The objective of this study is to evaluate the potential for streamflow estimation for major rivers using satellite remote sensing information.  The innovative aspect of this study is the coupling of water level information from satellite radar altimetry with an unsteady flow hydraulic model. The motivation for this approach is to exploit the dynamic information on the passage of flood waves through the river network provided by satellite observations at multiple sites.   This research compliments ongoing research efforts to develop at-site approaches, by exploring a system that fully integrates remote sensing estimates of water levels, available in-situ observations, and unsteady flow modeling, with a data assimilation scheme.

Ground tracks for the TOPEX/POSEIDEN satellite over North America and a schematic showing an application of the discharge estimation procedure.

Satellite altimetry estimates of water levels are available at discrete sampling locations (with orbital repeat periods of 10 days or longer) as the satellite passes over the river.   In-situ measurements of water levels and discharge may also be available at a few locations.  These estimates provide water level information at points within the domain. Macroscale hydrologic modeling may also be used to produce estimates of tributary and unobserved lateral inflows.   All this information can be integrated into a hydraulic model using a data assimilation scheme. Data assimilation updates estimates of water levels and discharge hydrographs by weighting observations with coincident predictions by the hydraulic model.

Modeling of unsteady river flows involves the solution of the continuity and momentum equations for the river. The one-dimensional form of the equations, known as the de Saint-Venant equations, will be solved using CHARIMA, a general-purpose computer code for the simulation of simple or complex systems of river channels.  Macroscale hydrologic modeling will be carried out using ground observed and satellite data, at a spatial resolution of one degree (latitude × longitude), for a case study in the Mississippi River basin between 1990 to 1995.

Flowchart showing the proposed river discharge estimation system, where water level information from satellite radar altimetry and other information on flows is integrated into the hydraulic model using a data assimilation approach.

Specific research tasks for this project include:

Development of data assimilation components suited to the observations of water levels from satellite and in-situ sensors. Although approaches have been developed for real-time prediction, dealing with the unique conditions of remote sensing observations will be a challenge (i.e., estimates of water levels for updating are made infrequently and at different times for points along the river).

Evaluation of the discharge estimation procedure based on extensive simulation experiments with idealized conditions. Simulations of remotely sensed water levels (representative of SRA measurements) will provide a means for assessing the sampling requirements for satellite-estimates of water levels and the data requirements for river channel representation.

Development and assessment of water level estimates from the ERS-1 and TOPEX/POSEIDON satellites for case study applications.

Application and evaluation of the system in both data rich and data sparse environments. Our first application will be in the data rich area of the Mississippi River basin. A second application will be developed (in the latter stages of the research) for a data sparse region.

Can the dynamic information on flow movement through a river network, obtained by repeated estimates of water levels by satellite remote sensing, be exploited in river discharge estimation? What accuracy is required of remotely sensed water levels for this framework?  Does macroscale hydrology modeling provide valuable information on tributary and local flows for improved estimation?

Are there sufficient physical data available in data sparse regions to permit the coupled approach? Are approximations of river channel bathymetry sufficient where detailed surveys of river cross-sections are unavailable?  Can satellite remote sensing of land cover and surface fluxes be used to facilitate macroscale hydrologic modeling in data sparse regions?

The simulation experiments and case study applications carried out in this research are designed to answers these questions.  The results should help in assessing whether satellite estimates of water levels, and other remote sensing information, can be combined with hydraulic and hydrologic modeling, to produce accurate estimates of discharge in data sparse regions of the world.