Joint mapping of water balance components in a large Chinese basin

A technique is presented for a joint interpolation of water balance components that considers estimated or observed values with their a priori uncertainty and yields balanced interpolated values, which respect the water conservation at all levels within the hydrographical network. The approach allows building a consistent water balance pattern, where both the observed values and water balance constraints are given full recognition. The framework presented has strong similarities with assimilation, as it is understood in meteorology (i.e. tuning a model to observed data). It proved in this study to be an excellent diagnostic tool for data in an exploratory phase when investigating patterns of variability. Patterns of spatial variation of precipitation, runoff and actual evapotranspiration have been studied at an annual time scale with data from the drainage basin upstream Bengbu in Huaihe River, China. Stochastic interpolation using data from the meteorological and hydrological networks were used to produce a set of prior digital maps of precipitation, actual evapotranspiration and runoff. In a second step an algorithm for balancing errors was applied to produce a set of posterior maps that satisfied the water balance. The network of runoff stations was sparse giving rise to high uncertainties in the interpolated runoff map with an average relative interpolation error of 15%. The precipitation map was more accurate (average interpolation error 10%), having a more dense observations network. The correction algorithm with the support in the more accurate precipitation map was able to compensate for the gaps in the runoff gauging net and smoothed out inconsistencies in the observed runoff values.