Advancements, measurement uncertainties, and recent comparisons of the NOAA frost point hygrometer

Abstract

The NOAA frost point hygrometer (FPH) is a balloon-borne instrument flown monthly at three sites to measure water vapor profiles up to 28 km. The FPH record from Boulder, Colorado, is the longest continuous stratospheric water vapor record. The instrument has an uncertainty in the stratosphere that is < 6 % and up to 12 % in the troposphere. A digital microcontroller version of the instrument improved upon the older versions in 2008 with sunlight filtering, better frost control, and resistance to radio frequency interference (RFI). A new thermistor calibration technique was implemented in 2014, decreasing the uncertainty in the thermistor calibration fit to less than 0.01 °C over the full range of frost - or dew point temperatures (93 to +20 °C) measured during a profile. Results from multiple water vapor intercomparisons are presented, including the excellent agreement between the NOAA FPH and the direct tunable diode laser absorption spectrometer (dTDLAS) MC-PicT-1.4 during AquaVIT-2 chamber experiments over 6 days that provides confidence in the accuracy of the FPH measurements. Dual instrument flights with two FPHs or an FPH and a cryogenic frost point hygrometer (CFH) also show good agreement when launched on the same balloon. The results from these comparisons demonstrate the high level of accuracy of the NOAA FPH.

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/ doi:10.5194/amt-9-4295-2016 Author(s) 2016. CC Attribution 3.0 License.

Emrys G. Hall1,2, Allen F. Jordan1,2, Dale F. Hurst1,2, Samuel J. Oltmans2, Holger Vmel3, Benjamin Khnreich4,5, and Volker Ebert4,5

1Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA

2NOAA Earth System Research Laboratory, Global Monitoring Division, Boulder, Colorado, USA

3National Center for Atmospheric Research, Earth Observation Laboratory, Boulder, Colorado, USA

4Physikalisch-Technische Bundesanstalt, Braunschweig, Germany

5Center of Smart Interfaces, Technische Universitt Darmstadt, Darmstadt, Germany Correspondence to: Emrys G. Hall ([email protected])

Received: 11 May 2016 Published in Atmos. Meas. Tech. Discuss.: 17 May 2016 Revised: 26 July 2016 Accepted: 16 August 2016 Published: 5 September 2016

Abstract. The NOAA frost point hygrometer (FPH) is a balloon-borne instrument own monthly at three sites to measure water vapor proles up to 28 km. The FPH record from Boulder, Colorado, is the longest continuous stratospheric water vapor record. The instrument has an uncertainty in the stratosphere that is < 6 % and up to 12 % in the troposphere. A digital microcontroller version of the instrument improved upon the older versions in 2008 with sunlight ltering, better frost control, and resistance to radio frequency interference (RFI). A new thermistor calibration technique was implemented in 2014, decreasing the uncertainty in the thermistor calibration t to less than 0.01 C over the full range of frost or dew point temperatures (93

to +20 C) measured during a prole. Results from multi

ple water vapor intercomparisons are presented, including the excellent agreement between the NOAA FPH and the direct tunable diode laser absorption spectrometer (dTDLAS) MC-PicT-1.4 during AquaVIT-2 chamber experiments over 6 days that provides condence in the accuracy of the FPH measurements. Dual instrument ights with two FPHs or an FPH and a cryogenic frost point hygrometer (CFH) also show good agreement when launched on the same balloon. The results from these comparisons demonstrate the high level of accuracy of the NOAA FPH.

Advancements, measurement uncertainties, and recent comparisons of the NOAA frost point hygrometer

1 Introduction

Water vapor is the most abundant and important greenhouse gas in the atmosphere and contributes to many processes and feedback mechanisms (Dessler et al., 2008). Though the vast majority of this highly variable invisible gas is found near the surface, water vapor in the upper troposphere and stratosphere can signicantly inuence climate (Held and Soden, 2000).

The concentration of water vapor in the stratosphere is controlled by transport through the tropical tropopause layer (TTL) rendering the stratospheric air extremely dry through a freeze-drying process (Brewer, 1949). The seasonal cycle of the tropical tropopause temperature controls the seasonal cycle of water vapor entering the stratosphere which is advected upward and poleward (Mote et al., 1996). Methane oxidation is a photochemical source of water vapor in the atmosphere but can only account for 2534 % of the net stratospheric water vapor increase between 1978 and 2003 (Rohs et al., 2006). Water vapor plays a role in cirrus cloud formation, which impacts Earths radiation budget. High thin cirrus clouds may cool the surface while producing a net warming in the stratosphere by absorbing outgoing longwave radiation in the atmosphere (Lee et al., 2009).

The Boulder stratospheric water vapor record shows a net increase of 1 ppmv (27 %) since 1980 (Hurst et al., 2011a)

with a sharp decrease of roughly 10 % in 2001 to 2005 (Randel et al., 2006; Rosenlof and Reid, 2008). Solomon et

Published by Copernicus Publications on behalf of the European Geosciences Union.

In situ balloon ight proles complement satellite sensors monitoring the global distribution of water vapor around the planet. Monthly NOAA FPH balloon measurements provide high-vertical-resolution proles. Satellite sensors, such as the Aura Microwave Limb Sounder (MLS) (Read et al., 2007), measure water vapor in the UTLS and above (Lambert et al., 2007) with almost daily global frequency but with a vertical resolution of 23 km. Combining the data from these two measurement techniques provides important distribution information of water vapor for models, to help detect trends, and identify future changes in climate. Differences between the NOAA FPH and the MLS have been explored in depth by Hurst et al. (2014), showing good agreement in the stratosphere (6826 hPa) at all three NOAA Global Monitoring Division (GMD) sites between August 2004 and December 2012. Previous coincident cryogenic frost point hygrometer (CFH) balloon observations were used in the validation of the earlier versions of the MLS satellite (v1.5 and v2.2) between 2005 and 2007 (Vmel et al., 2007c). Vmel et al. (2007c) found that MLS and CFH proles agreed to within 2.7 8.7 % from 68 to 22 hPa. The CFH is a small

balloon-borne hygrometer developed in 2003 which uses the same fundamental measurement principle as the NOAA FPH (Vmel et al., 2007a). The two instruments have many similarities but also differ in subtle ways including electronics, sunlight ltering, and frost control algorithm.

This paper provides detailed information regarding the description and uncertainty of the current digital FPH along with a brief history of the evolution of the sonde. Comparisons in the laboratory along with dual instrument balloon ights are presented.

2 Instrument description and history

The NOAA FPH is an in situ balloon-borne chilled mirror hygrometer capable of measuring vertical proles of frost point temperature up to 28 km. The basic mea

surement principle and calibration method have remained unchanged throughout the time series although signicant modernization of the instrument occurred over the ongoing 36 year record (Mastenbrook and Oltmans, 1983; Oltmans, 1985; Vmel et al., 1995; Oltmans and Hofmann, 1995;Hurst et al., 2011a). The chilled mirror principle relies on maintaining a thin, stable layer of condensate on a mirror disk through rapid feedback control. A copper cold nger immersed in a dewar containing liquid cryogen provides cooling power throughout the prole (Fig. 1). A polished mirror disk resides at the opposite end of the cold nger with ambient air passing over it at 36 m s1. A nichrome heater wrapped around the narrow shaft of the continuous cold nger and mirror piece provides heat to the mirror. An optical source and detector, composed of an infrared light-emitting diode (LED) and a photodiode, is used to monitor the mirrors reectivity as condensate accumulates in the form of

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

4296 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

al. (2010) estimated that this sharp drop in stratospheric water vapor in 2001 decelerated the global surface temperature increase by approximately 25 % compared to what would have been caused by well-mixed greenhouse gases between 2000 and 2009. It is imperative to monitor the distribution of water vapor, especially in the stratosphere where it represents an important driver of decadal worldwide surface climate change (Solomon et al., 2010).

Accurate measurements of upper tropospheric and lower stratospheric (UTLS) water vapor are pivotal for understanding changes in stratospheric water vapor and their impact on the radiative forcing (Forster and Shine, 2002) of our climate. Mller et al. (2016) argue for a large network of frost point hygrometers (FPHs) with global coverage spanning many decades because of the climatic importance of water vapor. Worldwide there are more than 800 radiosonde stations launching balloon instruments twice daily, totaling > 500 000 vertical proles of wind speed/direction, pressure, temperature, and humidity each year. These measurements are predominately used for weather forecasting. While radiosonde wind, pressure, and temperature measurements are accurate up to balloon burst (> 30 km), the relative humidity measurements often lack the precision and accuracy necessary for climate research and are most useful in the low to mid-troposphere. Stratospheric water vapor is deceptively hard to measure, despite the relatively high concentration (> 1 ppmv) compared to most trace gases in the atmosphere.Vertical proles of UTLS water vapor have been measured by satellite, aircraft, lidar, and balloon-borne instruments, revealing considerable disagreements ( 25100 %) between a

core group of stratospheric water vapor measurements (Kley et al., 2000; Weinstock et al., 2009). A 2011 intercomparison of two frost point hygrometers and ve in situ aircraft hygrometers show reduced but still statistically signicant differences of 0.40.8 ppmv (1020 %) (Rollins et al., 2014).Corroborating these ndings, Meyer et al. (2015) show that agreement between a core group of hygrometers has improved over the past 2 decades from 30 % or more to ap

proximately 520 % for mixing ratios under 10 ppmv. Fu

ture laboratory and airborne intercomparisons are necessary to further understand the remaining differences while helping to identify errors in calibration and measurement techniques.

