ARM Manus Radiosonde L3 Data 

A revised version of this L3 data set was release on 8 August 2013.  This
version corrected an error in the latitude data for the soundings from 29
September to 31 December 2011.  For these soundings the original L3 data 
set had the release location at 2.06N rather than 2.06S.  The release 
location as well as the latitude within the data section have been corrected.

This L3 version was released 22 February 2013.  This document summarizes the
surface data adjustments made by CSU to this version of the data.

The Manus site had two "co-located" sources of surface data. Source one came from an AWS (Automated Weather System) 
using an RTD sensor for temperature and HMP45 sensor for RH from an enclosed Stevenson screen enclosure with no forced 
air ventilation. Source two came from an ARM forced air ventilated surface MET system.

Observations from the AWS system were used to baseline Manus soundings from start of AMIE operations until and 
including 12 February 23Z after which time observations were switched to using instantaneous MET system data based. 
This switch to using MET data was based on Dr Long's recommendation as discussed in Long and Holdridge (2012). 
In this ARM report, it  was noted the that T and RH differences between the sonde surface values and those from 
the independent surface instruments were smaller when using the MET system values. This investigation of the 
Manus site was motivated by analyses based on field phase GTS-resolution Manus data which showed near-surface 
moisture gradients with specific humidity increasing with height in the mean. This raised a red flag 
since, to our knowledge, such conditions had not been observed in long-term mean records at other tropical locations.

To create a consistent data record at Manus, sondes for the entire AMIE period (24 September to 31 March) 
were reprocessed in the following manner. One minute averaged P,T,U data from the SMET system for the minute 
closest to sonde launch (taken from meta data information in sonde files) replaced the surface values found 
in the sonde file. A note was placed in the header lines for each sonde stating that "the surface data was 
replaced using 1-minute averaged data from the MET system" and original surface P,T, U, values were listed.

To examine the impact of this data change, an analysis of near surface moisture gradient (dq) was repeated but 
this time using the hi-res (2-sec) sonde data. For this analysis,  qd = q_sfc - q_pbl where q_pbl represents 
the mean q in the layer from first point above sfc to 950 hPa. At Manus, analyses were computed both using 
AWS and SMET sfc data for the period from late Sept to Feb. 12. For comparison, results at Gan are listed 
below for the period from late Sept. to Dec. 31.

Results show:

mean dq at Gan is +0.95 g/kg, varies from +1.4 during day to +0.6 at night.

mean dq at Manus is -0.09 g/kg using AWS data, varies from +0.47 during day to -0.52 at night.

mean dq at Manus is +0.49 g/kg using MET data, varies from +1.1 during day to 0.0 at night.

In other words use of AWS data gives one the impression that in the mean, q increases slightly from the sfc 
to the pbl (planetary boundary layer). However, using the MET sfc data this in no longer the case with the 
mean conditions showing the surface being ~0.5 g/kg moister than the pbl, which seems more consistent with 
other tropical sites. The near zero moisture gradient at nite may be related to a shallow inversion which often 
forms at night at Manus. This in turn causes a haze formation (and near surface water vapor removal) as the 
temperature drops below the deliquescence point for salt aerosals (Long et al. 2012).

Also note that that SMET mean surface data was cooler (0.42C), moister (0.59 g/kg), and had higher pressure 
(0.1 mb) than that from the AWS data for the period (24 Sept. 2011 - 12 Feb. 2012).

 
References:

Long, C. N. and D. J. Holdridge. 2012. "Investigations of Possible Low-Level Temperature and Moisture 
Anomalies During the AMIE Field Campaign on Manus Island." ARM Technical Report, DOE/SC-ARM/TR-119, available 
at http://www.arm.gov/publications/tech_reports/doe-sc-arm-tr-119.pdf

Long, C. N., D. J. Holdridge, and P. Ciesielski. 2012. "Evidence for a Manus Persistent Near Surface Night 
Inversion." Fourth Annual ASR Science Team Meeting Proceedings, Potomac, Maryland, 18-21 March 2012.