Compiled Tower/Microwave Radiometer Data Set


Observed and ECMWF model-predicted downwelling longwave radiation, for January-March 1998 (Bretherton et al. 2000, Fig. 14).


Daily-averaged liquid water path from the MWR and from the ECMWF, and ice water path from ECMWF, for March 1998. (Bretherton et al. 2000, Fig. 13).


Summary of all variables in





Julian day (1997)

decimal day


latitude SHEBA site



longitude SHEBA site



surface pressure



wind velocity at 2.5 m



wind velocity at 10 m



wind direction at 2.5 m

[degree north]


wind direction at 10 m

[degree north]


temperature at 2.5 m



temperature at 10 m



specific humidity at 2.5 m



specific humidtiy at 10 m



relative humidity with respect to ice at 2.5 m



relative humidity with respect to ice at 10 m



surface temperature (best estimate) (1)



longwave downwards



longwave upwards



shortwave downwards



shortwave upwards



shortwave upwards, albedo Perovich (2)



albedo from tower shortwave measurements



albedo from Perovich data (2)



precipitation (optical raingauge) (3)



precipitation (weighing gauge) (3)



precipitation, SPO corrected (3)



friction velocity (eddy correlation)



sensible heat flux (eddy correlation)



latent heat flux (eddy correlation)



ustar from tower at 2.5 m



ustar from tower at 10 m



sensible heat flux from tower at 2.5 m



sensible heat flux from tower at 10 m



latent heat flux from tower at 2.5 m



latent heat flux from tower at 10 m



MWR water vapor path



MWR liquid water path (4)


(1)Three surface temperature measurements were measured from a General Eastern, an Eppley radiometer and a Barnes radiometer. The Eppley is the most reliable, though there are periods when the other two are also reasonable, and one period (May) when the Eppley data may be slightly off. Tsfc is the best estimate of the surface temperature, and is principally based on slight corrections to the Eppley temperatures and the Barnes temperatures when the Eppley was known to be wrong.

(2) SWucor has been computed from the observed downwelling shortwave radiation multiplied by the line-averaged albedo measured by Don Perovich for the period between 1 June 1998 and 27 September 1998. The line albedo for this period is substantially lower than the tower albedo and more representative for the area-averaged albedo.

(3) The SHEBA Project Office (SPO) maintained a Nipher shielded snow gauge system about 300 m from the SHEBA ship, which was visited daily around 20 UTC. Corrections were made for wind, losses due to (temperature-dependent) evaporation and gauge wetting, and precipitation amount in intervals in which the reported weather is blowing (rather than falling) snow are set to zero. We regard the corrected SPO precipitation time series as the most trustworthy currently available for SHEBA.

The flux tower group had both a ETI NOAH-II weighing gauge (which sends a pulse every time it has accumulated 0.254 mm of water equivalent in an bucket with antifreeze in it), and an STI model 815 optical raingauge. The optical raingauge data can not be used for quantitative measurement of water equivalent during the nine months of the SHEBA year when snow was common, and even during rainy periods in the summer season, it indicated 2-3 times as much precipitation as the other raw observations.

(4)There is still some controversy about the accuracy of the microwave radiometer (MWR) liquid water path (LWP) retrievals, since the most reliable in-situ FIRE-ACE aircraft measurements of LWP over SHEBA (taken in May-July 1998) average only half as large as the simultaneous MWR LWP (Judith Curry, personal communication).


The default value for all variables is 9999, except for wvp and lwp (default = -99.99). The netCDF file contains hourly averaged data collected by the SHEBA Atmospheric Surface Flux Group and Dr. J. Liljegren from the ARM project. The file can be downloaded by anonymous ftp: ftp, directory pub/roode (netCDF, 2.6 Mb)




"We thank our colleagues in the SHEBA Atmospheric Surface Flux Group, Ed Andreas, Chris Fairall, Peter Guest, and Ola Persson for help collecting and processing the data. The National Science Foundation supported this research with grants to the U.S. Army Cold Regions Research and Engineering Laboratory, NOAA's Environmental Technology Laboratory, and the Naval Postgraduate School."


Bretherton, C. S., S. R. de Roode, C. Jakob, E. L Andreas, J. Intrieri, R. E. Moritz, and P. O. G. Persson, 2000: A comparison of the ECMWF forecast model with observations over the annual cycle at SHEBA. J. Geophys. Res., submitted 12/99, revised 5/00. Postscript version

Persson, P. O. G., C. W. Fairall, E. L. Andreas, and P. S. Guest, 2000: Measurements of the meteorological conditions and surface energy budget near the atmospheric surface flux group tower at SHEBA, J. Geophys. Res., submitted.

Stephan de Roode (
Chris Bretherton (

University of Washington
Dept of Atmospheric Sciences
Box 351640
Seattle, WA 98195-1640 USA


Here follows the text of the readme file of the data set (main_file3_hd.txt) that was used to make the netCDF data file. (source:, 9 May 2000.)


