1. Hydrologic Study of the Sandies & Elm Watershed
Jessica L. Watts
Good Afternoon
My name is Jessica Watts
The subject for my semester project, which is a portion of my research, started out as a Hydrologic Study of the Sandies and Elm Watershed, but rather ended up being an exercise in obtaining and working with Environmental and Weather related data.

2. Research Project Background
TCEQ TMDL Study
Bacteria
Low Dissolved Oxygen
Sandies River
Project Background
This is Texas
(Click)
The Sandies and Elm Watershed is located in the south central portion of Texas.
It is the focus of a TCEQ TMDL Study.
The major concerns with the watershed are:
????
And low dissolved oxygen
As a first step in the TMDL process a model needs to be created that correctly resembles what is happening in the watershed hydrologically, so that as the process continues the non-point source pollution can be properly modeled and regulated.
Elm River

3. Semester Project Overview
Objective:
To create an HSPF data model that simulates the gauging station flow for the years 2000 through 2004.
Tasks:
Arc Hydro Watershed Delineation
Arc Hydro To HSPF
HSPF Forcing Data
Interpolation Comparison
HSPF Output Calibration
The objective of my semester project is just that. To create an HSPF model that simulates the USGS gauging station flow for the years 2000 to As I said before this turned out to be an exercise in data collection.
The tasks for this project include:
To delineate the watershed using Arc Hydro
To translate the watershed from Arc Hydro into HSPF
To obtain forcing data
And finally to get an HSPF model and begin the calibration process

4. Arc Hydro Watershed Delineation
First Task

5. Arc Hydro Input Requirements
Digital Elevation Model
Seamless ( )
United States
1/3” National Elevation Dataset
NHD Flowlines
USGS – National Hydrography Dataset
( )
High Resolution
Arc Hydro requires these two data inputs, which I gathered from the above listed sources, Seamless and USGS-NHD

6. Arc Hydro Subbasin Delineation
Terrain Preprocessing
Watershed Processing
35 Subbasins
(limited by HSPF)
I applied the terrain preprocessing and watershed processing tools to create 35 subbasins.
The number of subbasins is limited by HSPF by the number of distinct land use types. I have 11 distinct land uses, therefore I can have up to ???? Total subbasins.
The oddly smaller subbasins were created so that there would be an outlet point, or point of calibration at the various monitoring points around the watershed. The monitoring point at the bottom right is the location of the USGS gaging station that I will be calibrating my model to.

7. Arc Hydro to HSPF
After creating the subbasins the next step was to get the Arc Hydro information into HSPF.

8. Created by Nate Johnson
Arc Hydro to HSPF
Tools and Models have been built to convert information from Arc Hydro into the HSPF format.
To do this I had the use of these wonderful tools, created by our own Nate Johnson. These tools convert information from Arc Hydro into an HSPF uci (input) file.
Arc Hydro To HSPF
Models and Tools
Created by Nate Johnson

9. Subbasin Delineation From Arc Hydro Subwatershed
The first input required for the ArcHydro2HSPF transition is the Subbasin delineation.
This dataset was required to have a unique HydroID for each subbasin area.
With unique HydroID for each area

10. Drainage Lines From Arc Hydro Drainage Lines
The second input requirement is a drainage line dataset where the DrainID of the Drainageline is the same as the HydroID of the Subbasin it is within.
With DrainID of DrainageLine = HydroID of Subbasin

11. Land Use Shapefile 1992 Land Use Downloaded from Seamless
Used ArcGIS Tools
To Convert from
Raster to a Polygon
To Clip Polygon into shape of Watershed
The third input requirement is a landuse shapefile. This can either be the land use land cover data from the ???? Or, what I used, 1992 land use from Seamless. This dataset needed to be in a polygon shapefile. So I needed to convert it from a raster to a polygon then clip the polygon into the shape of the watershed to remove the extraneous information.

12. Land Use Tables Defines how each type of land use is treated in HSPF
The final input requirement for translating the Arc Hydro information into HSPF is the Landuse definition and classification tables, which are used to define how the landuse areas in the land use shapefile are defined in HSPF.

