Sensitivity of Surface Analyses to Temporal Observational Constraints

The quality of operational analyses can be affected by pragmatic decisions regarding the breadth of the temporal window within which to accept data as well as the length of time to wait before beginning the analysis. The latter constraint depends in turn on latencies between the time at which an observation is valid and when it is eventually received by an operational center. The Real-Time Mesoscale Analysis (RTMA) of the National Centers for Environmental Prediction (NCEP) includes 5 km resolution hourly analyses of temperature, wind, and moisture that depend on the availability of surface observations from a variety of sources. Since these analyses are designed to support field operations of the National Weather Service in a timely manner, observations used in the RTMA must fall within 6 minutes of the top of the hour and be received at NCEP by 25 minutes past the hour.

The latencies associated with surface observations collected from over 12,000 stations around the nation as part of MesoWest are examined as a function of network type. The fraction of observations valid within 6 minutes of the top of the hour is compared to that within broader temporal windows. The potential impacts of the choice of temporal window and delay prior to beginning the analysis are examined in a set of sensitivity experiments. Surface temperature and wind speed analyses from Spring 2007 are generated using the University of Utah version of the ARPS Data Assimilation System (ADAS). These analyses are performed at the same resolution as the RTMA but restricted to the western United States.