The NOAA FPH captures monthly water vapor proles from three sites around the globe. The longest continuous record of upper atmospheric water vapor abundance began in Boulder, Colorado, in 1980. The second site was Lauder, New Zealand, which started monthly soundings in 2004.Hilo, Hawaii, initiated soundings in late 2010. These three sites were chosen to represent the northern and southern mid-latitudes and the tropics, respectively. The typical instrument package own at these three sites consists of a NOAA FPH, an electrochemical concentration cell (ECC) ozonesonde, and a radiosonde which wirelessly transmits the data back to the receiving station on the ground.

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4297

Table 1. The four versions of the NOAA FPH own between the present day and 1980 along with the radiosondes, weights, and data acquisition method.

FPH version Period Radiosonde frequency and model Data acquisition method Weight without cryogen (g)

FPH V1 19801991 1680 MHz VIZ A Analog strip chart recorder 1550 FPH V2 19912004 403 MHz Vaisala RS-80 Digital Strato software 1500 FPH V3 20042008 403 MHz Vaisala RS-80 Digital Strato software 475 FPH V4 2008present 403 MHz InterMet iMet-1-RSB Digital SkySonde Client/Server 450

Figure 1. Schematic of the NOAA FPH instrument. The electronics board and optics block are enclosed by insulating foam along with the battery pack while the lens, intake tubes, and mirror head are all exposed to ambient air. The microcontroller uses the photodiode signal to regulate the mirror temperature such that the reectivity of the frost on the mirror is constant over the ight.

dew or frost. A biconvex lens focuses the light returned off the mirror into the photodiode. A calibrated thermistor embedded in the mirror accurately measures the frost point temperature. The frost point temperature is achieved when a stable frost layer is present, showing equilibrium between water vapor in the air sample and the condensate on the mirror. Due to the material and geometry, the surface of the mirror is considered to be uniform in temperature across the 7.1 mm diameter (Mastenbrook and Oltmans, 1983). The GoffGratch formulation of the ClausiusClapeyron equation (Goff and Gratch, 1946; Goff, 1957) is used to calculate the water vapor partial pressure with respect to ice. Dividing the water vapor partial pressure by the dry atmospheric pressure determines the mixing ratio by volume. The chilled mirror principle is unique in that only the thermistor needs to be calibrated against National Institute of Standards and Technology (NIST) traceable temperature standards. This eliminates water vapor calibration scales or standards which are difcult to create, maintain, and use in the eld.

The NOAA FPH is own as a disposable instrument that can be reused if returned. The typical instrument package,

including an ozonesonde, radiosonde, and liquid cryogen, is lightweight (< 1.9 kg) and typically own using a 1200 g latex balloon. Triuoromethane (CHF3, R23) has been used as the liquid cryogen since the mid-1990s when the switch was made away from CClF3 (R13). Both compounds have ideal physical properties over a wide range of temperatures (100 to 30 C) and pressure (10 to 1050 hPa) to provide liq

uid cooling of the cold nger in all conditions encountered during balloon ights.

The ight train consists of a balloon with a valve in its neck and a parachute attached to a long string unwinder (30 50 m). Nearly all NOAA FPH instrument packages continue to be own with a valve in the neck of the balloon, allowing slow, controlled descent proles to be acquired (Masten-brook, 1966; Kruchi et al., 2016). The instrument package resides at the bottom of the unwinder that separates the balloon and parachute from the instrumentation in an effort to reduce water vapor contamination from balloon outgassing during the ascent.

Over the 36 years of FPH measurements by NOAA, four versions of the instrument have been own with each change reecting advancements in the capabilities of the instrument (Table 1). However, each version adhered to the same NIST-traceable temperature calibration standard.

2.1 FPH V1: analog hygrometer

Balloon-borne frost point hygrometers have been used to collect vertical proles of water vapor abundance and distribution in the atmosphere since the late 1950s (Mastenbrook and Dinger, 1961). Monthly proles were acquired in Washington, DC between 1964 and 1980 using a 4.5 kg balloon-borne frost point hygrometer specically developed for the program. The water vapor program was moved to Boulder, Colorado, in 1980 following signicant instrument improvements that included implementation of solid-state electronics (Mastenbrook and Oltmans, 1983). The components from a 1680 MHz VIZ A radiosonde were integrated inside the FPH electronics package, excluding the carbon hygristor humidity sensor, which was intentionally left off (Fig. 2a). This included the specially designed baroswitch and electronics of the radiosonde with the thermistor protruding on a stick off to the side. The baroswitch controlled hygrometer gain changes during the ight and served as the main pressure source for this version of the FPH. The data were teleme-

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

4298 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

Figure 2. The NOAA FPH V1 hygrometer integrated the baroswitch and electronics from a VIZ A radiosonde inside the FPH instrument package with the thermistor protruding on a stick off to the side (a). The VIZ A 1680 MHz transmitter (white cone located inside protective loop) was own 3 m above the instrument to avoid RFI problems. The sun shield for V1 hygrometers utilized a ping pong ball surrounded by a stainless steel shroud to block unwanted sunlight. A 60 cm foam boom separated the Vaisala RS-80 radiosonde and the FPH V2 hygrometer own without an ECC ozonesonde (b). A Vaisala RS-80 radiosonde and an ECC ozonesonde were attached to the FPH V3 instrument (c). Both V2 and V3 instruments were own with stainless steel sun shields. The NOAA FPH V4 is currently own with an InterMet iMet-1-RSB radiosonde and an ECC ozonesonde (d).

tered wirelessly and recorded on a paper strip chart. The VIZ A radiosonde transmitter was own 3 m above the rest of the instrument package to avoid radio frequency interference (RFI) (Mastenbrook, 1981).

2.2 FPH V2: digital interface to Vaisala RS-80 radiosonde

In February 1991, a digital interface was added utilizing a 403 MHz Vaisala RS-80 radiosonde and a separate 8 bit analog-to-digital converter (ADC) board which was replaced in October 1991 by a 12 bit ADC board. No shifts were detected in the data recorded digitally instead of on the strip chart recorder formerly used with the 1680 MHz VIZ A radiosonde (Oltmans et al., 2000). Starting in 1993, several steps were taken to minimize problematic 403 MHz RFI, including adding a 60 cm horizontal foam boom to separate the radiosonde from the other instrumentation and adding ferrite beads to any wires on the FPH (Fig. 2b). Silver mirrors used before 1977 were replaced with rhodium-plated copper mirrors on the Boulder instrument until nickel- and gold-plated copper mirrors became the standard late in 1992. In 1998 the custom-made aneroid barometer incorporated on the FPH became obsolete, forcing its replacement by a solid-state sen-

sor in the new supporting electronics designed in 2000 and improved upon in 2001. This sensor was only used for gain changes as the Vaisala RS-80 radiosonde measured pressure for all calculations for FPH V2 and V3 hygrometers.

2.3 FPH V3: lightweight hygrometer with upgraded electronics

In 2004, a modern electronics board was added utilizing surface mount components that eliminated RFI from the attached Vaisala RS-80 radiosonde. Modied mechanical and optical parts were combined with new foam packaging to create an analog FPH boasting a 70 % reduction in weight (Fig. 2c). These improvements enabled the second NOAA/GMD stratospheric water vapor station to be started at Lauder, New Zealand, in 2004.

Prior to the 2004 version of the hygrometer the optical system consisted of two pairs of optical elements in close thermal proximity. The specular pair looked down at the mirror through a lens and provided a current proportional to the light reected off the mirror while the bias pair looked directly at each other. The two pairs were matched according to their temperature coefcients and attached to the optics block with thermal epoxy to minimize temperature-driven signal drift in the differential measurement. Before ight, the bias signal was set at 21 % by adjusting a setscrew to match the specular signal setpoint. With the introduction of temperature stabilization of the aluminum optics block the bias pair was deemed unnecessary on all FPH instruments own after 2004, allowing a xed resistor to replace the bias pair on the electronics boards.

2.4 FPH V4: digital microcontroller hygrometer

Between 2005 and 2008, a digital version of the hygrometer was developed that offered greater precision and ease of use, while decreasing the weight, cost, power consumption, and manufacturing time. The skill required to successfully launch and collect water vapor data with the FPH V4 is equivalent to that needed to prepare and release an ozonesonde with an attached radiosonde, of which there are approximately 3000 own each year by different research groups around the world. The older analog instrument required more in-depth knowledge and time to successfully produce a high-quality water vapor prole.

A major advancement to the digital FPH was the introduction of the modulating lter that removes unwanted sunlight from the photodiode signal during a daytime balloon ight, similar to the CFHs earlier implementation (Vmel et al., 2007a). Earlier versions of the FPH utilized a stainless steel sun shield to block the sunlight except for ights done between 1997 and 2003. The sun shield was effective at blocking sunlight but often contributed to stratospheric contamination on the ascent portion of the ight. The new switching sunlight lter provides clean stratospheric frost control by

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4299

reducing contamination. Since 2010, the FPH LED is modulated at 24 Hz while the photodiode continually measures the signal from the mirror. The microcontroller calculates the difference between the LED on and off signals in real time during the ight. This enables the sunlight captured by the photodiode during the LED off periods to be removed from the LED on signal to maintain unvarying condensate coverage during a ight.

