Wisconsin HIRS 6.5 Year Cloud Climatology

Description of the cloud statistics

Statistics on clouds have been compiled from the HIRS sensors on the NOAA satellites since June 1989. This is an on going process and will be updated in the future. The HIRS channels with partial CO2 absorption from 13-15 microns, provides a better means of detecting semi-transparent clouds than other cloud analyses The method has also been referred to as "CO2 Slicing." A description of the technique is given in:

Wylie, D.P., W. P. Menzel, H. M. Woolf, K. I. Strabala, 1994: Four Years of Global Cirrus Cloud Statistics Using HIRS., J. of Climate, 7, 1972-1986

Also, Wylie, D. P., and W. P. Menzel: 1999: Eight years of high cloud Statistics using HIRS. J. Climate, 12, 170-184.

The data presented here have been converted from height units of pressure to kilometers. A discussion of this conversion is given below. Corrections also have been made for known biases in the cloud height assignments.

These cloud statistics were compiled as a compliment to the International Satellite Cloud Climatology Program (ISCCP). The ISCCP currently does not use the HIRS data for cloud detection and height determination. The Wisconsin HIRS Cloud Climatology reports more clouds in the upper troposphere than the ISCCP. A comparison of the two data sets can be found in Jin, Rossow and Wylie, 1996, "Comparison of the Climatologies of High-Level Clouds from HIRS and ISCCP." Ask one of the authors for the latest version of this paper.

This web page offers the global distribution of clouds from the 6.5 years of HIRS data that have been analyzed. Cloud data continue to be taken and the monthly summaries will be updated.

The probability of clouds are presented for each of the 12 calendar months. The cloud probabilities have been accumulated from the top of the atmosphere downward - the manner in which the satellite views the clouds. You can select a month and height level to get the probability of clouds at and above the selected level.

Since the CO2 Slicing technique is sensitive to clouds of optical depths > 0.1 (visible), the cloud frequencies presented here are larger than other cloud climatologies. The frequencies of clouds that are opaque to terrestrial infrared radiation (v sible optical depths > 6) are also given as separate statistics.

Height Conversion

The working equation for the CO2 Slicing algorithm solves for the pressure height of the cloud. The equation is formulated in pressure units because pressure is a measure of the atmospheric mass from the working level to the top of the atmosphere and thus indicates the magnitude of atmospheric mass that radiation must pass through from the working level to the satellite sensor. A vertical coordinate of pressure is a natural unit for radiative transfer calculations.

To convert height units from pressure to kilometers, the density of the the atmosphere had to be included in the calculation. The atmospheric density profile was calculated from a standard atmospheric temperature profile. However the density profile varies greatly because of temperature gradients between the tropics and poles, and between different locations with in weather systems. The height conversion starts as a function of latitude. The standard atmosphere was modified for known latitude gradients in temperature. Than extra modifications were made using temperature data reported with the cloud heights. The technique reported the cloud level temperature with the height. This temperature was used to adjust the pressure to kilometer conversion, raising the height in warm areas and lower them in colder areas.

Height Correction

The cloud heights were corrected because comparisons with lidar showed that the cloud heights reported by the CO2 Slicing technique tended to be lower than true cloud tops for some cloud forms. For clouds that are very dense near their tops, the CO2 Slicing cloud top should agree with other cloud top measurements. However, for semi-transparent clouds the CO2 Slicing technique reports a level that is closer to the cloud center than the top. This difference occurs because the cloud level is calculated from the absorbtion and re-emission of infrared radiation propagating vertically through the cloud. When the satellite sensor views very dense clouds, all the vertically propagating IR radiation is comming from the cloud top or very close to it. But the radiation upwelling to the satellite through semi-transparent clouds comes from a combination of the ground or cloud below, the atmosphere below, and the cloud. The top and bottom of the cloud can not be determined uniquely from radiative transfer measurements. To do this, the vertical profile of absorbtion and transmission in the cloud have to be known. The only unique solution that can be made with out assumptions of the content of the cloud, is a one layered solution. All the absorbtion and re-emission of the cloud are assumed to occur at one level. For uniform density clouds that are semi-transparent, the CO2 Slicing solution will be the middle of the cloud or a center of radiative divergence.

Lidars have shown that semi-transparent clouds can be very thick - often over 5 km. However, our height correction is designed to be conservative so we assume that the maximum height error is 2 km. Our height correction varies for the density of the cloud. We known that height errors for very dense clouds are < 1 km, the resolution of the data summary. Thus no correction is made for the dense clouds. For very thin clouds, the lidar data indicate that radiatively thin clouds are also physically thin and usually 1 km or less. Thus the thin clouds also do not have any corrections. A summary of the height corrects is given below.

O.D. is the estimated visible optical depth computed from the infrared emissivity (NE) derived from the CO2 Slicing technique.

Visible O.D. = 2 X IR O.D. = 2 X (-ln(1.0-NE))

22 Year Summary

Some images and text files from the 22 year record are available on the same ftp site as the older data.
Ftp to ftp.ssec.wisc.edu
Login as anonymous
Password is your email address
cd wylie/22_Year_Average

This directory has images of the frequency of all clouds, high clouds, and opaque clouds for each calendar month. Also in the same directory are text files that have been unix compressed, (*.txt.Z). These files have the frequency of clouds summed from the top of the atmosphere downward to 9 pressure levels. The frequency of only the clouds opaque to IR radiation also are in the same files. On my Windows XP system Winzip also will uncompress the files.

The monthly averages for the 22 year HIRS record is very similar to the 6.5 year record. Trends were very small. I'm writing a paper on this and will give more details later.

Obtaining Digital Copies of the 11 year record.

A copy of the cloud frequency statistics in ASCII are available at an FTP site.
Ftp to ftp.ssec.wisc.edu
Login as anonymous
Password is your email address
cd wylie

There are two forms of the cloud frequency statistics. The km directory has the vertical coordinate of km and the press directory has pressure for a vertical coordinate. Both directories have the same cloud frequency statistics which are cloud frequencies vertically cumulated from either the top of the atmosphere downward, or the surface upward. All data files are ascii text. Each directory has a file 0-format.txt which gives details on the formats.

The statistics are taken from 11 years of HIRS data from 1979-2000. They will revised in 2003 using a re-analysis of HIRS in the NOAA Pathfinder program. The revisions are expected to be small in these statistics.

Don Wylie
Phone 608 263 7458
Fax 608 262 5974