28–29 May 2020
WebEx
Europe/Brussels timezone

Multiple linear regression and dynamic linear model applied to Hohenpeissenberg total ozone

29 May 2020, 17:00
10m
WebEx

WebEx

Speaker

Wolfgang Steinbrecht (Deutscher Wetterdienst)

Description

Multiple linear regression does a surprisingly good job at reproducing the time series of annual mean total ozone since 1968 at Hohenpeissenberg. Hockey-stick and EESC fit give virtually the same long-term trends: significant decline from 1968 to 1997 and consistent increase from 1998 to now. Recently I also tried the Alsing & Ball DLM model - with very similar long-term variation. In my presentation I'll show these results, which are based on annual means. If I get around to it, I'll also show (noisier) results based on monthly means. Ultimately, I want to apply the regression and the DLM to ozone profile time series and have a closer look at trends in the lower stratosphere.

Primary author

Wolfgang Steinbrecht (Deutscher Wetterdienst)

Presentation materials