HR: 14:10h
AN: H43H-03    [Abstracts]
TI: Exploiting Equilibrium Tendencies of Soil Moisture Dynamics for Evaporation Estimation
AU: * Salvucci, G D
EM: gdsalvuc@bu.edu
AF: Boston University, Dept. of Earth Sciences 685 Commonwealth Ave., Boston, MA 02215 United States
AB: Two aspects of soil moisture dynamics that can be exploited to parameterize the relation between hydrologic fluxes and soil moisture (s) at both the point-scale and large-scale are explored. They are: 1) the expectation of soil moisture change (ds/dt) conditioned on moisture level is equal to zero. Applying this principle to the water balance equations allows direct, empirical estimation of the relation between soil moisture and total water loss (the sum of drainage d, evapotranspiration et, and runoff ro ) from precipitation measurements conditionally averaged according to soil moisture level; and 2) the variance of the sum of model-estimated soil moisture changes (i.e. var [p-et(s)-d(s)-ro(s)]) will contain a term that saturates to the actual soil moisture variance and a term that grows with time. The first term saturates because of the strong negative correlation between increments: i.e., days with positive perturbations in accumulation physically cause days with negative perturbations to occur in the future. The second term grows as a consequence of summing errors due (in part) to poor parameter specification. Thus by minimizing the total variance, we can estimate model parameters, and thereby partition the total water loss term estimated from principle 1 into the evapotranspiration and drainage/runoff components. Most critically, this methodology requires only intermittently sampled forcing data (e.g. p and s), no calibration data (e.g., continuous measurements et, d, or ds/dt), and no computationally expensive simulations. The requirements are more parsimonious than alternative methods because the objective function that we minimize is not a measure of actual prediction error, but rather a measure of how nonstationary (and thus error prone) the associated modeled soil moisture process would be for a given set of parameters and forcings. Applications will be presented.
DE: 1816 Estimation and forecasting
DE: 1818 Evapotranspiration
SC: Hydrology [H]
MN: Fall Meeting 2005