|
Abstract:
|
Hydrological routing models use river connectivity information to propagate the localized lateral inflows of surface and subsurface water runoff into downstream flows. The resulting modeled flows can be used for planning and risk analysis, which has motivated the determination of standard errors for flows. We describe computational tradeoffs among several approaches for determination of streamflow uncertainties, which generally correspond to different assumptions about the spatial/temporal covariance of inflows from runoff. We introduce a "reach random effects" model to account for large-scale error correlation, as may be caused by spatially-correlated errors in precipitation forcing. We describe implementation of uncertainty propagation using RAPID (David et al. 2011) applied over the 650,000 reaches of the Western Contiguous United States covered by the NHDPlus network. Finally, we observe that new space missions should provide novel remote-sensing observations of flows at sparsely-sampled points in the river network. We use the accessibility of the full space-time flow covariance to understand the constraints on network flows offered by these new observations.
|
