JUNE 2004 VANCE ET AL. 921 Comparison of In Situ Humidity Data from Aircraft, Dropsonde, and Radiosonde A. K. VANCE,J.P.TAYLOR,T.J.HEWISON, AND J. ELMS Met Of®ce, Exeter, United Kingdom (Manuscript received 14 January 2003, in ®nal form 3 November 2003) ABSTRACT Results are presented from the Measurement of Tropospheric Humidity (MOTH) Tropic and MOTH Arctic airborne ®eld experiments, comparing a number of in situ humidity measurements. Good agreement is shown between the Total Water Content probe on board the C-130 aircraft, and the Vaisala RS90 and ``new'' Vaisala RS80 radiosondes; ``old'' Vaisala RS80 radiosondes and Vaisala RD93 dropsondes show the dry bias noted by others. An empirical correction for RD93 dry bias is presented and is shown to produce good results with both MOTH and non-MOTH data. It was concluded that the aircraft and corrected dropsonde data agree (1s)to61 gkg21; these limits are due to atmospheric variability. The possibility of temperature measurement errors producing errors in RD93 relative humidities is not signi®cant compared to atmospheric variability. Meteolabor Snow White radiosondes are shown to exhibit a wet bias at high and low mixing ratios and possible reasons are discussed. Intercomparisons between the RS90s and other instruments, partitioned by day±night and by experiment, suggest de®ciencies in RS90 daytime radiation corrections. 1. Introduction forded the opportunity to intercompare ®ve hygrometers in both tropical and subarctic conditions. Water vapor is well established as the most important The value of data from radiosondes and dropsondes greenhouse gas and hence is crucially important in de- is well known and, despite the ongoing developments termining the radiation budget of the atmosphere. Its in satellite meteorology, data from radiosondes are, and vertical distribution has a signi®cant impact on local will continue to be, relied on for routine observations, radiative heating and cooling, and on the net ¯uxes at research, and satellite validation work, despite short- the surface and the top of the atmosphere. Its three- comings in these data (Guichard et al. 2000; Nash et al. dimensional distribution has a large impact on the dy- 2003; Wang et al. 2002; Weckwerth et al. 1999). An- namics and thermodynamics of the atmosphere and also derson et al. (2000) and Baker and Eskridge (2000) have controls the distribution of clouds. A thorough under- investigated the effect on numerical weather prediction standing of the distribution of water vapor is therefore of reducing the number of radiosonde observations and important to our monitoring and understanding of cli- found that this causes a signi®cant reduction in the qual- mate change, to our ability to interpret data from future ity of forecast produced. Other researchers (Szunyogh satellite instruments, and to operational numerical et al. 1999, 2000, and references therein; Amstrup and weather prediction processes. Huang 1999; Cardinali 2000) have reported signi®cant In 1999 the U.K. Met Of®ce carried out two Mea- improvements in numerical weather prediction during surement of Tropospheric Humidity (MOTH) ®eld ex- the Fronts and Atlantic Storm Tracks Experiment (FAS- periments, MOTH Tropic and MOTH Arctic, to inves- TEX), Tropical Cyclones 98, the North Paci®c Exper- tigate relevant radiative transfer issues using a micro- iment (NORPEX), and the Winter Storms Reconnais- wave radiometer, the Microwave Airborne Radiometer sance Program (WRS99) resulting from the inclusion of Scanning System (MARSS; McGrath and Hewison data from extra dropsondes and radiosondes launched 2001), and a mid-infrared interferometer, the Airborne in key locations during these experiments. Research Interferometer Evaluation System (ARIES; Differences exist between various sonde types cur- Wilson et al. 1999), ®tted to the Meteorological Re- rently in use, and accurate measurements of upper-air search Flight (MRF) C-130 aircraft. The detailed in situ humidity (especially at low temperatures) are lacking humidity measurements required for this exercise af- (Elliot and Gaffen 1991; Garand et al. 1992; Leiterer et al. 1997; GeÂrard and Saunders 1999). Jaubert et al. (1999) also note differences between sensors and a lack Corresponding author address: A. K. Vance, Met Of®ce, FitzRoy of opportunity to compare dropsondes and radiosondes Road, Exeter, Devon EX1 3PB, United Kingdom. during FASTEX. E-mail: alan.vance@metof®ce.com In this paper we present the results of this intercom- 922 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 21 parison (section 2) and propose a simple empirical structure''). It is possible neither to ``chase'' a drop- scheme for the correction of dry bias seen in the drop- sonde down with the aircraft (as such a spiral descent sondes used during the MOTH experiments (section 3). would result in unacceptable operating conditions for the TWC) nor to chase an ascending balloon. In many cases it will be obvious from inspection if two vertical 2. Comparison of in situ humidity data pro®les were measured in air with the same or different