• Four general problems with observations:
  1. stations and observations are spaced at irregular intervals
  2. there are biases in sampling:
     • station observations:
       • land areas better sampled than ocean areas
       • surface is better sampled than upper atmosphere
       • developed nations have denser coverage than undeveloped regions
       • midlatitudes sampled better than polar regions
     • satellites
       • soundings in clear sky
       • cloud tracked winds can't see clouds underneath higher clouds, etc.
     • etc. for other observing platforms...
  3. lack of representativeness (how well does the observation represent the larger scale pattern?)
     • station sited in a mountain valley or next to a lake
     • station on windward or lee side of a mountainous, but isolated, small island
     • sonde was released into a thunderstorm (!)
     • etc.
  4. instrument and transmission errors
     • inherent inaccuracy of the device
     • inaccuracy of the method of deducing the data
       • what level should a cloud tracked wind be assigned to?
       • inversion of IR signature to obtain a satellite T sounding has limitations...
     • incorrect location, e.g. of a ship or aircraft
     • reduction to sea level problems
     • mistyped data transmission
     
        
• Many sources of data:
  • station based data considered the most reliable
  • satellite information and other items listed in the table of observing systems are mainly used to fill in the gaps of the station based network.

Back to ATM 111 homepage