An Adaptive Spatial Model for Precipitation Data from Multiple Satellites Over Large Regions

• Activity Number: 505
• Type: Invited
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: International Indian Statistical Association
• Author(s): Bani Mallick*+
• Companies: Texas A&M
• Keywords: large data computation ; nonstationary spatial model ; predictive process ; random knots ; satellite measurements.

Abstract

Satellite measurements have of late become an important source of infor- mation for climate features such as precipitation due to their near-global coverage. In this presentation, we look at a precipitation dataset during a 3-hour window over tropicalSouth America that has information from two satellites.We develop a exible hierarchi-cal model to combine instantaneous rainrate measurements from those satellites while accounting for their potential heterogeneity. Conceptually, we envision an underlyingprecipitation surface that inuences the observed rain as well as absence of it. The surface has been specied using a mean function centered around a set of knot locations, to capture the local patterns in the rainrate, combined with a residual Gaussian process to account for global correlation across sites. To improve over the commonly used prexed knot choices, an ecient reversible jump scheme has been used to allow the number of such knots as well as the order and support of associated polynomial terms to be chosen adaptively. To facilitate computation over a large region, a reduced rank approximation for the parent Gaussian process has been employed.