A Reduction in Systematic Errors of a Bayesian Retrieval Algorithm

Seo, Eun-Kyoung Florida State University
Liu, Guosheng Florida State University
Kim, Kwang-Yul Texas A&M University

Category: Cloud Properties

Using satellite ice water path (IWP) retrievals as an example, a technique is described to reduce systematic errors in retrieving meteorological variables having discontinous and heavily skewed data distributions from a supporting database using Bayesian theorem. Discontinuity in the data distribution occurs at since there are no negative IWPs, which cause a positive bias in IWP retrievals near ice-free conditions. We propose to reduce this positive bias at by extending the database to negative IWPs that mirror the positive IWP data points. On the other hand, since clouds with small IWPs occur more often than clouds with large IWPs in nature, data points in the database are heavily populated in the low IWP range if we build the supporting database based on observed atmospheric conditions. This skewed data distribution leads to a negative bias. To reduce the negative bias, we introduce an empirical transform function , so that is the variable to be directly retrieved by the Bayesian retrieval algirthm instead of IWP itself. Although it is explained using IWP retrieval as an example in this study, the technique with an appropriate transform function representing each database can be easily adopted for retrieving other variables using Bayesian type algorithms.

This poster will be displayed at the ARM Science Team Meeting.