Contracting authority (client):
2015 - 2019
Seamless Hydrological prediction of East Indian summer monsoon and Variance Analysis of its meteorological and hydrological uncertainty
The implementation of so-called "seamless predictions" represents a new method of numerical weather prediction in which all time scales, from days to several months, are treated uniformly. In this way, each scale benefits from the research progress of the other scales. On the other hand, probabilistic statements using ensemble calculations are the de facto standard in operational forecasts in both meteorology and hydrology. The underlying physics and statistics are, however, very different for the entire forecast range for both disciplines. Meteorological predictability is mainly limited by the uncertain initial state. The total uncertainty in the forecast period is represented by initialising each ensemble member once appropriately and then letting it evolve according to deterministic dynamics. For any hydrological forecast, this meteorological uncertainty represents an essential boundary condition. However, there are other factors that are at least as important: the structure of the hydrological model and its parameters, which together are usually summarised as hydrological uncertainty. This is mainly due to the much more heterogeneous object of research, because a river basin has significantly more independent components than the global atmosphere. The combination of meteorological and hydrological uncertainty for a "seamless prediction" of runoff is therefore a considerable difficulty and is addressed in two of a total of three parts in the SHIVA project proposal.The third part analyses such predictions for the respective shares of meteorological and hydrological uncertainty. These can be quantified with the so-called Analysis Of VAriance (ANOVA). There are three important aspects here.
1) special predictors are defined for the ANOVA for the short, medium and long term;
2) possible sources of error are revealed by the size of each proportion for a given forecast time;
3) the resulting relationship between forecast time and the uncertainty proportions provides a general picture of the principle limits to the predictability of discharges for the entire forecast range.
The Mahanadi catchment (A_E= 141,500km²) in eastern India is an ideal study object for this question. Firstly, because strong couplings between ocean and atmosphere in the tropical monsoon belt make seasonal forecasts possible in the first place (in contrast to conditions in the mid-latitudes, for example). Secondly, runoff forecasts are generally of utmost importance for water management there, across the entire time range from daily to seasonal forecasts: for example, for flood warning, the control of multiple-use reservoirs, and irrigation management.