13. HSPF
So after these inputs are put through a number of Nate’s ArcHydro2HSPF tools one can get out a UCI file that will open in HSPF as shown here with reaches connected to certain watershed areas that are broken down into specific land use areas.

14. Forcing Data
The next step is to get the forcing data that drives the HSPF model.

15. Precipitation Data NCDC Daily Precipitation Data
Years 2000 through 2004
The first of the two required data inputs for HSPF is precipitation. There is only one NCDC station that lies completely within my watershed area out of the 15 in the 5 county area.

16. Evaporation Data NCDC Daily Evaporation Data Years 2000 through 2004
Canyon Dam
SeaWorld SAT
The same situation is true for the NCDC evaporation stations available in my area.
Therefore interpolation of the available data over my watershed area is necessary.
Coleto Creek Res.
Choke Canyon DAM
Beeville 5 NE

17. Interpolation Comparison
To try and figure out what interpolation method would best serve to imitate precipitation. I checked the different interpolation methods against one day of NEXRAD data. The NEXRAD image is of August 27, 2001, when there was a significant storm over my basin. The first thing that is noteworthy is that there are no gauging stations at the center of the maximum of the storm. Therefore ANY interpolation method is bound to get it wrong.
But, we will give it a try.
So, the question ends up being, what method should we use. Kriging, IDW, or Spline?
NEXRAD, August 27, 2001

18. Data Interpolation Choices
NEXRAD
Kriging
First, Kriging.
The NEXRAD information and Kriging information are expressed over the same color value range. Therefore it is obvious that the Kriging method is not representative of the precipitation on this day at all.
The Kriging method seemed to take the data points average them and then put that value over most of the area and apply a variation directly at the points.
This is NOT the way precipitation acts.

19. Data Interpolation Choices
NEXRAD
IDW
Second, Inverse Distance Weighting
This is the one that I thought had the best chance. It put a larger weight on the points with a gradient between it and its neighbors.
This one would probably have worked well if the gages had been closer together or more representative of the storm it was checked against.
When averaged over catchments there is a 61% average difference.

20. Data Interpolation Choices
NEXRAD
Spline - Regular
Finally, Spline – Regular
This is probably the least representative of what precipitation does overall, but somewhat accurate over the watershed.
This subject is worthy of a LOT more study, but alas it was only a side-subject in my overall project.
Not that it is the best, but Spline-Regular was the interpolation choice I made.
Negative Rain !!
When averaged over catchments there is a 35% average difference.

21. HSPF Forcing Data Arc GIS HSPF NCDC Monthly Data Zonal Statistics
WDM File
NCDC
Monthly Data
Arc GIS
Zonal Statistics
So, here is an overview of the steps required to take data from NCDC though to HSPF WDM file.
Starting with obtaining NCDC Monthly data
Utilizing an Excel spreadsheet to transform the NCDC output into timeseries
Then using Jon Goodall’s Space-Time Tools to change the timeseries into RasterSeries with the Spline Interpolation method
Again using the Space-Time tool which uses Arc GIS’s Zonal Statistics to change the RasterSeries into an Attribute Series
Finally, using a tool created by Nate Johnson that translates a timeseries into the forcing data file for HSPF

22. HSPF Output & Calibration
So, finally this forcing data can be put into the HSPF model and an outflow model can be computed.

23. HSPF Output
First Try.
The model output, shown in green, does not represent what the gage is showing. In some areas the model peak leads the observed one and in others the model peak trails it. Overall there is too much water in the river. There are some calibration techniques that I have begun going through. So, just as an example I changed some of the suggested parameters.

24. HSPF Output Future Work Calibration Second try.
The first year is starting to look better, but the rest is still wrong.
Therefore my future work on this project will be
(Click)
Calibration

25. Questions ? Thanks… Nate Johnson Jon Goodall Gil Strassberg
Carrie Gibson
And
Dr. David Maidment
I would like to say a great big thank you to
Nate Johnson, Jon Goodall and all the rest at CRWR for all their help with this project.
Finally, any questions.