Eliminating the sun shield, incorporating a lens heater, and implementing a exible frost point-dependent gain schedule (Vmel et al., 2007a) were signicant improvements linked with the digital FPH. A digital calibrated pressure sensor was incorporated into the hygrometer, adding the exibility to perform tasks throughout the ight based on pressure and also served as a backup source when the radiosonde had issues.

Prior to 2008 the LED in the analog version of the FPH had a wide viewing angle and a peak wavelength centered at 940 nm. This LED and the photodiode were replaced with an 880 nm peak LED and a 900 nm centered photodiode, both with narrow viewing angles. Peak wavelength matching is not necessary because the spectral bandwidth at half power is 40 nm, providing ample overlap. The optics block is tem

perature controlled to 32.00 0.01 C throughout the bal

loon ight. Great care is taken to maintain temperature stability of the photodiode and LED because both are highly temperature sensitive. Since 2009, a lens heater similar to that own on the CFH (Vmel et al., 2007a) was implemented to help reduce condensation, which may form on the lens in humid conditions.

The FPH is insulated by several pieces of foam, with a 300 mL foam Dewar on one side of the instrument and a foam battery enclosure housing the optics block, electronics board, and battery pack on the opposite side. Very thin, hydrophobic stainless steel inlet tubes with a diameter of 2.25 cm are glued above and below the sensor housing to deliver clean air across the polished mirror disk in the middle of the sensor housing (Fig. 2d).

2.4.1 Mirror feedback controller, electronics, and battery power

The FPH uses a custom-made electronics board with an on-board pressure sensor and an 8 bit Atmel AVR microcontroller. The electronics remain between 20 and 30 C during a ight by capturing heat from the nearby optics block and batteries. The FPH is powered by a 9 V battery pack with3.5 Ah capacity. A nichrome heater wrapped around the mirror allows for a maximum heating power of 22 W, more than sufcient to measure dew or frost point temperatures anywhere on Earth when R23 cryogen is used for cooling.

A 16 bit ADC converter, oversampled to 19 bit, provides continuous mirror reectivity feedback to the microcontroller from the photodiode. A dynamic schedule of proportionalintegralderivative (PID) gain values, based on

Figure 3. The frost point temperature measured by the NOAA FPH mirror thermistor (blue) remains unchanged when varying the amount of condensate on the mirror by adjusting the reectivity set-point (black). The measured reectivity of the mirror is shown in red. Ambient laboratory air was sampled during this experiment.

a slow moving average of the measured frost point temperature, enhances the ability of the FPH to control the frost layer over an extremely wide range (> 105) of water vapor number densities in the atmosphere. The modern gain schedule was developed to allow the instrument to be own anywhere in the world using the same PID gain table.

The older analog version of the NOAA FPH utilized a simple proportional controller with one gain change within a prole. It was not uncommon for the analog instrument to experience frost control oscillations in the rst few kilometers of the prole and directly after the gain change that historically occurred in the middle troposphere. The digital instrument, with the dynamic gain schedule, provides nearly uninterrupted water vapor prole measurements that are used to determine integrated precipitable water columns.

Under stable conditions the FPH controls dew or ice on the mirror such that 79.1 % of the light that would reect off a bare mirror reaches the photodiode. This value has remained constant since 1980. Laboratory testing has shown that increasing the setpoint value by > 7 % does not affect the frost point temperature measured by the instrument (Fig. 3).Maintaining a constant condensate layer is much more important than the actual setpoint value used.

2.4.2 Radiosonde telemetry and data collection

The FPH has been own with the InterMet Systems iMet-1-RSB radiosonde since summer 2009 because the Vaisala RS-80 radiosonde used previously became obsolete. This radiosonde provides measurements of pressure, temperature, and humidity (PTU) (Hurst et al., 2011b) along with any externally connected instrument data. The iMet-1-RSB sup-

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

4300 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

plies GPS location information that enables payload tracking as well as the calculation of wind speed and direction.

The iMet-1-RSB transmits 1 Hz wireless data at 1200 baud allowing up to 120 bytes of data to be sent each second. The iMet-1-RSB radiosonde employs an open-source telemetry protocol (XDATA) that allows multiple instruments to be connected via serial daisy chain. Roughly 80 bytes of externally connected instrument data can be sent down along with the PTU and GPS data of the radiosonde. The data are recorded on the ground by a custom-made software program SkySonde that was designed and written by NOAA/GMD engineers to receive and demodulate telemetered data from the iMet-1-RSB.

3 Thermistor calibration

Although the equipment used to calibrate thermistors has changed over the years, the thermistor calibration method has remained unchanged, providing greater stability in the absolute calibration of FPHs over the decades. Details regarding the different data acquisition systems, curve tting routines, and NIST-traceable thermometers are discussed below.

The frost point temperature is measured by a bead thermistor embedded in the mirror disk. Each thermistor is individually calibrated in small batches in a 4 L insulated dewar containing pure ethanol. A laboratory stirrer uniformly maintains the alcohol bath temperature during the calibration. The bath temperature is measured with a NIST-traceable thermometer. The thermistors are inserted approximately 16 cm into the alcohol bath and are attached to a digital multimeter outside of the dewar. Between 1980 and 2013, nely ground dry ice chips were manually added to the bath to cool and maintain the alcohol temperature at three targets: 0, 45, and 79 C. Great care was taken while adding small amounts of dry ice to maintain a 0.005 C tolerance around each of the

three setpoints to minimize curve tting errors. Calibration coefcients were calculated from the measured resistances at the three target temperatures and used in an equation to calculate mirror temperature during a ight (Clark, 1961). Starting in 2014, calibrations were done by continuously measuring the bath temperature and the resistance of each thermistor as the alcohol bath slowly warmed from 93 to +19 C. The

new continuous resistance measurement technique utilizes a six-point t instead of the historic three-point t eliminating the extrapolation curve tting correction necessary for the older three-point t (see Sect. 3.1 below).

A NIST-traceable Hewlett-Packard quartz thermometer was employed as the temperature reference between 1980 and 1998. This probe had accuracies of 0.075 C between 80 and 50 C and 0.040 C between 50 and +150 C

with a resolution of 0.001 C over the full range. When the quartz thermometer broke in 1998 a platinum resistance thermometer (PRT, Hart Scientic, USA) became the temperature reference for all calibrations after 1998. This probe

Figure 4. Difference between two different calibration ts and the actual measured bath temperature during a calibration run. Each curve is comprised of the mean of 39 thermistors t errors. The SteinhartHart three-point t is shown with (blue) and without (red) the low temperature correction. All ights between 1980 and 2013 utilize the three-point t with the low temperature correction. Flights starting in 2014 do not need a low temperature correction since the new six-point thermistor t (green) was implemented. Dashed lines show the six calibration points.

has a stated accuracy of 0.01 C with 0.001 C resolution

over the full range of NOAA FPH thermistor calibrations.

The Hart PRT probe has been recalibrated by the manufacturer every 6 years since 1998 and still serves as the NIST-traceable standard today.

Until late 1998 the resistance of each thermistor was hand recorded from an HP digital multimeter. Then a new data acquisition system was developed using a Campbell Scientic CR10 data logger that allowed 20 thermistors to be calibrated at once. During the summer of 2004 an Agilent 34970A data acquisition switching unit was set up as the new primary calibration system, allowing 40 thermistors to be calibrated at a time.

3.1 Three-point thermistor extrapolation correction below 79 C frost point temperature: 19802013

The traditional three-point thermistor curve t was utilized for thermistor calibrations performed from 1980 to 2013. Between 0 and 79 C the curve t was better than 0.06 C

(Fig. 4) that translates to at most a 0.51.1 % difference in water vapor mixing ratio over the entire ight (+20 to 93 C) or in terms of absolute differences 0.06 ppmv when analyzing stratospheric data. Colder than 79 C there ex

isted a one-sided extrapolation error with a maximum error of 0.16 C at 90 C equivalent to an absolute difference of

0.16 ppmv or < 3 % in water vapor mixing ratio. Based on a batch of six thermistors calibrated down to 93.3 C, a lin

ear correction Eq. (1) was applied to frost point temperatures colder than 79 C for all NOAA FPH ights prior to 1991

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

and 79 C, the bath is initially cooled to 93 C with liquid

nitrogen and allowed to slowly warm (Vmel et al., 2007a), eliminating the need to extrapolate below 79 C. An ex

panded version of the SteinhartHart formulation now derives fth-order polynomial ts of thermistor resistances to mirror temperature (Steinhart and Hart, 1968) at six calibration temperatures: 91, 80, 59, 39, 15, and +19 C.

These temperatures were selected to minimize t errors, with greater weighting for the important stratospheric temperature range. The NIST-traceable thermometer temperatures and the resistance values for all 40 channels are recorded continuously on an Agilent 34970A digital multimeter until the bath reached > 19 C, a process that takes 50 h because the de

war is well insulated. The probe temperature is measured before and after the two second sweep (< 0.003 C temperature change during each sweep) of each thermistor and then interpolated to determine the probe temperature during each of the 40 thermistor measurements. The new six-point t reduces errors over the full range to better than 0.01 C com

pared to 0.06 C for the historical three-point t (Fig. 4).

The new six-point t converts to at most a 0.080.2 % difference in water vapor mixing ratio over the entire ight or 0.01 ppmv, in terms of absolute differences, when looking specically in the stratosphere. The discontinuity near

70 C is caused by the custom data acquisition software switching between the 1 M[Omega1] and the 100 k[Omega1] range setting on the Agilent 34970A digital multimeter in order to measure the lower thermistor resistances as the bath warms. The discontinuity does not affect the balloon ight data as there are no calibration points near 70 C and the mirror thermis

tor is measured with the onboard ADC, which does not have this issue.

