Detecting Non-Gaussian and Lognormal Characteristics of Temperature and Water Vapor Mixing Ratio
Abstract
Many operational data assimilation and retrieval systems assume that the errors and variables come from a Gaussian distribution. This study builds upon previous results that shows that positive definite variables, specifically water vapor mixing ratio and temperature, can follow a non-Gaussian distribution and moreover a lognormal distribution. Previously, statistical testing procedures which included the Jarque-Bera test, the Shapiro-Wilk test, the Chi-squared goodness-of-fit test, and a composite test which incorporated the results of the former tests were employed to determine locations and time spans where atmospheric variables assume a non-Gaussian distribution. These tests are now investigated in a "sliding window" fashion in order to extend the testing procedure to near real-time. The analyzed 1-degree resolution data comes from the National Oceanic and Atmospheric Administration (NOAA) Global Forecast System (GFS) six hour forecast from the 0Z analysis. These results indicate the necessity of a Data Assimilation (DA) system to be able to properly use the lognormally-distributed variables in an appropriate Bayesian analysis that does not assume the variables are Gaussian.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFMNG31A0150K
- Keywords:
-
- 3315 Data assimilation;
- ATMOSPHERIC PROCESSES;
- 1910 Data assimilation;
- integration and fusion;
- INFORMATICS;
- 1990 Uncertainty;
- INFORMATICS;
- 3260 Inverse theory;
- MATHEMATICAL GEOPHYSICS