Tropical upper-tropospheric dry regions from TOVS and rawinsondes

Water vapor (WV) radiances from the TIROS-N Operational Vertical Sounder (TOVS) and rawinsondes during January 1979 are intercompared within the equatorial dry zone (EDZ) southwest of Mexico and the subtropical high (STH) southwest of Hawaii. Discrepancies are found when inferring upper-tropospheric moisture over the EDZ from different instrument platforms. This paper reconciles these discrepancies between radiances and operationally measured moisture profiles, quantifies the impact of trace amounts of upper-tropospheri WV on satellite-measured radiances, and summarizes limitations in the use of WV channel data for integra ed precipitable water (PW) estimates. Contradictory upper-tropospheric moisture information is obtained over the EDZ when operational rawinsonde data and outgoing longwave radiation (OLR) are compared to TOVS WV radiances. However, no such contradiction is observed over the STH when all three data sources indicate dry conditions in the upper troposphere. This dryness is depicted in histograms of 6.7- and 7.3-mu m WV brightness temperatures T-b where a subset of excessively warm (dry) values exists within the STH but not the EDZ. It is shown that variations in small quantities of subtropopause moisture account for much of the WV T-b, discrepancy between these two regions; this moisture is not reported in operational rawinsonde data or detected in OLR. An upper-layer moisture boundary is defined linking TOVS and rawinsonde observations. This moisture boundary is the pressure level reached when moisture is integrated downward from 300 mb to a PW content of 0.6 mm. Operational rawinsonde limitations require that 300 mb be used for the top of our model atmosphere. Correlations imply that 70% of the upper-boundary elevation pressure (elevation) derived from rawinsonde data can be explained by WV T-b variations. However, large standard deviations up to 55 mb are still present in boundary pressure estimated from WV radiances. Calculations from a radiative transfer model suggest that trace moisture between 100 and 300 mb, not normally available in operational soundings, causes the discrepancy. Analysis of special ship-launched rawinsondes confirm this hypothesis and demonstrate that up to 50% of the TOVS WV weighting functions can be attributed to tropospheric moisture above 300 mb; associated PW amounts may be as little as 0.2 mm. Accounting for this moisture could reduce errors in moisture boundary locations, estimated from TOVS, to less than 10 mb.