Though far more comprehensive, the new continuous-measurement calibration technique has a complication due to the changing temperature of the bath as it warms. The PRT probe has a larger thermal mass than the thermistors and therefore warms at a slower rate. At each sampling time, the measured probe temperature will be slightly colder than the true temperature of the thermistors. To account for this thermal lag a small correction is required.

The thermal lag correction was determined by performing both a historical three-point calibration and a new continuous-measurement calibration on the same set of thermistors. At the three historical static temperature setpoints there is no thermal lag of the PRT probe despite its larger thermal mass by holding the bath in equilibrium for an extended time at each temperature (> 3 min). The results are compared to those from the new continuous-measurement calibration to determine the thermal lag correction for the probe. The correction is dependent on the warming rate of the bath, so an equation is created mapping warming rate to temperature lag correction for use in future calibrations (which may warm at different rates due to varying lab conditions and equipment). Before applying the lag correction,

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4301

(Mastenbrook, 1981):

TFpT,corr = 1.013(TFpT) + 1.046, (1)

where TFpT is in C. Scherer et al. (2008) developed an improved empirical t to correct frost point temperatures below

79 C for all ights between 1991 and the last three-point thermistor calibration in 2013. The correction was based on the 30 thermistor calibration from 100 to 20 C (Vmel et

al., 2007b).

TFpT,corr = TFpT (0.029(TFpT + 79) + 0.083)2 (2)

Both corrections slightly lowered frost point temperatures < 79 C and reduced water vapor mixing ratios < 3 % in

the stratosphere. After the Scherer et al. (2008) correction was applied the calibration curve ts between 1991 and 2013 were better than 0.06 C over the full temperature range.

No corrections were applied to the frost point temperatures > 0 C for any of the FPH data over this period as they would be much smaller than the uncertainties of the FPH measurements in the lower troposphere.

3.2 Thermistor self-heating correction applied to FPH instruments calibrated before 1987

Prior to 1987, all thermistor calibrations were performed using an HP3490A multimeter that applied a known current to each thermistor and determined its resistance by measuring the voltage drop. For resistances < 10 k[Omega1], the meter applies an excitation current of 0.85 mA that is large enough to heat the small FPH thermistors by 1.5 C at the 0 C cali

bration point. Calibrations performed after 1987 used a different multimeter with a 0.1 mA excitation current, reducing the self-heating. The self-heating at the two colder calibration points (45 and 79 C) was minimal because both old

and new multimeters applied smaller excitation currents at those calibration points.

The self-heating problem at the 0 C setpoint was documented during a calibration in 1987. A group of six thermistors was calibrated using the original 0.85 mA excitation current and the lower 0.1 mA setting. When using the three-point calibration t the warm bias at 0 C produces a cold bias of < 0.21 C at frost point temperatures of 90 C (Scherer et

al., 2008). Utilizing these data, Scherer et al. (2008) derived an empirical correction that was applied to FPH frost point temperatures < 79 C for all ights before 1987:

TFpT,corr = TFpT + (0.0203(TFpT + 61.9))2 0.119. (3)

Applying the empirical correction increases water vapor mixing ratios by < 4 % at 90 C. All data published since

2008 have these two corrections applied where applicable.

3.3 Six-point thermistor calibration: 2014 to present

In 2014 a new calibration technique was implemented. Instead of manually maintaining the calibration bath at 0, 45,

4302 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

Table 2. Overview of the sources contributing to the FPH uncertainty budget. Combining the frost stability, manufacturing variability, and thermistor calibration uncertainties in quadrature can be calculated for a xed interval in the FPH prole.

Uncertainty parameter Value R/S1

Frost control stability, u_Frost_Control 0.11.5 K R Temperature uniformity of the mirror, u_Mirror_ Uniform < 0.1 K S FPH electronics board thermistor measurement, u_FPH_ ADC < 0.02 K (stratosphere), < 0.17 K (troposphere) S2 Manufacturing uncertainty total, u_Manuf = (u_Mirror_Uniform

2 +u_FPH_ADC2)1/2 S2 Thermistor calibration t, u_Cal_Fit 0.06 K (three-point t), 0.01 K(six-point t) S Thermistor calibration repeatability, u_Cal_Repeat 0.043 K (three-point t), 0.028 K (six-point t) S Reference thermometer, u_Ref_Therm 0.01 K after 1998, 0.075 K before 1998 S Thermistor calibration total, u_Therm_Cal = (u_ Cal_Fit

2)1/2, S30.1 K < 1990, 0.07 K > 1990 (three-point t), 0.03 K (six-point t)

NOAA FPH uncertainty total, u_FPH = (u_ Frost_ Control

2 +u_Cal_Repeat2+ u_Ref_ Therm

2 + u_Manuf

2 + u_Therm_Cal

2)1/2

1 Random or systematic; 2 dependent on frost point temperature; 3 constant in prole.

Figure 5. Temperature correction applied to raw NIST-traceable PRT thermometer readings to account for the lag between the larger PRT thermometer and the small thermistors in the alcohol bath.

a line is t to the calculated warming rate to avoid inserting extra noise into the corrected probe temperature data (Fig. 5).

4 Measurement uncertainty

Errors and uncertainties are treated as outlined in the Guide to the expression of uncertainty in measurement by the working group 1 of the Joint Committee for Guides in Metrology (JCGM/WG 1, 2008). Uncertainty in the FPH measurements is determined through the evaluation of random and systematic errors. Uncontrolled factors in measurements that cause normally distributed uctuations are characterized as random errors and their uncertainties can be reduced by increasing the number of measurements. An example of a random uncertainty in the FPH is the frost control stability. Alternatively, systematic errors remain regardless of the number of measurements, producing biases in the measurements if not corrected. Even if systematic errors are properly understood and corrected there will be a residual error as no correction is perfect. The uncertainty of the mirror

temperature measurement by the FPH electronics board is an example of a systematic error. All of the FPH uncertainties are summarized in Table 2.

The largest measurement uncertainties during a sounding arise from instabilities in frost control. The magnitudes of these terms are calculated using the standard error of the estimate of frost point temperature over a xed interval(0.25 km) to determine the residual error from the linear t (2 ). The uncertainty is calculated for the balloon ascent and the descent separately. The uncertainty over the 0.25 km interval range is between 0.1 to 1.5 C in the prole, depending on the instrument performance at any given time.

The mirror temperature uniformity and location of the thermistor embedded in the mirror have been demonstrated to induce systematic uncertainties < 0.1 C in the CFH frost point temperatures (Vmel et al., 2007a). Uncertainties for the FPH are the same because it uses the same thermistors as the CFH and has mirrors of the same thickness and diameter.

The uncertainty of the mirror temperature measurement on the FPH circuit board varies with the frost point temperature due to the ADC. These uncertainties, along with the mirror temperature uniformity and location uncertainties, are combined in quadrature to produce the manufacturing uncertainty in each instrument.

Thermistor calibration t, calibration repeatability, and the reference thermometer uncertainty are combined in quadrature to yield total uncertainty estimates for both the three-point and six-point calibration ts. All FPHs since 2014 use the six-point t and have a combined calibration uncertainty of 0.03 C. All prior instruments using the three-point t have a combined uncertainty of 0.1 C before 1990 and 0.07 C after 1990 and before the six-point t.

Two years of FPH data from Boulder, Colorado, are analyzed to show the contributions of the various uncertainty estimates in Fig. 6. Panel a shows the average in frost point temperature (K) with frost control stability uncertainties dominating the total to 15 km with the other two

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4303

Figure 6. Contributions of specic uncertainty terms to the total FPH uncertainties in frost point temperature (a) and water vapor mixing ratio (b). The FPH uncertainties shown are the mean values from 24 ights in Boulder, CO. Radiosonde pressure uncertainties (magenta) are based on InterMets sensor specications that decrease at 8 km, producing the discontinuity.

terms contributing more above 15 km to the top of the ight. The mixing ratio uncertainties shown in Fig. 6b include those computed from the manufacturers specications for radiosonde pressure sensor uncertainties in magenta, and the total with and without the pressure sensor uncertainty (black and dashed gray lines, respectively). Manufacturer-quoted radiosonde pressure sensor uncertainties are used for the iMet-1-RSB (1070400 hPa: 1.8 hPa; 4004 hPa: 0.5 hPa), the Vaisala RS-80 (10803 hPa: 1.0 hPa), and

the VIZ A (10505 hPa: 2.0 hPa) at the 2 accuracy limit

(Stauffer et al., 2014; Richner and Phillips, 1981; Tarasick et al., 2016). The uncertainties below 15 km are < 12 % due to both atmospheric variability and less stable frost control. In the stratosphere total uncertainties are < 6 % (Fig. 6b).

5 Mirror condensate

Condensate, either dew or ice, deposits on the mirror when cryogen is added before the ight. Even at mirror temperatures 0 to 40 C, supercooled liquid water may reside

on the mirror before homogeneous nucleation will spontaneously freeze the liquid. The uncertain phase of the condensate between 0 and 40 C is inherent to all chilled mirror

frost point hygrometers (Fujiwara et al., 2003; Vmel et al., 2007a). Correct identication of the condensate phase is important because the vapor pressures above liquid and ice are very different within this temperature range.

