Planetary Wave Flux vs. Polar Temperature

Introduction

This diagnostics is designed to evaluate the response of a model's polar stratosphere to the forcing by planetary waves that propagate into the stratosphere from the troposphere. The calculation was first described in Newman et al. (2001). The temperature of the polar lower stratosphere during spring is the key factor in changing the magnitude of ozone loss in the Arctic polar vortex. The daily averaged eddy heat flux in middle to late winter (NH and SH) is highly correlated with the mean polar stratospheric temperature during late winter. Strong midwinter planetary wave forcing leads to below average spring temperatures, while weak midwinter forcing leads to above average temperatures.

This diagnostic was employed in the Austin et al. (2003) paper to determine if a variety of models could simulate this forcing.

Calculation Description

The heat flux is calculated from the temperature (T) and meridional wind (v). T = [T]+T', where [T] is the longitudinal average (or more commonly called the zonal mean) of the temperature, and T' is the deviation from that longitudinal average. It is critical that the heat flux ([v'T']) be calculated from the product of the zonal mean deviations.

Plots

The NH plot (date file in next section) shows the data as described above.

postscript file

The SH plot.

postscript file

Data

The data are derived from the NCEP/NCAR reanalysis 2 data set. The data used in Newman et al. (2001) and Austin et al. (2003) are slightly different than these numbers since they used the reanalysis 1 data set.

The NH data are the: [T] Feb.-Mar., 60-90N, 50hPa area weighted averages, and [v'T'] Jan.-Feb., 40-80N, 100hPa area weighted averages.

The SH data are the: [T] Aug.-Sep., 60-90S, 50hPa area weighted averages, and [v'T'] July-Aug., 40-80S, 100hPa area weighted averages.

The Temperature vs. Heat Flux Figures 4 & 5 in Austin et al. (2003) have an error, since the heat flux from the NCEP/NCAR reanalyses are calculated for a Jan.-Mar. average instead of the Jan.-Feb. that is indicated in the Figure 4 caption. Similarly, the heat flux average in the SH is for July-Sept., rather than the July-Aug. that is indicated in the figure. The correct values are in the files above.

Data Errors

There is a certain amount of error in the numbers used for estimating both the temperature and the heat flux.

The NH data and SH data show flux-temperature lines of the NCEP/NCAR reanalysis 2 data and the ECMWF ERA40 data. The datas are not quite identical, but yield fairly close results for the flux temperature relations.

The time series of NH data and SH data are shown for 5 different data sets. The NCEP MRF62 analyses, NCEP/NCAR REAN2 analyese, the UKMO analyses, the NCEP/CPC analyses, and the ECMWF ERA40 analyses. In the NH, temperatures are show agreement to within 0.5K. The NH flux esimates are typically within 2 Km/s (15%). In the SH, the temperatures show larger discrepancies of approximately 1K, while heat flux errors are typically about 2 K m/s (20%) errors.


Back to the CCMVal main page

Back to the CCMVal Evaluation Table



Last Updated: 2005-10-24
Author: Dr. Paul A. Newman (NASA/GSFC, Code 916) (newman@notus.gsfc.nasa.gov)
Responsible NASA organization/official: Dr. Paul A. Newman, Atmospheric Chemistry and Dynamics Branch