Estimating the global terrestrial hydrologic cycle through modeling, remote sensing, and data assimilation [INVITED]

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 25 January 2011: 8:30 AM
Estimating the global terrestrial hydrologic cycle through modeling, remote sensing, and data assimilation [INVITED]
612 (Washington State Convention Center)
Eric F. Wood, Princeton University, Princeton, NJ; and M. Pan, A. Sahoo, J. Sheffield, and T. Troy

Documentation of the global water cycle and its evolution over time is a primary scientific goal of the Global Energy and Water Cycle Experiment (GEWEX) and fundamental to assessing global change impacts. In developed countries, observation systems that include in-situ, remote sensing and modeled data can provide long-term, consistent and generally high quality data sets of water cycle variables. The export of these technologies to less developed regions has been rare, but it is these regions where information on water availability and change is probably most needed in the face of regional environmental change due to climate, land use and water management. In these data sparse regions, in-situ data alone are insufficient to develop a comprehensive picture of how the water cycle is changing, and strategies that merge in-situ, model and satellite observations within a framework that results in consistent water cycle records is essential. Such an approach is envisaged by the Global Earth Observing System of Systems (GOESS), but has yet to be applied.

The goal of the underlying research is to develop consistent Climate Data Records that will enable one to quantify the variation and changes in the global water cycle over the past 50 years. We evaluate the global water cycle using a variety of independent large-scale data sets of hydrologic variables that are used to bridge the gap between sparse in-situ observations, including remote-sensing based retrievals, observation-forced hydrologic modeling, and weather model reanalyses. A data assimilation framework is used to blend these disparate sources of information together in a consistent fashion with attention to budget closure, resulting in a ‘best-estimate' of the global water cycle and its variation. The data assimilation framework consists of a constrained Kalman filter applied to the water budget equation. With imperfect estimates of the water budget components, the equation additionally has an error residual term that is redistributed across the budget components using error statistics, which are estimated from the uncertainties among data products. The constrained Kalman filter treats the budget closure constraint as a perfect observation within the assimilation framework.

The water budget components are estimated as follows. Precipitation using gauge observations, reanalysis products, and (for below 50ºN) remote sensing products. Evapotranspiration is estimated in a number of ways: from (i) the VIC land surface hydrologic model forced with a hybrid reanalysis-observation global forcing data set, (ii) using remote sensing retrievals, and (iii) using an atmospheric water budget approach using reanalysis products for the atmospheric convergence and storage terms and our best estimate for precipitation. Terrestrial water storage changes, including surface and subsurface changes, are estimated using estimates from both VIC and the GRACE remote sensing retrievals. From these components, discharge can then be calculated as a residual of the water budget and compared with gauge observations to evaluate the closure of the water budget. Through the use of these largely independent data products, we estimate both the mean seasonal cycle of the water budget components and their uncertainties for a set of 20 large river basins across the globe.