Bayesian Analysis of Orographic Precipitation Sensitivity to Upwind Sounding and Terrain

Orographic precipitation is a key component of the hydrologic cycle. The amount and distribution are known to depend on mountain and atmospheric sounding characteristics: mountain height and width, wind speed, stability, temperature, and humidity. Numerical models, including the Weather Research and Forecasting (WRF) model and the CM1 cloud-resolving model (Bryan and Fritsch, 2002), are useful tools for exploring the response of precipitation to atmosphere and mountain characteristics, and have been used to model moist neutral flow over a mountain (Miglietta and Rotunno, 2005, 2006, 2009, 2010). However, exploring all possible combinations of mountain and sounding configurations with a numerical model is computationally exhausting.

Bayesian techniques, namely Markov chain Monte Carlo (MCMC) algorithms, were recently used to explore parameter sensitivities in remote sensing retrievals and model parameterizations in a computationally efficient way (Posselt et al., 2008; Posselt and Vukicevic, 2010; Posselt and Mace, 2014). However, use of MCMC to explore iteractions in the physical system is an original concept. Here we introduce a MCMC-based sensitivity analysis that, in conjunction with the CM1 model, provides insight into combinations of atmospheric and mountain parameters consistent with a given precipitation distribution.