Attached are two files. Consider this a draft needing your comments. I would like to come up to NCEP as soon as possible to go over all of this. It is quite clear that these OSSEs are very critical to the NOAA/NASA decision process. I was out to visit with Tom Schlatter and Mike Hardesty last week and shared with them the list of metrics (currently eight). Those metrics seem to be right on target. We may want to do some study on how those metrics would apply to current observing systems to establish a baseline.

For Jim's sake, I'm including the content of my earlier memo.

I finished writing an explanation of the OSSEs as they relate to the range of DWL concepts that we envision. When I get input back from Sid, I'll send it on to you. The following are significant expansions on the plan:

1. Our simulated data sets will contain LOS measurements based upon the accumulation of shots that were distributed over the entire 200 km x 200 km processing grid AND a LOS measurement based upon clustering all the shots near a single point within the 200 km x 200 km area. The first data product will be more representative of the average wind within the 200 km x 200 km grid cell while the second product will be more like that which you get from rawinsondes today. All observations in our data set that have an N of -1 will be of the latter type.

2. I have added a "Reference" OSSE that ignores all cloud effects and returns perfect wind soundings from the one Nature Run grid point that is nearest the center of the 200 km x 200 km target area. This product does not map to any known DWL system being envisioned, but does satisfy your desire to look at observations that are not the result of averaging. It also will allow us to assess the importance of clouds in changing the value of DWL data products.

3. I have replaced (Should say recommend) the fourth OSSE that was to illustrate the "worst case" with a simulation that has more traceability to DWL data products, especially the one expected from a proposed ESE concept.The data results from a non-scanning lidar that makes a single LOS measurement every 200 km along its track. There is no cross-track coverage. It is the most extremely low horizontal coverage that we could expect (~ 2500 km between observations at the equator). This last set of observations can be subsetted at NCEP to obtain even worse data coverage scenarios.

4. We are defining two general series of bracketing OSSEs: "Coverage Series" and the "Accuracy Series". The "Accuracy Series" method for adding differing levels of measurement and total observation errors to the near perfect measurements provided for the "Coverage Series" has been explained. In general, we need to be very careful in how we define accuracy if we are to communicate with the instrument designers.

Dave Emmitt
President and Senior Scientist
Simpson Weather Associates, Inc.
809 E. Jefferson St.Charlottesville, Va. 22902
804-979-3571804-979-5599 fax
gde@swa.com