The NOAA FPH compares the relative humidity measured by the attached radiosonde to help determine the phase of the condensate on the mirror. Radiosondes measure atmospheric relative humidity with a thin lm polymer capacitive sensing

Figure 7. The liquid to ice transition on the FPH mirror can be seen at 4.2 km on the ascent. Using the appropriate vapor pressure formulations below and above the point of condensate transition brings the RH values of the FPH and the Vaisala RS-92 into agreement.

surface. This sensor always reports relative humidity with respect to liquid and does not have ambiguity issues like chilled mirror frost point hygrometers. Relative humidity proles are calculated from the FPH mirror temperatures (0 to 40 C)

using the different vapor pressure equations for liquid and ice (Fig. 7), then each is compared to the radiosondes relative humidity prole to identify when the liquid to ice transition occurs on the mirror. When the condensate starts as liquid this transition occurs around 31 C. The transition is not

instantaneous and may exhibit a lag of as much as 2 min during some ights. A sharp dip in the frost point temperature occurs when the condensate changes to ice, further helping to distinguish the correct transition point.

According to Murphy and Koop (2005) metastable forms of ice, such as cubic and amorphous ice, may exist colder than 200 K in the atmosphere. These have a higher vapor pressure than the most stable form, hexagonal ice. At temperatures > 200 K only hexagonal ice is formed on the frost point hygrometer mirror. Below the 200 K threshold cubic ice nucleates rst; this is estimated to have a vapor pressure 3 to 11 % greater than hexagonal ice, depending on which enthalpy measurement is used. The cubic ice eventually transforms to hexagonal ice over a period of minutes to days. If the ice on the mirror is cubic instead of hexagonal the frost point temperatures must bias warm to compensate for the greater equilibrium vapor pressure above cubic ice.This leads to high biases in the partial pressures calculated from the GoffGratch equation and therefore high biases in the water vapor mixing ratios. For this reason, during every balloon ascent when the frost point temperature reaches

53 C, the ice on the mirror is completely sublimed by a heat pulse (mirror clear) followed by rapid frost regrowth within 50 s, as is done during each CFH prole (Vmel et al., 2007a). Hexagonal ice is quickly reformed on the mirror and

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

4304 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

is maintained for the duration of the ight because transformation into metastable ice is not possible under atmospheric conditions. The mirror clear is performed at 53 C (220 K),

providing a large window to safely grow hexagonal ice instead of cubic ice which can be formed below 200 K. This procedure helps to eliminate ambiguity during a ight. Prior to 2008 the older instruments switched from low to high gain around 55 C, creating a similar sublimation and regrowth

of the frost layer due to frost control oscillations instead of the dedicated mirror clear at 53 C. It is unlikely that cubic

ice or any other metastable form of ice forms on the FPH mirror after the mirror clear or gain change. To date, there have not been any proles where cubic ice has been identied.

Thornberry et al. (2011) conducted experiments using a laboratory version of the CFH to study the potential interference of gas-phase nitric acid (HNO3) on chilled mirror frost point hygrometers. Mixtures of moist air with up to 4 ppbv of HNO3 for 150 min presented no detectable change in the measured mirror temperatures, showing it is unlikely that HNO3 can change the frost point temperatures signicantly during a typical atmospheric sounding.

6 Pressure activated balloon valve system

Obtaining uncontaminated water vapor proles during a slow descent (i.e., no balloon burst) is accomplished using a valve situated in the neck of a balloon. This technology has been utilized on frost point hygrometer ights since 1964 when Mastenbrook (1966) developed a reliable and simple valve system to release helium from the balloon. A tting on the valve in the neck of an expandable latex balloon was opened at a predetermined pressure. The valve system evolved out of necessity for two main reasons. First, moisture captured on the balloon during ascent in the wet troposphere can contaminate the ascending hygrometer measurements. This contamination is from outgassing of the balloon and ight train in the dry stratosphere and is intermittent because the FPH pendulums in and out of the wake of the balloon. Once the valve opens the balloon ascent slows then starts to descend, and now the FPH leads the ight train with clean uncontaminated air below. Equally important, the valve provides slow and constant descent rates of 5 m s1 compared to a burst bal

loon that descends rapidly (> 40 m s1) in the stratosphere despite the parachute. The parachute is attached in case the valve fails to open or the balloon bursts before the designated pressure has been attained.

Kruchi et al. (2016) show an example of stratospheric contamination on an ascent prole starting near 25 km with uncontaminated measurements from the descent extending the prole above 27 km. When analyzing 141 ights from Lauder, New Zealand, instruments own without a sun shield encounter contamination starting in the stratosphere below 25 km 15 % of the time whereas instruments own with

a sun shield prior to 2010 see contamination 52 % of

the time. Although eliminating the sun shield has signicantly improved stratospheric data collection with regards to contamination on the ascent, the controlled uncontaminated valved descent proles continue to provide stratospheric data reaching above the contaminated ascent data.

The current version of the valve system consists of two pieces: a PVC valve and a digital microcontroller circuit board with a calibrated pressure sensor (Kruchi et al., 2016). The 175 g valve is constructed from PVC pipe with a plastic lid secured with waxed string. Separate battery packs are used to power the electronics board and to cut the waxed string with a heated nichrome wire at the desired pressure. Custom software allows the user to set the cut down pressure and apply a one-point pressure offset determined from a laboratory pressure standard. The overall accuracy of the pressure sensor, after the one-point offset is applied, is 1 hPa

with 0.01 hPa resolution. For proles over Boulder, Lauder, and Hilo the valves are set to open at 16.0 hPa, or roughly28.5 km.

The new version of the valve and electronics has increased

the success rate from 76 to 94 % for all balloon ights between 1991 and 2014 that did not burst before the cut down pressure was achieved. At the same time the new system decreased the valve system weight by 60 %, which saves helium.

7 Intercomparisons

7.1 AquaVIT-2

The Karlsruhe Institute of Technology (KIT) operates a large volume (84 m3) aerosol and cloud simulation chamber (AIDA) with the ability to vary pressure, temperature, and water vapor mixing ratio in a highly controlled fashion. In April 2013 a 2-week-long water vapor measurement inter-comparison experiment (AquaVIT-2) was performed at the AIDA chamber. AquaVIT-2 built off a previous intercomparison (AquaVIT-1; Fahey et al., 2014) at the AIDA chamber in October 2007 in which the NOAA FPH did not participate. A special laboratory version of the balloon FPH was constructed specically for AquaVIT-2 to externally measure chamber air extracted through a heated tube by a downstream pump. The laboratory instrument has three main differences from the balloon version. First, liquid nitrogen was used as the cryogen in the laboratory instrument, requiring a Teon sleeve to partially insulate the copper cold nger from the colder liquid nitrogen. Next, the sensor housing and intake tubing were modied to connect to the heated extraction tube without any leaks since most of the chamber experiments were done at sub-ambient pressures. Finally, the ow rates through the laboratory instrument were reduced when compared to the balloon sonde to conserve the chamber air.

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4305

Figure 8. Daily time series of temperatures (a) and pressures (b) for all data during the six non-blind experiments are shown. The relationship between chamber temperature and pressure with 36 individual stable water vapor mixing ratio segments is shown in panels (c) and (d), respectively. Areas contained by dashed lines indicate the range of water vapor and temperature in the actual atmosphere.

The purpose of AquaVIT-2 was to intercompare state-ofthe-art and prototype atmospheric hygrometers with each other and with independent humidity standards to determine precision and accuracy of the different instruments. Each day the chamber was held at one specic temperature while pressure and water vapor mixing ratio were changed throughout the day. Daily chamber temperatures ranged from 190 to 233 K over the 10-day experiment (Fig. 8a). Dynamic temperature excursions of nearly 10 K, caused by adiabatic

expansion and compression when air was removed from or added to the chamber, were not included in the analysis. The water vapor mixing ratios varied between 0.15 and

1000 ppmv while the pressure in the chamber started each day between 50 and 100 hPa and increased stepwise to a maximum of 3001000 hPa before stepping back down to 80100 hPa at the end of each day. Typically there were six distinct pressure steps throughout the day to intercompare the instruments during quasi-static conditions. Temperature and pressure in the chamber were stable within 0.3 K (1 )

and 0.35 hPa (1 ), respectively, during each of the 36 mea

surement segments examined. Three of the 10 experiment days were established as blind intercomparison days when data were sent directly to independent and impartial referees. Here we focus on comparing 6 of the 7 non-blind days between the NOAA FPH and the MC-APicT-1.4, an open-path, in-cloud, direct tunable diode laser absorption spectrometer (dTDLAS) (Khnreich et al., 2015). MC-APicT-1.4 is a successor of the initial APICT instrument described in Ebert et

al. (2005) and selectively measures, without any gas sampling process, the interstitial water vapor mixing ratio within the AIDA chamber (even within an ice cloud) with help from an open-path white cell integrated into the AIDA chamber. All MC-APicT-1.4 data were evaluated by the German National Metrology Institute (PTB) based on a proprietary, calibration-free rst-principle dTDLAS approach. The validity of this data evaluation was recently successfully demonstrated by comparing PTBs extractive, airborne dTDLAS eld hygrometer, SEALDH, with the German primary humidity standard at PTB (Buchholz et al., 2014). The 3 blind days are excluded because the data are still not available.

