HR: 0800h
AN: C41A-0180 [Abstracts]
TI: Quantifying Uncertainties in Large Scale Water Budget: Case Study in Siberia
AU: * Joe, S
EM: sjoe@umd.edu
AF: University of Maryland, College Park, Dept. of Civil & Environmental Engineering
1173 Glenn L. Martin Hall, College Park, MD 20742
United States
AU: Brubaker, K L
EM: klbrubak@umd.edu
AF: University of Maryland, College Park, Dept. of Civil & Environmental Engineering
1173 Glenn L. Martin Hall, College Park, MD 20742
United States
AB:
Assessment and prediction of Arctic River flows' effects on ocean
circulation and climate are hindered by lack of knowledge about the
terrestial water balance in remote regions. In
this study, we quantify the components of the annual water
budget for a large Siberian river basin and -- most importantly -- the
uncertainty in the components. The water budget for a watershed can be
simplified to basic inputs and outputs: Precipitation (P), Streamflow (Q),
and Evapotranspiration (E). Over the long term, assuming negligible
change in storage, inputs and outputs should balance, P = Q + E. However,
errors in measuring and estimating the components lead to a nonzero
closure error, CE = P - Q - E. The uncertainty in the water balance can
be quantified by the variance of CE, which is equal to the
sum of the component variances (assumed independent).
The closure error and its variance were estimated for the 57000 km$^2$ Tom
River basin (a subbasin of the Ob River) for five water years, 1981-
1985. We hypothesized that (a) the CE would be negative due to
underestimation of P by the sparse, low-elevation precipitation network,
and (b) statistical hypothesis testing would show that the CE
is not significantly different from 0, due to uncertainty
in the components. The basin mean and variance of P were estimated by
kriging station observations. The annual mean Q was obtained from
discharge measurements at Tomsk, Russia; the uncertainty in Q was based on published estimates of rating curve error bars.
The
basin mean and variance of E were computed from a derived distribution
based on Monte Carlo simulation of the Penman Monteith model, driven by
measured meteorological data at Tomsk, and accounting for variation in
elevation and vegetation. Annual CEs were negative,
ranging from -160 to -325 mm, and the standard deviations
ranged from 50 to 60 mm. The CE was significantly different from 0 for all five water years, supporting the
belief that annual P is underestimated by the gage network.
The level of
uncertainty, largely due to E, makes it difficult to assess the accuracy of P. The
uncertainties in the components and the CE are an indication of
the confidence in statements about the water balance of the region under
current or changed conditions.
DE: 9315 Arctic region
DE: 9320 Asia
DE: 1836 Hydrologic budget (1655)
SC: Cryosphere [C]
MN: 2004 AGU Fall Meeting