Distributed watershed modeling

Status:

Distributed watershed modeling will be conducted using the TIN-based Real-time Integrated Basin Simulator (tRIBS; Ivanov et al. 2004; Vivoni et al. 2007). The first step in the application of tRIBS is to delineate the basin boundaries and the stream network derived from a Digital Elevation Model (DEM). A 29 meter resolution DEM obtained from the USGS National Elevation Dataset (NED; http://seamless.usgs.gov/) was processed using ArcGIS. Figure 1 shows the resulting basin boundaries for the Lower Santa Cruz and San Pedro River Basins. This figure also indicates the location of available USGS stream-gaging stations and a few geographical indicators such as towns and the US-Mexico border.

The preliminary basin boundaries and stream networks shown in this figure may be modified as other DEM products (SRTM and ASTER DEM) are also being evaluated. In addition to the figure, we also attach a ZIP file to this posting that contains the GIS shapefiles for the basin boundaries in Decimal Degrees (DD) and Universal Transverse Mercator (UTM Zone 12 North) geographic coordinate systems. These basin boundaries (from the USGS NED 29 m DEM) can be used by other participants for masking the study basins or for their own purposes. See the Download Link at the end of the post.

Figure 1. NSF-WSC Study Basins with Elevation, Boundaries and Stream Network.

Discussion:

The next step in the model setup process is to discuss the appropriateness of the basin boundaries as well as the identification of the critical reaches in the stream network. The details on the boundaries and river network are important for the groundwater modeling and hydraulic river routing portions of the integrated modeling. We would appreciate comments and suggestions from others in the group. We would also like to divide the river network into reaches for the purposes of identifying common points of interaction for the demographic scenarios and the riparian habitat efforts. Input on this in the form of existing reach delineations would be helpful. Finally, we would like to be able to overlay the basin boundaries and the gridded output mask from the second downscaled WRF model runs to see how the climate information will be mapped onto the watershed domain. This effort is currently in progress.

Future work:

1. Derive the triangulated irregular network (TIN) for each study basin with the finalized basin boundaries and channel networks.

2. Obtain and process available high-resolution soil and vegetation data layers for the entire river basins. Begin process of parameter value identification.

3. Setup numerical model in ASU Saguaro cluster for basins. Estimate model performance times for basins for specified number of processors.

4. Derive methods for transferring downscaled WRF model output to tRIBS meteorological model input.

Download Shapefiles:

http://vivoni.asu.edu/wsc/NSF_WSC_Boundary_Shapefiles.zip

References:

Ivanov, V.Y., Vivoni, E.R., Bras, R. L. and Entekhabi, D. 2004. Catchment Hydrologic Response with a Fully-distributed Triangulated Irregular Network Model. Water Resources Research. 40(11): W11102, 10.1029/2004WR003218.

Vivoni, E.R., Entekhabi, D., Bras, R.L., and Ivanov, V.Y. 2007. Controls on Runoff Generation and Scale-dependence in a Distributed Hydrologic Model. Hydrology and Earth System Sciences. 11(5): 1683-1701.

Climate simulations

The simulations with the regional climate model WRF (weather research forecast model) for historical and future climate involves a two-step approach.

First Dynamical Downscaling

The output from the Hadley centers’ HAD-CM3 global climate model for the period 1968-2080 is downscaled to a 35km resolution over North America. The future climate data is generated based on the A2 emission scenario, currently the likeliest scenario for the future. This data is available at a 6-hourly temporal resolution and provides approximately 100 variables including atmospheric and land-surface data. However, this data doesn’t have the appropriate spatial or temporal resolution to be used by the hydrologic model tRIBS. For this reason, a secondary downscaling procedure is needed.

• Status: This first downscaling has been completed for the HAD-CM3 model. The output is stored locally.
• Future work: We have the forcing data fore the second GCM (MPI-ECHAM5), and are currently performing pre-processing to get the data into WRF.

Second Dynamical Downscaling

The secondary downscaling uses the output of the first simulations (6-hry - 35km) as forcing data for WRF, and generates data at a 10km resolution every hour. The area for the secondary downscaling encompasses most of Arizona, western New Mexico and part of northern Mexico (see Figure 1). This domain was selected based on conversations between E. Vivoni, F. Dominguez and E. Rivera. The team has decided not to do continuous simulation but three “time-slice” experiments: 10 years in the past (1991-2000), 10 years in the “near” future (2031-2040) and 10 years in the “far” future (2071-2080).

Figure 1: Domain for secondary downscaling

While there are approximately 100 variables that are available from the secondary downscaling, only a subset are needed for tRIBS. E. Vivoni identified the forcing variables needed by tRIBS, and E. Rivera located them in the WRF output files. These are their corresponding names:

Table 1: tRIBS required forcing variables and their

corresponding names in the WRF output files.

• Status: We have made considerable progress in getting the output from the first dynamical downscaling to the correct format to be input into WRF. We have successfully completed one year of simulation with no problems. It is estimated that each of the 10-year simulations takes approximately 4 days, which makes it feasible to complete the three time slices in 1 month (including all the pre-processing). Each output file contains 13 days (at hourly timestep) and is currently 4GB.
• Future work: We will perform the three time-slice experiments in the coming month.