table 1 lists the 17 streamgages that have been selected for inclusion in this study and figure 1 shows their location. All of these streamgages were included in the 2013 StreamStats flood-estimation study for Iowa in which 59 selected basin characteristics were measured for each streamgage using 1:24,000-scale data from stream networks, basin boundaries, and 10-meter DEMs (Eash and others, 2013). Watersheds for the 17 streamgages will be enforced using the method developed by Gelder (2015) for the Iowa Highway Research Board (IHRB) and at least four stream initiation methods will be used to define channel initiation points and the downstream flowpaths. Possible stream initiation methods include (1) channelization (determined by local elevation difference), (2) channelization (determined by profile curvature), (3) aspect change, (4) grid order (similar to Strahler stream order, but starting at flow accumulation = 0), (5) analyst derived streamlines using the LiDAR as a reference, and (6) snapping 1:24,000-scale stream initiation points to high-flow accumulation cells. The stream initiation methods will then be used to define channelized flowpaths on the hydrologically enforced LiDAR DEMs, creating multiple sets of selected basin-characteristic values that will be measured for each streamgage. The 4-6 different quantitative methods to define stream initiation will be tested side-by-side for three watershed delineations (1) the total drainage-area delineation (which should be similar to a 0.2-percent AEP 24-hour rainfall), (2) an effective drainage-area delineation of basins based on a 2-percent AEP 24-hour rainfall, and (3) an effective drainage-area delineation based on a 20-percent AEP 24-hour rainfall producing 12-18 different data sets of basin-characteristic values for each streamgage watershed. Basin-characteristic values for stream density (STRDEN), relative stream density (RSD), total stream length (STRMTOT), constant of channel maintenance (CCM), the number of first-order streams (FOSTREAM), and drainage frequency (DRNFREQ) will be measured for each streamgage watershed from at least four stream initiation methods and LiDAR DEMs. The 4-6 sets of LiDAR-measured basin-characteristic values for total drainage area will be evaluated and compared to 1:24,000-scale StreamStats-measured basin-characteristic values for total drainage area for determining optimum stream-channel delineations from LiDAR data. Because a comparison of LiDAR-measured basin-characteristic values and 1:24,000-scale StreamStats-measured basin-characteristic values may not adequately determine optimum stream-channel delineations from LiDAR data, additional selected basin characteristics will be measured for each streamgage to also test optimum stream-channel delineations from LiDAR data using flood-estimation regression analyses. Expected moments algorithm/multiple Grubbs-Beck test (EMA/MGB), AEP streamgage analyses (Eash and others, 2013) will be updated through the 2014 water year for the 17 streamgages and regression analyses will be performed to identify which of the 12-18 sets of LiDAR-measured basin-characteristic values from the 4-6 stream initiation methods and the three watershed delineation methods are the most significant for the estimation of 50-, 20-, 10-, 4-, 2-, 1-, 0.5-, and 0.2-percent AEPs for the Des Moines Lobe landform region for drainage areas less than 50 square miles.