Statistics – Applications
Scientific paper
2011-02-21
Statistics
Applications
Scientific paper
A hierarchical Bayesian dynamic model for spatio-temporal data is presented and applied to obtain short term predictions of rainfall. The model incorporates physical knowledge about the underlying processes, such as advection, diffusion, and convection. It is based on a temporal autoregressive convolution with spatially colored and temporally white innovations. By linking the advection parameter of the convolution kernel to an external wind vector, the model is temporally non-stationary. Further, it allows for non-separable and anisotropic covariance structures. The temporal Markovian structure offers computational benefits and uses the inherent order in the time domain. With the help of the Voronoi tessellation, a natural parametrization, that is space as well as time resolution consistent, for data lying on irregular grid points is obtained. The model is used to predict three-hourly precipitation. It performs better than a separable, stationary, and isotropic version, and it also outperforms a deterministic numerical weather prediction model. In addition, we introduce a new tool, the "primary posterior predictive density" (PPPD), for assessing the fit of Bayesian models.
Künsch Hans R.
Sigrist Fabio
Stahel Werner A.
No associations
LandOfFree
A Dynamic Non-stationary Spatio-temporal Model for Short Term Prediction of Precipitation does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with A Dynamic Non-stationary Spatio-temporal Model for Short Term Prediction of Precipitation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and A Dynamic Non-stationary Spatio-temporal Model for Short Term Prediction of Precipitation will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-520558