VOSClim Progress

1
VOSClim Progress

• Status report (Kent, North, Woodruff Berry)
• 5th VOSClim meeting (VOSClim-V) was held in Brest
  as part of the 3rd SOT meeting (SOT-3, 7th-13th
  March 2005, Brest, France, JCOMM MR35, 2005).
• General agreement of the importance of VOS for
  providing climate observations
• But is it too late?

2
VOSClim at/after SOT-3

• Only 100 ships recruited (initial minimum 200)
  
• Problems with real-time data access at NCDC has
  hampered use of the data
• Delayed mode data transmission is not working
  (due to difficulty in applying MQCS to IMMT-2 and
  IMMT-3 data before forwarding to the GCCs)
• Now have metadata (WMO47) to end 2004 (but not
  images)
• Now entering VOSClim evaluation phase (2005 to
  2007) but. lack of data means lack of users

3
Berry Kent (2005 in press)
4
Differences from background field
Skewness of distribution indicates how good the
screen exposure is.
5
Skewness v Screen exposure
6
Recommendations

• VOSClim
• Real time data NCDC requires adequate resources
  for the effective operation of the VOSClim DAC.
• Delayed mode data National Meteorological
  Services to implement MQCS for IMMT-2 or IMMT-3
  format delayed mode data to enable delivery of
  VOSClim delayed mode parameters or GCCs to
  consider accepting data without MQCS.

7
Errors in Air Temperature Gridded Datasets
Black line monthly error estimates (C) Red
line 12-month running mean
0.3o
0.2o

• The plot shows preliminary (measurement
  sampling) error estimates for the HadMAT air
  temperature dataset in a well-sampled 5 degree
  box in the North Atlantic.
• Errors are 50 greater now than in the 1970s

8
Errors in Air Temperature Gridded Datasets

• The red line shows number of observations at the
  6 hourly synoptic intervals (the black line is
  total number of obs)
• Recent increases in error are due to a persistent
  decrease in observations
• The increase in error may be underestimated as
  the calculation assumes a set correlation between
  the observations.
• The number of hourly observations from AWS is
  increasing resulting in increased correlation
  between observations in recent years.

9
VOS Sampling in ICOADS (January values)

• The number of ship observations of SST and
  pressure per month in ICOADS have dramatically
  increased (are these really ships?).
• Ship observations of air temperature, wind speed
  and humidity are greater, or similar to, 1970s
  levels.
• But numbers of ship observations from which we
  can calculate latent heat flux (requires all of
  the variables) have decreased.

10
VOS Sampling in ICOADS

• Surprising decrease in humidity observations

11
VOS Sampling in ICOADS
6 hourly
3 hourly

• Number of January latent heat flux observations
• The decrease in 'flux observations' is made worse
  by the decrease in the proportion of observations
  taken 6-hourly.
• More observations are being taken hourly,
  successive observations are highly correlated -
  we therefore need more observations to reduce
  sampling errors.

12
Recommendations

• VOS
• Urgently need GCOS observing system specification
  for the VOS fleet (not just based on observation
  numbers per area)
• The importance of ship observations which provide
  the full range of variables for surface flux
  calculation should be highlighted and the decline
  in the number of these observations reversed.
• Express concern that targeting observations for
  NWP may threaten the observations system for
  climate (eg Minos systems - P Tair only)
• Monitoring of VOS data against climate
  requirements should be implemented as a matter of
  urgency
• WMO to provide adequate resources for the
  development and electronic dissemination of
  Publication No. 47.
• VOS Operators to provide current VOS metadata to
  WMO.

13
(No Transcript)
14
Screen exposure - James Clark Ross