IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 43, NO. 5, MAY 2005 Abstract—We show that observed biases in retrievals of temperature and water vapor profiles from a twelve-channel microwave radiometer arise from systematic differences between the observed and model-calculated brightness temperatures at five measurement frequencies between 22 and 30 GHz. Replacing the value for the air-broadened half-width of the 22-GHz water vapor line used in the Rosenkranz absorption model with the 5% smaller half-width from the HITRAN compilation largely eliminated the systematic differences in brightness temperatures. An a priori statistical retrieval based on the revised model demonstrated significant improvements in the accuracy and vertical resolution of the retrieved temperature and water vapor profiles. Additional improvements were demonstrated by combining the MWRP retrievals with those from the GOES-8 sounder and by incorporating brightness temperature measurements at off-zenith angles in the retrievals. Index Terms—Microwave remote sensing, spectroscopy, thermodynamic profile retrieval, water vapor absorption. I. INTRODUCTION HE Atmospheric Radiation Measurement (ARM) Program has operated a twelve-channel microwave radiometer profiler (MWRP) [1] since February 2000 at its Southern Great Plains (SGP) site near Lamont, OK. The MWRP provides real-time vertical profiles of temperature, water vapor, and limited-resolution cloud liquid water from the surface to 10 km in nearly all weather conditions, at approximately 5-min intervals. In contrast to radiosondes, the MWRP provides substantially improved temporal resolution but coarser vertical resolution that declines in proportion to the height above ground level. In this regard, the MWRP data may be more appropriate inputs to numerical weather, climate, and cloud models, which have Manuscript received May 13, 2004; revised June xx, 2004. This work was supported by the Climate Change Research Division, U. S. Department of Energy, Office of Science, Office of Biological and Environmental Research, under contract W-31-109-Eng-38, as part of the Atmospheric Radiation Measurement Program. Argonne National Laboratory is operated by the University of Chicago for the U. S. Department of Energy. J. C. Liljegren is with Argonne National Laboratory, Argonne, IL 60439 USA (phone: 630-252-9540; fax: 630-252-9792; e-mail: jcliljegren@anl.gov). S. A. Boukabara, K. Cady-Pereira, and S. A. Clough are with Atmos- pheric and Environmental Research (AER), Inc., Lexington, MA. time steps ranging from 10 s to 30 min but only 20-50 vertical levels [2]. Modelers generally reduce the vertical resolution of the soundings by averaging over the vertical layers of the model. In evaluating the MWRP for the ARM Program, Liljegren [3] observed significant biases, in comparison with radiosonde data, in the water vapor and temperature profiles retrieved from the MWRP with the artificial neural network algorithms supplied by the manufacturer [4], which were based on the Rosenkranz absorption model [5]. This finding is in agreement with the previous results of Gueldner and Spaenkuch [6]. II. REVISED ABSORPTION MODEL We compared brightness temperatures measured in the five K-band channels (22.235, 23.035, 23.835, 26.235, and 30.0 GHz) that span the water vapor resonance centered at 22.235 GHz with calculations based on the absorption model of Rosenkranz [5] with recent modifications [7] (hereinafter R03). To ensure that any dry bias in the radiosondes used in the R03 model calculations did not affect the brightness temperature comparison, ARM’s scaled radiosonde product (sgplssondeC1.c1) was used. In this product the relative humidity of the radiosonde is scaled linearly, so that the integrated precipitable water vapor (PWV) matches the PWV derived from a collocated two-channel microwave radiometer (MWR) that measures brightness temperatures at 23.8 and 31.4 GHz. In addition, using the scaled soundings substantially reduces the variability in the model calculations arising from the variability in the radiosonde humidity calibrations. Prior to April 2002, the retrieval used by ARM to obtain PWV from the MWR brightness temperatures for scaling the radiosondes was based on the Liebe and Layton [8] absorption model. After April 2002 the retrieval – and sonde scaling – was based on [5], which yielded 2% larger values of PWV. For this study we have rescaled the radiosondes from the period after April 2002 to be consistent with [8]. Because radiosondes do not measure cloud liquid water amount, the comparisons were limited to liquid-water-cloud- free conditions by requiring that the magnitude of the retrieved liquid water path (LWP) was less than 50 g/m 2 (the expected The Effect of the Half-Width of the 22-GHz Water Vapor Line on Retrievals of Temperature and Water Vapor Profiles with a Twelve- Channel Microwave Radiometer James C. Liljegren, Sid A. Boukabara, Karen Cady-Pereira, and Shepard A. Clough T