Temperatures and water vapor mixing ratios during nearly all the static air temperature segments were within the ob-served range measured by the NOAA FPH in the atmosphere. The exception was experiment 3 on 10 April 2013 (Fig. 8c), which was colder and drier than the real stratosphere. Only a quarter of the water vapor mixing ratios during stable pressure segments were within the range observable by the NOAA FPH (Fig. 8d). A total of 36 stable sections were analyzed, each lasting an average of 28 min. Excluding data from experiment 3, the agreement between the FPH and the MC-APicT-1.4 instrument was 1.0 5.5 % (2 ) for realis

tic atmospheric conditions between 2 and 600 ppmv. From here forward, the uncertainties presented with mean values are always twice the standard deviation. Below 1.5 ppmv the agreement is better than 30 % (Fig. 9a). Although the 15 30 % differences for experiment 3 appear to be large, the ab-

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

4306 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

Figure 10. Correlation plot showing all data collected between 0.1 and 10 ppmv over the 6 static non-blind days of experiments. A linear trend line (black) is shown with slope and offset.

mental issues. In these cases, the descent data are agged bad and removed during quality control after the ight.

Similarly, a dual ight with a CFH and an FPH using a single radiosonde and balloon was own in Boulder, Colorado, on 13 April 2015 (Fig. 12). The agreement in water vapor mixing ratio was 0.35 10 % even when taking into account

the differences larger than 10 % that occurred when the mixing ratio was changing quickly. Only ascent data were compared due to an unexplained cryogen loss at the top of the ight. The CFH is tuned to be a fast responding instrument with some oscillations in order to quickly respond to changes in water vapor during a prole. The error bars are larger on the CFH compared to the FPH during this ight (Fig. 12c) partially due to incomplete sunlight ltering coupled with fast responding stratospheric PID controller values (Vmel et al., 2016).

Figure 11a and b are plotted in absolute frost point temperature and frost point temperature differences to show the principal raw measurement of the instrument over an entire prole. However, the CFH FPH dual ight data are plotted in mixing ratio and mixing ratio percent difference to show the full range of water vapor mixing ratio measured during a ight.

8 Summary

Since 1980 the NOAA FPH has acquired high-quality monthly vertical proles of water vapor over Boulder, CO.The instrument has evolved over the years but the physical measurement principal and the calibration process has not deviated, making the stability of FPH measurement accuracy suitable for long-term monitoring and process stud-

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

Figure 9. Difference between NOAA FPH and MC-APicT-1.4 mixing ratios in (a) relative % and (b) absolute (ppmv) units for mixing ratios between 0.1 and 600 ppmv over the 6 days of non-blind experiments.

solute differences are small, ranging from 0.02 to 0.43 ppmv (Fig. 9b).

The FPH data collected over the non-blind days in the region of interest (0.110 ppmv) show good agreement with the MC-APicT-1.4 instrument. Even including the six stable sections < 1.5 ppmv where the FPH was measuring 1530 % higher than MC-APicT-1.4, the linear trend between the two instruments shows a slope of 1.014 and near-zero y intercept with minimal scatter as seen in Fig. 10.

7.2 Dual frost point hygrometer balloon ights

To exclude the atmospheric variability and pressure measurement differences, two FPHs were own using a single balloon and radiosonde on 8 November 2011. When comparing ascent and descent data the mean differences in frost point temperature and water vapor mixing ratio (0.25 km bin averages) between the two hygrometers were 0.019 0.25 C

and 0.163.9 %, respectively (Fig. 11). Instrument 1105

had frost control oscillations after the valved balloon turn on this ight, resulting in missing descent data until 21 km. Although the descent data often capture high-quality data above the ascent data, there are times when the descent data suffer, depending on atmospheric conditions or unknown instru-

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4307

Figure 11. Frost-point temperature differences from a dual FPH ight on the same balloon in Boulder, CO, are shown for the ascent (black) and descent (green). The frost point proles (b) along with the stratospheric water vapor mixing ratios between 3 and 7 ppmv (c) show agreement over the entire ight.

Figure 12. Mixing ratio deviations are shown between the CFH and FPH (a) beside the full stratospheric water vapor mixing ratio plot (b). Stratospheric water vapor mixing ratios between 2 and 7 ppmv are show in panel (c) where the error bars are 1 for the 0.25 km bins in

gray (CFH) and black (FPH).

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

4308 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

ies. In 2004, a second FPH sounding site was added at Lauder, New Zealand, initiating what is now the longest upper-atmospheric water vapor record in the Southern Hemisphere. A third site at Hilo, Hawaii, was started late in 2010 to measure tropical water vapor proles, complementing the measurement programs at the existing northern and southern midlatitude locations.

Improvements to the microcontroller version of the instrument in 2008 lead to more high-quality ights at the three NOAA/GMD water vapor sites. The biggest improvements to the FPH were implementing digital sunlight ltering to eliminate the sun shield, adding a lens heater, and creating a frost point-dependent gain schedule.

The various contributions to the uncertainty of the instrument are calculated for each prole. The overall uncertainty in the stratosphere is typically < 6 %, increasing to < 12 % in the troposphere. The majority of the uncertainty estimate resides in the frost control stability during a ight.

In 2014 a new continuous mirror thermistor calibration technique was adopted. This allows a six-point calibration t that reduces t errors to 0.01 C compared to 0.06 C

for the older three-point method. A lag correction was also developed to account for the larger PRT thermometer, with a slower response time than the tiny FPH thermistors. This correction was determined by comparing the results from the old and new calibration techniques for the same batch of 39 thermistors.

Comparison data during the AquaVIT-2 chamber experiment in Karlsruhe, Germany, with the MC-APicT-1.4 direct tunable diode laser spectrometer show good agreement over 6 non-blind days of experiments. For water vapor mixing ratios between 2 and 600 ppmv the agreement was 1.05.5%.

Dual FPH/FPH and FPH/CFH balloon ights show excellent consistency between instruments and do not give any indication of a bias.

9 Data availability

Vertical water vapor proles for Boulder, Lauder, and Hilo are available on the GMD public ftp: ftp://aftp.cmdl.noaa.gov/data/ozwv/WaterVapor/

Web End =ftp://aftp.cmdl.noaa. ftp://aftp.cmdl.noaa.gov/data/ozwv/WaterVapor/

Web End =gov/data/ozwv/WaterVapor/ . AquaVIT-2 data are not yet available for the public. Per the data agreement signed by the campaign participants, all data sets will be released to the public at a date to be agreed on by the referees and the instrument principal investigators. No date has been determined at this time.

Acknowledgements. The authors would like to thank Frank Polacek III, Ron Thorne, Michael ONeill, David Zim Sherman, Patrick Cullis, Hamish Chrisholm, Alan Thomas, David Nardini, Darryl Kuniyuki, Wills Dobson, Richard Querel, and all of the other people who have helped build and launch water vapor sondes since the program started in 1980. We thank the US GCOS Program for continued funding support for Lauder, New

Zealand, and the NOAA climate program ofce for support with operation costs at Hilo, Hawaii. Support from the NASA Upper Atmospheric Research Program provides imperative funding to all three NOAA/GMD sites. The AquaVIT-2 experiment was partially supported by the EMRP project METEOMET (ENV07) and FP7 project EUROCHAMP2. The EMRP is jointly funded by the EMRP participating countries within EURAMET and the European Union. The authors thank Chance Sterling for helpful conversation and comments on this manuscript.

Edited by: A. HofzumahausReviewed by: three anonymous referees

References

Brewer, A. W.: Evidence for a world circulation provided by the measurements of helium and water vapor distribution in the stratosphere, Q. J. Roy. Meteor. Soc., 75, 351363, 1949.

Buchholz, B., Bse, N., and Ebert, V.: Absolute validation of a diode laser hygrometer via intercomparison with the German national primary water vapor standard, Appl. Phys. B, 116, 883 899, doi:http://dx.doi.org/10.1007/s00340-014-5775-4

Web End =10.1007/s00340-014-5775-4 http://dx.doi.org/10.1007/s00340-014-5775-4

Web End = , 2014.

Clark, L. E.: Report of the determination of exactness of t of thermistor to the equations logR = A + B/(T + 2) and

logR = A + B/(T + 2) + CT, Tech. Rep. 2168, US Army Signal

Engineering Laboratories, 1961.

Ebert, V., Teichert, H., Giesemann, C., Saathoff, H., and Schurath, U.: Fibre-Coupled In-situ Laser Absorption Spectrometer for the Selective Detection of Water Vapour Traces down to the ppb-Level, Tm Technisches Messen, 72, 2330, doi:http://dx.doi.org/10.1524/teme.72.1.23.56689

Web End =10.1524/teme.72.1.23.56689 http://dx.doi.org/10.1524/teme.72.1.23.56689

Web End = , 2005.

Dessler, A. E., Zhang, Z., and Yang, P.: Water-vapor climate feedback inferred from climate uctuations, 20032008, Geophys.Res. Lett., 35, L20704, doi:http://dx.doi.org/10.1029/2008GL035333

Web End =10.1029/2008GL035333 http://dx.doi.org/10.1029/2008GL035333

Web End = , 2008.Fahey, D. W., Gao, R.-S., Mhler, O., Saathoff, H., Schiller, C.,

Ebert, V., Krmer, M., Peter, T., Amarouche, N., Avallone, L. M., Bauer, R., Bozki, Z., Christensen, L. E., Davis, S. M., Durry, G., Dyroff, C., Herman, R. L., Hunsmann, S., Khaykin, S. M., Mackrodt, P., Meyer, J., Smith, J. B., Spelten, N., Troy, R. F., Vmel,H., Wagner, S., and Wienhold, F. G.: The AquaVIT-1 intercom-parison of atmospheric water vapor measurement techniques, Atmos. Meas. Tech., 7, 31773213, doi:http://dx.doi.org/10.5194/amt-7-3177-2014

Web End =10.5194/amt-7-3177-2014 http://dx.doi.org/10.5194/amt-7-3177-2014

Web End = , 2014.