TOWER README - Atmospheric Surface Flux Group Data from "Met City"
Our current version is ASFG2.0 and was completed in October 1999. Fluxes are calculated using the observed surface pressure (at Florida) rather than an assumed constant one. Wind direction is true wind direction accounting for the rotation of the tower during the year. Both objective and some subjective editing has been done at various stages of the data processing. Hourly averages were calculated as long as at done at various stages of the data processing. Hourly averages were calculated as long as at least 4 10-minute periods during the hour contained 2 or more minutes of good data. Fluxes are also eliminated when the airflow was from the ship or through the tower. The data does include intercomparison calibrations done during the year. I am currently in the process of writing up a detailed description of the data processing procedure. One file (prof_file_all3_ed_hd.txt) contains data from all 5 levels on the tower as well as the radiometer and surface measurements (include the Barnes, GE and the two thermistors). The tower data includes sensible heat and momentum fluxes as well as the RH, wind speed, temperature data. This file also includes variances, structure functions, etc for each level. The other file (main_file3_hd.txt) is derived from the first file by taking the median fluxes and interpolating the temperature, humidity and wind to 2.5 and 10-m. This file does not contain an y of the variances, etc, but does contain bulk estimates of latent and sensible heat fluxes at 2.5 and 10 m. Each file has a header that explains each column of data. Note that rhi is relative humidity with respect to ice recalculated from the original measuremen t of relative humidity with respect to water. twr_orien is the orientation of the tower sonic boom a rms, which was used to calculate the true wind direction wd. Three surface temperature measurements are available from the General Eastern, the Eppley radiometer and the Barnes radiometer. The Eppley is the most reliable, though there are periods when the other two are also reasonable, and one period (May) when the Eppley data may be slightly off. Tsfc is our bes t estimate of the surface temperature, and is principally based on slight corrections to the Epple y temperatures and the Barnes temperatures when the Eppley was known to be wrong. Check with one of the PIs on the latest status of the estimated errors. The measurements of stress an d sensible heat flux are the median values of the levels with "good" measurements. Eddy correlation measurements of the latent heat flux from the Ophir instruments are not yet availabl e. The bulk estimates of stress, sensible and latent heat flux are calculated using a modified COARE flux algorithm that computes fluxes over the ocean or sea ice. For ice it uses Andreas 1987 for Ch and Ce; presently it sets zo=4.5e-4 m. Other data files are also available from the four PIs. These data files include similar data as above but with higher temporal resolution, files of covariance and quadrature spectra for each hour and level, and sodar data. This data is currently at various stages of development. We are currently working on a detailed description of the processing that was done for each parameter. In the meantime, let us know if you have any questions, and especially let us know i f you find anything odd or an obvious error.


Ola Persson


Edgar L. Andreas
U.S. Army Cold Regions Research and Engineering Laboratory (CRREL)

Christopher W. Fairall
NOAA Environmental Technology Laboratory (ETL)

Peter S. Guest
Naval Postgraduate School (NPS)

P. Ola G. Persson
ETL / Cooperative Institute for Research in Environmental Sciences (CIRES)


The hourly averages in are based on a data set kindly provided by Dr. J. Liljegren. The original file has 757 measurements per day (every 114 seconds one measurement). Hourly averages were computed only if 67% of the data during this hour had flag=0 (OK). If the total number of data with flag = 0 was less then this threshold value the data for these periods were set to default = -99.99.


Here follows the text of the original readme file.



AUTHOR: James Liljegren

PERIOD: 5 December 1997 - 9 September 1998

The MWR data during SHEBA that appear in the netCDF files from ARM named "shbmwrtip" include the zenith brightness temperatures (TB) derived from elevation angle scans ("tip curves"). These are valid only when the sky is clear. A noise diode provides a calibrated reference when the sky is cloudy. The clear sky results are used to calibrate the noise diode. This is the calibration that I have performed. Subsequently I re-applied the statistical retrievals to to these data to derive the correct precipitable water vapor (PWV) and liquid water path (LWP).

I developed the statistical retrievals using radiosondes launched from Barrow, Alaska during 1990-1995 stratified by month. Because the retrievals represent monthly-averaged statistics, significant departures from these mean conditions can lead to biases in the retrieved quantities. The most noticeable example is non-zero LWP during clear sky conditions. When the surface presure and temperature are such that the air density is significantly greater than the monthly mean implicit in the retrieval, a positive LWP bias is observed; as a low-pressure condition approaches the bias may become negative. This is a problem with the retrieval; it does not indicate a problem with the radiometer calibration. Improved retrievals which account for this and other problems will be applied to these data in the near future.


In examining the data so far I have determined that there are several problems:

1. Prior to 5 December 1997 the radiometer was operated in zenith-viewing line-of-sight (LOS) mode. The only calibration scans prior to 5 December 1997 were acquired for a period of 4 days in mid-October. Unfortunately, the radiometer configuration appears to have been incorrect during this calibration period. (I don't know how this happened; it was correct when I sent it to Barrow in March 1997.) I may be able to recover these data, but they are they are not included in this release.

2. Subsequent to 5 December 1997 the radiometer was operated in a continuous scanning (calibration) mode at my recommendation. I also provided the correct configuration file and a new executable file to replace those lost in a disk crash on 20 November 1997.

3. The additional insulation I provided with the radiometer for very cold periods was not installed. Consequently the thermal stabilization was affected at low temperatures (below about -25 C) which occurred frequently during December, January and February plus early March. Fortunately these cold temperatures occurred only when the skies were clear of liquid water clouds and so the tip curve data provide valid measurements during these periods. Unfortunately, because the radiometer software recognized this thermal stabilization failure, data are only available for approximately 15 minutes per hour during these periods of extreme cold.

4. The radiometer is equipped with a continuously-operating external blower that prevents snow and other debris from accumulating on its window. A 750 Watt heater mounted in the blower housing is controlled by a moisture sensor and is designed to prevent moisture from accumulating on the microwave window. Unfortunately the heater, which is on a separate power circuit so as not to burden a UPS, was plugged into the wrong outlet and was not powered. I did not discover this until late June when I was investigating why moisture kept accumulating on the window. Consequently there are periods when the data are invalid due to water on the microwave window. Periods when this problem appear likely have been flagged.