Untitled DocumentFine resolution coupled land-atmosphere model of the Rio Grande basin
Macro Theme Area:
Integrated Modeling [Project ID: M03]
PI:
Everett Springer
CO-PI(s):
N/A
Basin focus:
Rio Grande
Specific area in
basin /
field sites:
Rio Ojo Caliente
Summary/Goals: This project provides a scientific computational foundation for water resource decisions by coupling detailed physical models of the atmosphere, land surface hydrology (including plant communities) and groundwater hydrology of the Rio Grande Basin.
Activities and outcomes during past year:
The newest sequential version of Triangular Irregular Network-(TIN) based Real-Time Integrated Basin Simulator (tRIBS) from New Mexico Tech University (NMT) was integrated with the Los Alamos National Laboratory (LANL) parallel implementation. The major enhancement to tRIBS was the addition of a snow accumulation and melt routine that created new classes and modified others classes. A critical objective of this project is to assess the importance of coupled surface and subsurface flow in semi-arid regions. To this end we leverage the strengths of two state-of-the-art simulation codes that have previously focused on either surface or subsurface flow. Specifically, state-of-the-art surface flow codes, such as tRIBS, are designed to handle large river basins, with significant topographic features while interfacing well with geographic information system (GIS) data. However, the subsurface flow model in tRIBS is highly simplified. In contrast, state-of-the-art subsurface flow codes, such as LANL Finite Element Heat and Mass (FEHM), are well suited for complex subsurface geometries and both saturated and unsaturated flow scenarios, but have either simplified or nonexistent surface flow. After assessing the tRIBS and FEHM codes and relevant algorithms we determined that the best approach to coupling the codes was at the physical surface of the ground. Previous grid incompatibilities were corrected by adding a surface mesh capability in FEHM with simplified hydrologic routing. The FEHM code uses control volume finite element (CVFE) discretizations. This allows the grid used in tRIBS to be represented exactly in FEHM and greatly enhances the mass conservation of the overall system. We continued to collect data for model forcing, model parameters, and model testing for the Rio Ojo Caliente Basin in New Mexico. The University of California at Irvine has generated a set of weather data for the southwestern U. S. on a 4 x 4 km grid, and we have extracted the data for the Rio Ojo Caliente. These data will be the model forcing data for our simulations. We are also looking for data to test our simulation results. The Global Land Data Assimilation System (GLDAS) provides distributed data from different land surface models at various resolutions. The highest resolution available is 1/8° for the MOSAIC land surface model. Although these grid cells are large compared to the resolution of our models, they do provide us with some distributed data to check our simulations.
Plans for the upcoming year:
The deliverable for this project is the coupled model for the Rio Ojo Caliente Basin. The following tasks are required to achieve this goal. Task 1 is the parallelization of the component codes to attain the needed computational efficiency. The completion of the benchmark simulations for the parallel version of tRIBS and a series of simulations demonstrating the effectiveness of the parallel implementation on a river basin will complete this parallelization task for this project. The test will involve developing a finer grid representation and simulating the basin response for a given period. We will examine changes in simulation time to determine the effectiveness of the parallel version.
Task 2 is the coupling of parallel versions tRIBS and FEHM. The study of surface/subsurface coupling at the LANL Ponderosa site drives our development of the two state-of-the-art tools, tRIBS and FEHM, towards a common interface at the earth's surface. In FEHM recent developments facilitate the use of a common surface triangulation; in addition, FEHM has a simple surface hydrologic routing capability. Thus, each code can simulate the surface/subsurface flow at the Ponderosa site. However, it is anticipated that the different inputs and resulting flow regimes will play to the strengths of one code versus the other. Thus, we plan to conduct calibration and uncertainty studies of the Ponderosa site for both tRIBS and FEHM independently, and then in a coupled state. During this work we will develop a common driver for both tRIBS and FEHM that will facilitate sequentially coupling these codes. Data transferred will consist of combinations of head data, saturation (moisture content), and water flux. We anticipate the strength of the surface/subsurface coupling to be different in the wet and dry years, and plan to explore automatic adjustment of the time step size and the frequency of data passing Task 3 is the development of the geographical data, model parameter fields, and input forcing fields for the Rio Grande using available geographical analysis tools such as the ArcGIS™. Task 4 is estimating parameter values for the Rio Ojo Caliente basin using hard and soft data in approaches such as machine learning or neural networks. Efforts will continue to use remotely sensed data to provided vegetation terms for the surface hydrology. Task 5 is model testing to demonstrate that the model responds correctly or at least within identified bounds. The simulation of the period from 1996 - 2006 for the Rio Ojo Caliente Basin will be completed during the next year. The forcing data will be formatted for tRIBS and additional model parameters will be developed. Data for comparison of the model results will include snow cover, snow water equivalent, soil moisture, and streamflow. Soil parameters will be estimated using the neural network algorithms developed by Texas A&M.
