A Bayesian state space modelling approach to probabilistic quantitative precipitation forecasting

Abstract

The generation of very short range forecasts of precipitation in the 0-6 hours time window is traditionally referred to as nowcasting. Most existing nowcasting systems essentially extrapolate radar observations in some manner, however, very few systems account for the uncertainties involved. Thus deterministic forecast are produced, which have a limited use when decisions must be made, since they have no measure of confidence or spread of the forecast. This paper develops a Bayesian state space modelling framework for quantitative precipitation nowcasting which is probabilistic from conception. The model 1 treats the observations (radar) as noisy realisations of the underlying true precipitation process, recognising that this process can never be completely known, and thus must be represented probabilistically. In the model presented here the dynamics of the precipitation are dominated by advection, so this is a probabilistic extrapolation forecast.The model is designed in such a way as to minimise the computational burden, while maintaining a full, joint representation of the probability density function of the precipitation process. The update and evolution equations avoid the need to sample, thus only one model needs be run as opposed to the more traditional ensemble route. It is shown that the model works well on both simulated and real data, but that further work is required before the model can be used operationally.

NORA–Nowcasting of Orographic Rainfall by means of Analogues

“…Lagrangian extrapolation techniques make use of different methodologies (e.g. Austin and Bellon, 1974;Bellon and Austin, 1978;Burlando et al, 1996;Seed, 2003;Germann and Zawadzki, 2004;Cornford, 2004). However, as extensively reported by Reyniers (2008), there are two main families of Lagrangian persistence nowcasting systems: cell tracking and area tracking.…”

On the propagation of uncertainty in weather radar estimates of rainfall through hydrological models

“…In using data from different sources it is clearly necessary to combine these different data. One approach to this problem is to use a stochastic state-space model with a Kalman filter procedure allocating weights to each data form, based upon the respective uncertainty of each observation type and of the predictions (see Grum et al, 2002;Cornford, 2004). Recognizing that flow predictions made with such a model will nevertheless remain uncertain, a Bayesian post processor (Krzysztofowicz, 1999) may be used to analyse components of the output error associated with particular data input types as demonstrated by Collier and Robbins (2008) for an urban drainage modelling system.…”

A Review of Quantitative Precipitation Forecasts and Their Use in Short- to Medium-Range Streamflow Forecasting

“…Several studies have addressed ensembles generated from radar data (e.g., Cornford 2004;Germann et al 2009) but the authors are unaware of any examples of operational streamflow forecasts that use ensemble precipitation from a blend of radar and NWP model outputs. Germann et al (2009) generated an ensemble radar precipitation product and fed this ensemble to a semidistributed hydrologic model for flash flooding in a mountainous Alpine catchment.…”