Forster, P. M. and Shine, K. P.: Assessing the climate impact of trends in stratospheric water vapor, Geophys. Res. Lett., 29, 1086, doi:http://dx.doi.org/10.1029/2001GL013909

Web End =10.1029/2001GL013909 http://dx.doi.org/10.1029/2001GL013909

Web End = , 2002.

Fujiwara, M., Shiotani, M., Hasebe, F., Vmel, H., Oltmans, S. J.,

Ruppert, P. W., Horinouchi, T., and Tsuda, T.: Performance of the Meteolabor SnowWhite chilled mirror hygrometer in the tropical troposphere: Comparisons with the Vaisala RS-80 A/H humicap sensors, J. Atmos. Ocean. Tech., 20, 15341542, 2003.Goff, J. A.: Saturation pressure of water on the new Kelvin temperature scale, T. Am. Soc. Heating Ventilating Eng., 63, 347354, 1957.

Goff, J. A. and Gratch, S.: Low-pressure properties of water from

160 to 212 F, T. Am. Soc. Heating Ventilating Eng., 52, 95 122, 1946.

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/

E. G. Hall et al.: Advancements of the NOAA frost point hygrometer 4309

Held, I. M. and Soden, B. J.: Water vapour feedback and global warming, Annu. Rev. Energ. Env., 25, 441475, 2000.

Hurst, D. F., Oltmans, S. J., Vmel, H., Rosenlof, K. H., Davis,S. M., Ray, E. A., Hall, E. G., and Jordan, A. F.: Stratospheric water vapor trends over Boulder, Colorado: Analysis of the 30 year Boulder record, J. Geophys. Res., 116, D02306, doi:http://dx.doi.org/10.1029/2010JD015065

Web End =10.1029/2010JD015065 http://dx.doi.org/10.1029/2010JD015065

Web End = , 2011a.

Hurst, D. F., Hall, E. G., Jordan, A. F., Miloshevich, L. M., White-man, D. N., Leblanc, T., Walsh, D., Vmel, H., and Oltmans, S.J.: Comparisons of temperature, pressure and humidity measurements by balloon-borne radiosondes and frost point hygrometers during MOHAVE-2009, Atmos. Meas. Tech., 4, 27772793, doi:http://dx.doi.org/10.5194/amt-4-2777-2011

Web End =10.5194/amt-4-2777-2011 http://dx.doi.org/10.5194/amt-4-2777-2011

Web End = , 2011b.

Hurst, D. F., Lambert, A., Read, W. G., Davis, S. M., Rosenlof, K.H., Hall, E. G., Jordan, A. F., and Oltmans, S. J.: Validation of Aura Microwave Limb Sounder stratospheric water vapor measurement by the NOAA frost point hygrometer, J. Geophys. Res.-Atmos., 119, 16121625, doi:http://dx.doi.org/10.1002/2013JD020757

Web End =10.1002/2013JD020757 http://dx.doi.org/10.1002/2013JD020757

Web End = , 2014.JCGM/WG 1: Evaluation of measurement data Guide to the expression of uncertainty in measurement, International Bureau of Weights and Measures, http://www.bipm.org/utils/common/documents/jcgm/JCGM_100_2008_E.pdf

Web End =http://www.bipm.org/utils/common/ http://www.bipm.org/utils/common/documents/jcgm/JCGM_100_2008_E.pdf

Web End =documents/jcgm/JCGM_100_2008_E.pdf (last access: September 2015), 2008.

Kley, D., Russell III, J. M., and Philips, C.: SPARC Assessment of Upper Tropospheric and Stratospheric Water Vapour, World Meteorology Organistaion (WMO), Geneva, 2000.

Kruchi, A., Philipona, R., Romanens, G., Hurst, D. F., Hall, E. G., and Jordan, A. F.: Controlled weather balloon ascents and descents for atmospheric research and climate monitoring, Atmos.Meas. Tech., 9, 929938, doi:http://dx.doi.org/10.5194/amt-9-929-2016

Web End =10.5194/amt-9-929-2016 http://dx.doi.org/10.5194/amt-9-929-2016

Web End = , 2016.Khnreich, B., Wagner, S., Habig, J. C., Mhler, O., Saathoff, H., and Ebert, V.: Time-multiplexed open-path TDLAS spectrometer for dynamic, sampling-free, interstitial H2 18O and H2 16O vapor detection in ice clouds, Appl. Phys. B, 119, 177187, doi:http://dx.doi.org/10.1007/s00340-015-6074-4

Web End =10.1007/s00340-015-6074-4 http://dx.doi.org/10.1007/s00340-015-6074-4

Web End = , 2015.

Lambert, A., Read, W. G., Livesey, N. J., Santee, M. L., Manney,G. L., Froidevaux, L., Wu, D. L., Schwartz, M. J., Pumphrey, H.C., Jimenez, C., Nedoluha, G. E., Coeld, R. E., Cuddy, D. T.,

Daffer, W. H., Drouin, B. J., Fuller, R. A., Jarnot, R. F., Knosp,B. W., Pickett, H. M., Perun, V. S., Snyder, W. V., Stek, P. C., Thurstans, R. P., Wagner, P. A., Waters, J. W., Jucks, K. W., Toon,G. C., Stachnik, R. A., Bernath, P. F., Boone, C. D., Walker, K.A., Urban, J., Murtagh, D., Elkins, J. W., and Atlas, E.: Validation of the Aura Microwave Limb Sounder middle atmosphere water vapor and nitrous oxide measurements, J. Geophys. Res., 112, D24S36, doi:http://dx.doi.org/10.1029/2007JD008724

Web End =10.1029/2007JD008724 http://dx.doi.org/10.1029/2007JD008724

Web End = , 2007.

Lee, J., Yang, P., Dessler, A. E., Gao, B.-C., and Platnick, S.: Distribution and radiative forcing of tropical thin cirrus clouds, J.Atmos. Sci., 66, 37213731, doi:http://dx.doi.org/10.1175/2009jas3183.1

Web End =10.1175/2009jas3183.1 http://dx.doi.org/10.1175/2009jas3183.1

Web End = , 2009.Mastenbrook, H. J.: A control system for ascent-descent soundings of the atmosphere, J. Appl. Meteorol., 5, 737740, 1966.Mastenbrook, H. J.: Operation Manual for Model 1012 Hygrometer,

Naval Research Laboratory Instruction Book, 176, 1981.

Mastenbrook, H. J. and Dinger, J. E.: Distribution of Water Vapor in the Stratosphere, J. Geophys. Res., 66, 14371444, 1961.Mastenbrook, H. J. and Oltmans, S. J.: Stratospheric water vapor variability for Washington, DC/Boulder, CO: 196482, J. Atmos.Sci., 40, 21572165, 1983.

Meyer, J., Rolf, C., Schiller, C., Rohs, S., Spelten, N., Afchine,A., Zger, M., Sitnikov, N., Thornberry, T. D., Rollins, A. W., Bozki, Z., Ttrai, D., Ebert, V., Khnreich, B., Mackrodt, P., Mhler, O., Saathoff, H., Rosenlof, K. H., and Krmer, M.: Two decades of water vapor measurements with the FISH uorescence hygrometer: a review, Atmos. Chem. Phys., 15, 8521 8538, doi:http://dx.doi.org/10.5194/acp-15-8521-2015

Web End =10.5194/acp-15-8521-2015 http://dx.doi.org/10.5194/acp-15-8521-2015

Web End = , 2015.

Mote, P. W., Rosenlof, K. H., McIntyre, M. E., Carr, E. S., Gille, J. C., Holton, J. R., Kinnersley, J. S., Pumphrey, H. C., Russell III, J. M., and Waters, J. W.: An atmospheric tape recorder: The imprint of tropical tropopause temperatures on stratospheric water vapor, J. Geophys. Res., 101, 39894006, doi:http://dx.doi.org/10.1029/95JD03422

Web End =10.1029/95JD03422 http://dx.doi.org/10.1029/95JD03422

Web End = , 1996.

Mller, R., Kunz, A., Hurst, D. F., Rolf, C., Krmer, M., and Riese,M.: The need for accurate long-term measurements of water vapor in the upper troposphere and lower stratosphere with global coverage, Earths Future, 4, 2532, doi:http://dx.doi.org/10.1002/2015EF000321

Web End =10.1002/2015EF000321 http://dx.doi.org/10.1002/2015EF000321

Web End = , 2016.

Murphy, D. M. and Koop, T.: Review of the vapour pressures of ice and supercooled water for atmospheric applications, Q. J. Roy. Meteor. Soc., 131, 15391565, 2005.

Oltmans, S. J.: Measurements of water vapor in the stratosphere with a frost-point hygrometer, Moisture and Humidity-1985-Measurement and Control in Science and Industry, 251258, 1985.

Oltmans, S. J. and Hofmann, D. J.: Increase in lower-stratospheric water vapour at a mid-latitude northern hemisphere site from 1981 to 1994, Nature, 374, 146149, doi:http://dx.doi.org/10.1038/374146a0

Web End =10.1038/374146a0 http://dx.doi.org/10.1038/374146a0

Web End = , 1995.

