C. Duffy (PSU)

The hypothesis of this research is that historical hydroclimatic observations, signal processing tools, and multi-scale dynamical models collectively provide a systematic strategy for scientific discovery for advancing our understanding of mechanisms, rates, and time scales of recharge and discharge fluxes within the Rio Grande. The modeling approach is based on a low-dimensional representation of local hillslope hydrologic processes, which is then upscaled to more complex terrain. The modeling is driven by multi-resolution GIS coverages for model input and a basin hydrogeologic conceptual model (HCM). This blueprint for modeling embraces data assimilation, optimization, and water resources forecasting.

Our data analysis show that low-frequency oscillations in seasonal to interannual and decadal climatic forcing may interact with the long time scales of deep soil-moisture and groundwater storage to amplify low-frequency modes in runoff in ephemeral, intermittent and perennial streams of the basin. Evidence suggests that low-frequency components in mountain front runoff are consistent with the El Nino-Southern Oscillation, quasi-biennial, and quasi-decadal signature. However, the physical role that the basin hydrogeology, topography and vegetation play is unresolved. Discussions with J. Wilson regarding his basin cross-section models currently under development at NMT, and the LANL Virtual Modeling Laboratory, will provide additional means of dynamical model comparison with high-resolution results. C. Duffy has had substantial interaction with the USGS (Stan Leake) regarding their regional investigation of recharge across the southwest. In the area of information management, our group is working with the University of Arizona and LANL scientists (H. Gupta, E. Springer) in developing a strategy for a multi-scale Geographical Information System (GIS) for the Rio Grande. Preliminary coverages (DEM, stream networks, geology, etc.) are found on the http://cataractis.cee.psu.edu/riogrande/ web site.

Activities and Results

The emphasis on this project has shifted from hydroclimatic data analysis to the dynamical modeling, parameter identification, and recharge estimation. Validation of the approach is being done using data sets generated by LANL for the Parajito plateau in cooperation E. Springer and B. Newman, the Rio Puerco watershed, and ephemeral mountain-front streams of the Sangre de Christo mountains at the Great Sand Dunes National Park, as well as numerical experiments using integrated surface-subsurface models (HMS-MOD). The genetic algorithm (GA) technique has been implemented to identify the model parameters with the observed daily precipitation (rain and snow), temperature and runoff. The GA method is shown to be useful where parameter ranges of the dynamical system can be specified a-priori. The model allows us to uncover nonlinear processes, feedbacks and resonance-like effects from observed hydroclimatic, groundwater and runoff data in the basin. We have developed a visualization procedure using 3-D block diagrams for hydrogeologic-conceptual model. A new version of the Rio Grande Web Site and digital field trip is near completion. This includes a GIS layers (geology, soils, topo, basin/watershed boundaries), historical hydroclimate data.

As part of our attempt to evaluate the space-time scales of mountain-front recharge and determine the prospect for storage-type dynamical models, we have used numerical experiments with Richards' Equation to guide our understanding of the recharge process in the field. For example, results for the Upland Recharge-Runoff Regime (Figure 7) where atmospheric input to the model is: forcing = noise + periodic terms. Singular spectrum analysis is then used to evaluate the signatures of infiltration, recharge, runoff, and deep recharge (leakage) for the numerical experiment. The results show a power-law spectra over a fairly broad range of frequency. The power-law structure compares favorably with the long-term soil moisture experiments at Los Alamos National Laboratory (Nyhan and Duffy, 1998). Our conclusion is that the local recharge-runoff process is low dimensional and relatively simple but nonlinear models may indeed describe such systems. The implications of the power-law behavior in the soil moisture is the prospect for scaling relations which we hope to discover and lead to a new dynamical theory for recharge estimation.

Plans

During the next two years we intend to finish our numerical experiments on surface water/groundwater-coupled modeling and complete the intercomparisons with field data, including the effects of macropore flow. We are currently carrying out numerical experiments on recharge at an intermediate scale within alluvial fans and deep valley aquifers. Field sites have been selected at Los Alamos, the Sand Dunnes National Monument in Colorado, and the Rio Puerco below Albuquerque. The goal of the next 2 years is to find a clear linkage between recharge and long-term climate oscillations across multiple space and time scales. Extension of the low-dimensional model to large regions is the long-term goal of the research. Instead of a regular grid such as in a Finite Difference method or the Triangulated Irregular Network (TIN) as in Finite Element method, our strategy is to decompose the river basin into sub-basin using GIS tools. This not only makes our model fully GIS-driven, but also makes it highly scalable. For example, a large watershed is decomposed into twenty elements and every element can be considered as a sub-watershed with low-dimensional representation (Figure 8). The multi-scale concept means that the model is a function of the "support" used to delineate the basin from the GIS. The hydrogeologic conceptual model will be an essential element of the GIS, since this provides the hydraulic geometry and parameters for large-scale surfacewater-groundwater model development.