The COOP network in its present state is an embarrassment to the NWS. Our best observers are generally over the age of 75, and are not getting any younger. The main problem in obtaining decent new observers is the requirement for manually taking observations on a daily basis, which just doesn't work in today's culture. The ideal *location* for COOP sites is at established government or educational institutions (Post Offices, Forest Service Ranger Stations, Universities, State/Local Government facilities, etc)... but these institutional sites tend to have the least reliable observers. If the systems were automated, this problem would be alleviated.

All stations should have a datalogger... which would eliminate the problems faced with once-daily observation times (Min Temperature "hangover") ... you could compute Min/Max values for any 24 hour period.

Ideally, you'd have two classes of stations. Stations in critical, data sparse areas would have real-time telemetry, less-critical stations would use memory sticks, etc.

As it is now, once-daily observations are inherently of far less value than realtime data. Realtime data fulfills the climate needs as well as forecast operation needs. If the cost of such a network is an issue, then reduce the overall number of stations. Half the current stations, with a significant number providing real-time data, would be more valuable than the current outdated, obsolete, and slowly-dying network we currently have.