Oltmans, S. J., Vmel, H., Hofmann, D. J., Rosenlof, K. H., and Kley, D.: The increase in stratospheric water vapor from balloonborne, frostpoint hygrometer measurements at Washington, D.C., and Boulder, Colorado, Geophys. Res. Lett., 27, 3453 3456, doi:http://dx.doi.org/10.1029/2000GL012133

Web End =10.1029/2000GL012133 http://dx.doi.org/10.1029/2000GL012133

Web End = , 2000.

Randel, W. J., Wu, F., Vmel, H., Nedoluha, G. E., and Forster, P.: Forster, Decreases in stratospheric water vapor after 2001: Links to changes in the tropical tropopause and the Brewer-Dobson circulation, J. Geophys. Res., 111, D12312, doi:http://dx.doi.org/10.1029/2005JD006744

Web End =10.1029/2005JD006744 http://dx.doi.org/10.1029/2005JD006744

Web End = , 2006.

Read, W. G., Lambert, A., Bacmeister, J., Coeld, R. E., Christensen, L. E., Cuddy, D. T., Daffer, W. H., Drouin, B. J., Fetzer, E., Froidevaux, L., Fuller, R., Herman, R., Jarnot, R. F., Jiang, J. H., Jiang, Y. B., Kelly, K., Knosp, B. W., Kovalenko,L. J., Livesey, N. J., Liu, H.-C., Manney, G. L., Pickett, H. M., Pumphrey, H. C., Rosenlof, K. H., Sabounchi, X., Santee, M. L., Schwartz, M. J., Snyder, W. V., Stek, P. C., Su, H., Takacs, L. L., Thurstans, R. P., Vmel, H., Wagner, P. A., Waters, J. W., Webster, C. R., Weinstock, E. M., and Wu, D. L.: Aura Microwave Limb Sounder upper tropospheric and lower stratospheric H2O and relative humidity with respect to ice validation, J. Geophys.

Res., 112, D24S35, doi:http://dx.doi.org/10.1029/2007JD008752

Web End =10.1029/2007JD008752 http://dx.doi.org/10.1029/2007JD008752

Web End = , 2007. Richner, H. and Phillips, P. D.: Reproducibility of VIZ radiosonde data and some sources of error, J. Appl. Meteorol., 20, 954962, 1981.

Rohs, S., Schiller, C., Riese, M., Engel, A., Schmidt, U., Wetter, T., Levin, I., Nakazawa, T., and Aoki, S.: Long-term changes of methane and hydrogen in the stratosphere in the period 19782003 and their impact on the abundance of

www.atmos-meas-tech.net/9/4295/2016/ Atmos. Meas. Tech., 9, 42954310, 2016

4310 E. G. Hall et al.: Advancements of the NOAA frost point hygrometer

stratospheric water vapor, J. Geophys. Res., 111, D14315, doi:http://dx.doi.org/10.1029/2005JD006877

Web End =10.1029/2005JD006877 http://dx.doi.org/10.1029/2005JD006877

Web End = , 2006.

Rollins, A. W., Thornberry, T. D., Gao, R. S., Smith, J. B., Sayres, D. S., Sargent, M. R., Schiller, C., Krmer, M., Spelten, N., Hurst, D. F., Jordan, A. F., Hall, E. G., Vmel, H., Diskin, G. S., Podolske, J. R., Christensen, L. E., Rosenlof,K. H., Jensen, E. J., and Fahey, D. W.: Evaluation of UT/LS hygrometer accuracy by intercomparison during the NASA MACPEX mission, J. Geophys. Res.-Atmos., 119, 19151935, doi:http://dx.doi.org/10.1002/2013JD020817

Web End =10.1002/2013JD020817 http://dx.doi.org/10.1002/2013JD020817

Web End = , 2014.

Rosenlof, K. H. and Reid, G. C.: Trends in the temperature and water vapor content of the tropical lower stratosphere: Sea surface connection, J. Geophys. Res., 113, D06107, doi:http://dx.doi.org/10.1029/2007JD009109

Web End =10.1029/2007JD009109 http://dx.doi.org/10.1029/2007JD009109

Web End = , 2008.

Scherer, M., Vmel, H., Fueglistaler, S., Oltmans, S. J., and Staehelin, J.: Trends and variability of midlatitude stratospheric water vapour deduced from the re-evaluated Boulder balloon series and HALOE, Atmos. Chem. Phys., 8, 13911402, doi:http://dx.doi.org/10.5194/acp-8-1391-2008

Web End =10.5194/acp- http://dx.doi.org/10.5194/acp-8-1391-2008

Web End =8-1391-2008 , 2008.

Solomon, S., Rosenlof, K. H., Portmann, R. W., Daniel, J.S., Davis, S. M., Sanford, T. J., and Plattner, G.-K.: Contributions of stratospheric water vapor to decadal changes in the rate of global warming, Science, 327, 12191223, doi:http://dx.doi.org/10.1126/science.1182488

Web End =10.1126/science.1182488 http://dx.doi.org/10.1126/science.1182488

Web End = , 2010.

Stauffer, R. M., Morris, G. A., Thompson, A. M., Joseph, E., Coetzee, G. J. R., and Nalli, N. R.: Propagation of radiosonde pressure sensor errors to ozonesonde measurements, Atmos. Meas. Tech., 7, 6579, doi:http://dx.doi.org/10.5194/amt-7-65-2014

Web End =10.5194/amt-7-65-2014 http://dx.doi.org/10.5194/amt-7-65-2014

Web End = , 2014.

Steinhart, I. S. and Hart, R.: Calibration curves for thermistors,Deep-Sea Res., 15, 497503, 1968.

Tarasick, D. W., Davies, J., Smit, H. G. J., and Oltmans, S. J.: A re-evaluated Canadian ozonesonde record: measurements of the vertical distribution of ozone over Canada from 1966 to 2013, Atmos. Meas. Tech., 9, 195214, doi:http://dx.doi.org/10.5194/amt-9-195-2016

Web End =10.5194/amt-9-195-2016 http://dx.doi.org/10.5194/amt-9-195-2016

Web End = , 2016.

Thornberry, T., Gierczak, T., Gao, R. S., Vmel, H., Watts, L. A., Burkholder, J. B., and Fahey, D. W.: Laboratory evaluation of the effect of nitric acid uptake on frost point hygrometer performance, Atmos. Meas. Tech., 4, 289296, doi:http://dx.doi.org/10.5194/amt-4-289-2011

Web End =10.5194/amt-4- http://dx.doi.org/10.5194/amt-4-289-2011

Web End =289-2011 , 2011.

Vmel, H., Oltmans, S. J., Hofmann, D. J., Deshler, T., and Rosen,J. M.: The evolution of the dehydration in the Antarctic stratospheric vortex, J. Geophys. Res., 100, 1391913926, 1995. Vmel, H., David, D. E., and Smith, K.: Accuracy of tropospheric and stratospheric water vapor measurements by the cryogenic frost point hygrometer: Instrumental details and observations, J. Geophys. Res., 112, 8305, doi:http://dx.doi.org/10.1029/2006JD007224

Web End =10.1029/2006JD007224 http://dx.doi.org/10.1029/2006JD007224

Web End = , 2007a. Vmel, H., Yushkov, V., Khaykin, S., Korshunov, L., Kyro, E., and

Kivi, R.: Intercomparison of stratospheric water vapor sensors: FLASH-b and NOAA/CMDL frost point hygrometer, J. Atmos. Ocean. Tech., 27, 941952, doi:http://dx.doi.org/10.1175/JTECH2007.1

Web End =10.1175/JTECH2007.1 http://dx.doi.org/10.1175/JTECH2007.1

Web End = , 2007b. Vmel, H., Barnes, J. E., Forno, R. N., Fujiwara, M., Hasebe,F., Iwasaki, S., Kivi, R., Komala, N., Kyr, E., Leblanc,T., Morel, B., Ogino, S.-Y., Read, W. G., Ryan, S. C., Saraspriya, S., Selkirk, H., Shiotani, M., Valverde Canossa, J., and Whiteman, D. N.: Validation of Aura Microwave Limb Sounder water vapor by balloon-borne Cryogenic Frost point Hygrometer measurements, J. Geophys. Res., 112, D24S37, doi:http://dx.doi.org/10.1029/2007JD008698

Web End =10.1029/2007JD008698 http://dx.doi.org/10.1029/2007JD008698

Web End = , 2007c.

Vmel, H., Naebert, T., Dirksen, R., and Sommer, M.: An update on the uncertainties of water vapor measurements using cryogenic frost point hygrometers, Atmos. Meas. Tech., 9, 3755 3768, doi:http://dx.doi.org/10.5194/amt-9-3755-2016

Web End =10.5194/amt-9-3755-2016 http://dx.doi.org/10.5194/amt-9-3755-2016

Web End = , 2016.

Weinstock, E. M., Smith, J. B., Sayres, D. S., Pittman, J. V., Spackman, J. R., Hintsa, E. J., Hanisco, T. F., Moyer, E. J., St. Clair, J.M., Sargent, M. R., and Anderson, J. G.: Validation of the Harvard Lyman- in situ water vapor instrument: Implications for the mechanisms that control stratospheric water vapor, J. Geophys. Res., 114, D23301, doi:http://dx.doi.org/10.1029/2009JD012427

Web End =10.1029/2009JD012427 http://dx.doi.org/10.1029/2009JD012427

Web End = , 2009.

Atmos. Meas. Tech., 9, 42954310, 2016 www.atmos-meas-tech.net/9/4295/2016/