GCOS GOOS WCRP/OOPC IX/3 

page 53 

However, none of these advances is complete! For example: 

   Millions of marine observations remain to be located and digitized from logbooks 

(e.g.,  Fig.  1),  and  millions  that  are  already  digitized  remain  to  be  blended  into 

ICOADS. 

   The  biases  in  marine  temperatures  around  1939 45  are  still  poorly  understood. 

Daytime MAT data need to be made useable. Our knowledge of biases for as much 

of  the  past  as  possible  needs  to  be  complemented  by  inclusion  of  appropriate 

metadata  in  data  sets,  so  that  proxy  and  historical  data  can  be  made  compatible 

with  modern  data;  also  enabling  future  data  to  be  made  compatible  with  current 

data. This is an application of the GCOS Climate Monitoring Principles (Appendix 

2  of  GCOS,  2003).  However,  it  is  recognised  that  finding  some  of  the  required 

metadata  will  be  difficult  and  may  need  augmenting  by  special  studies  of  the 

character of the data to make deductions about some of the observational practices. 

   We  are  still  improving  our  assessments  of  uncertainties  and  need  to  compare 

techniques  for  making  these  assessments;  we  also  need  to  specify  our  target 

accuracies.  

   The  global  observing  system  still  leaves  large  areas  unobserved  at  the  ocean 

surface and   especially   below. 

Other Boulder recommendations, such as creation of sub monthly analyses of SST and 

sea ice, and adjustment of historical wind speed data, are still at an early stage. CLIMAR II 

supported  the  need  for  sub monthly  (pentad)  analyses  because  they  provide  useful  ground 

truth even though they may be noisy or even impossible over most of the globe and most of 

the instrumental record because of the sparsity of observations. Pentad SST analyses based on 

satellite  data  (e.g.  Reynolds  et  al.,  2002)  are  very  valuable  but  require  in  situ  data  for 

validation and often for calibration also. Adjustment of historical wind speeds is particularly 

difficult without metadata. Some useful work has been done for the post Second World War 

period  (e.g.  Ward  and  Hoskins,  1996),  which  showed  that  the  problems  in  the  raw  data  are 

indeed serious, but this needs extending throughout the ICOADS period. QC techniques for 

all  parameters  need  to  be  fully  and  consistently  documented;  if  possible,  QC  methods  used 

throughout ICOADS should be homogeneous.  

There  were  seen  to  be  shortcomings  in  the  access  to  ICOADS  data.  Do  we  have 

optimal  methods  for  collecting,  preparing  and  providing  information?  There  are  many, 

overlapping  sources  of  data  and  products,  and  the  problem  of  optimising  data  provision  is 

complex.  Many  users  are  working  with  outdated  versions  of  COADS.  Often  data  are 

available,  but  it  is  difficult  for  the  uninitiated  to  discover  what  is  there.  There  should  be  a 

web based  route map  to the best available data which should be widely advertised to all the 

various user communities.  

The  Boulder  workshop  recommended  that  the  Voluntary  Observing  Ships  Climate 

(VOSClim) Project be extended, or a parallel project be initiated, to include buoys. CLIMAR 

II  discarded  this  recommendation.  With  the  planned  availability  of  buoy  metadata,  buoy 

versus  model  comparison  will  be  possible  from  existing  datasets.  Operationally  the 

monitoring of buoy data already takes place.