a. Summary of instruments compared humidity structures. Confusion may, however, arise in two ways that are not immediately obvious: 1) if the The Total Water Content (TWC) probe is a Lyman- humidity structures are actually different, but differ- a absorption hygrometer developed by the Met Of®ce ences in the measuring instruments are such that they for use on its C-130 aircraft. Its operating range, ac- (at least partially) offset otherwise obvious discrepan- curacy, and resolution are quoted as 0±20, 60.15, and cies, and 2) if the humidity structures are identical but 21 60.005 g kg , respectively (Nicholls et al. 1990). The not perceived as such because of differences between TWC is calibrated post¯ight, using data from an on- the measuring instruments. board General Eastern 1011B chilled-mirror hygrometer The following data selection strategy was adopted for (StroÈm et al. 1994). Note that although this instrument the purposes of this instrument intercomparison. Any will measure the combined amount of water in all phas- potential intercomparisons between instruments were es, it may be regarded as a hygrometer in this case rejected if their start or launch times were separated by because the data considered were gathered only in 2 h or more. Aircraft pro®les were recorded during as- cloud-free air. cents or descents along one or more straight ¯ight paths. The Vaisala RD93 (revision B) is a dropsonde devel- The furthest extremities of the longest of these pro®les oped by the National Center for Atmospheric Research (NCAR) in collaboration with Deutschen Zentrum fuÈr (in the horizontal plane) in each of the experiments were Luft- und Raumfahrt (DLR) and manufactured by Vaisala used to de®ne the maximum acceptable displacement of Oyj under license. Humidity measurement is by means any two measurements; this was 140 km in the case of of a pair of unheated ``H-Humicap'' capacitative sensors. MOTH Tropic, 230 km for MOTH Arctic. These criteria Relative humidity resolution is quoted as 1% and the do not in themselves guarantee that the two measure- uncertainty in soundings as 5% (Vaisala 2002a). ments being compared will be of the same humidity The Vaisala RS90 is a balloon-borne radiosonde hav- structure, and the situation will still arise where the ing a pair of alternately heated H-Humicap sensors. The humidity structure at one or more levels in a pro®le speci®cations of these are the same as for the RD93 differs signi®cantly between the two instruments due to (Vaisala 2002b). atmospheric effects rather than instrumental ones; these The Vaisala RS80-H is a balloon-borne radiosonde occurrences were handled as part of the comparison with a single unheated H-Humicap. Range and resolution process (section 2c). are the same as for the RS90 but the repeatability and In all cases Snow Whites were ¯own on the same reproducibility are quoted as 2% and ,3%, respectively balloon as an RS90, thus guaranteeing measurements of (Vaisala 2002c). RS80 radiosondes from two different the same humidity structure. During MOTH Tropic, calibration batches were used during MOTH Tropic. RS80s were also ¯own with RS90s, but during MOTH ``Old'' shall be used to refer to those calibrated in March Arctic the RS80s were ¯own separately. In the case of 1997 and ``new'' to those calibrated in January 1999. such dual ¯ights, unless there was evidence of one of The Meteolabor Snow White is a low-cost chilled- the sondes malfunctioning in a manner that would have mirror dewpoint±frost point hygrometer designed for been apparent had it been ¯own solo, the entire pro®le radiosonde use. As the Snow Whites only report dew- was used without question. or frost point, in the MOTH experiments RS90 tem- Table 1 shows the numbers of comparisons with the perature measurements were used to calculate mass mix- Vaisala RS90. The choice of the RS90 as a standard for ing ratios. Relative humidity (dependent on ambient the purposes of this study should not be taken to imply temperature) and mirror temperature ranges are quoted any a priori assumption of greater accuracy on the part as 2%±100% and 2808 to 1408C, respectively (Me- of the authors. Also shown are the number of compar- teolabor 2001); accuracy of mirror temperatures is not isons of the TWC with RD93 dropsondes. As a result stated but is believed to be 0.1 K when used with the of the application of these data selection criteria, a sig- RS90 (R. Maag 2003, personal communication). ni®cant quantity of data from the MOTH Arctic exper- iment, where the weather was more variable, was re- jected. Data from a subsequent ¯ight (number A814, b. Selection of data near Scarborough, United Kingdom) have been included The greatest potential source of error in these inter- in this intercomparison as it included speci®c intercom- comparisons is likely to arise as a result simply of the parison of the aircraft and dropsondes, and data recorded various instruments sampling air with a different water at very low mixing ratios and relatively low tempera- vapor mass mixing ratio structure (hereafter, ``humidity tures, were included.