Spatio-temporal Bayesian modeling of precipitation using rain gauge data from the Hubbard Brook Experimental Forest, New Hampshire, USA

Abstract

Estimating precipitation volume over space and time is essential for many reasons such as evaluating air quality, determining the risk of flood and drought, making forest management decisions, and developing strategies for municipal water supplies. It is imperative to employ sound statistical methods for modeling data from a network of sparsely located rain gauges with known confidence. This paper proposes a spatio-temporal Bayesian model for estimating precipitation volumes using observations from a network of gauges. Based on Gaussian processes, the Bayesian model is able to interpolate at a high spatial resolution at each time point. Such interpolations are used to obtain various spatio-temporally aggregated statistics, such as annual precipitation volume in a local area. Markov chain Monte Carlo based model fitting, employed here, allows estimation of uncertainty that can be used in decision making. These methods are applied to a large data set of weekly precipitation volumes collected over the years 1997-2015 at the Hubbard Brook Experimental Forest (HBEF) in New Hampshire, USA. Using the proposed methodology we estimate trends in annual precipitation volumes spatially aggregated over nine gauged watersheds in the HBEF. The proposed modeling is also used to demonstrate a method for determining how to downsize a rain gauge network.

More information

Identifiers

ORCID for Sujit Sahu: orcid.org/0000-0003-2315-3598

Catalogue record

Export record