BRACKETING OSSES: SERIES 1

The following is a revised plan for conducting bracketing OSSEs related to evaluating the bounds of the potential trade space for tropospheric wind observations obtained from space-based Doppler wind lidars. The original plan was outlined in a letter from Steve Lord to Steve Mango dated May 6, 1998.

After several iterations with NCEP personnel, it was decided to begin with a horizontal resolution of 200 km for the simulated DWL data sets. A 6 hour data set has been simulated, synoptic (00, 06) data sets packaged in BUFR and placed on SWA’s anonymous ftp server ( cyclone.swa). Building on the 200 km horizontal resolution, SWA proposes a follow-on sequence of experiments to allow us to get started on preparing the required data sets. Four basic experiments are defined in the following Table 1.

TABLE 1

EXPERIMENT >	1. BEST CASE	2.PBL/CLOUD	3.MID/UPPER	4.WORST CASE
Horizontal spacing	200 km	200 km	200 km	600 km
Vertical coverage	Layers 1,2,3	Layer 1 plus layers 2,3 when cloud coverage > 10%	Layers 2,3 when cloud coverage < 25%	Layers 1,2 ,3
Vertical resolution	250 m < 2km 1000m >2km	250m <2 km 250m w/clouds	1000m >2 km	1 observation per layer 25% of the time
Swath width	2000 km	2000 km	2000 km	1000 km
Comments	This is the "near perfect" simulated DWL data set already generated and put into BUFR	This data set will be generated selective subsampling the "near perfect" data set	This data set will be generated by selective subsampling the "near perfect" data set	This data set will be generated by randomly subsampling the "near perfect" data set on just the port side of the ground swath
Estimated data volume	340 Mbytes/day	140 Mbytes/day	200 Mbytes/day	.4 Mbytes/day

Notes on Table:

All basic wind observations are assumed to be the product of processing all lidar samples taken within 200 x 200 x 1 km3 volumes above 2 km and 200 x 200 x .25 km 3 below 2 km. The assumed LOS measurement error (instrument) is .5 RMSE. The error of representativeness varies with the number of samples which vary with could cover. The nominal number of lidar shots taken through a 200 x 200 km2 area is ~40.
The horizontal spacing is the distance between "basic wind observations" in both the along track and cross track directions within the viewing swath.
Vertical coverage is described by the three layers defined by Lord (May , 1998)
Vertical resolution is defined by the prescribed data processing resolutions of 250m below 2 km and 1000m above 2 km.
The ~ 2000 km swath width results from the choice of 833 km orbit and a 45 degree nadir scan angle.
The ~ 1000 km swath width is obtained by using only half of the 2000 km swath width data. This is only a first approximation, since the geometries for the two different swath widths would be, in reality, somewhat different.
The 25% observation probability used in the worst case experiment is simulated by first randomly picking one observation level within each of the three atmospheric layers and then randomly deciding on whether to keep the observation (.25 probability).
The data volumes are taken to be before writing into the BUFR.

G. D. Emmitt

14 October, 1998

Return to SWA NPOESS Home Page
Return to SWA Lidar Home Page
Return to SWA Home Page

This page managed by Sidney A. Wood

Last modified: 14 Oct. 1998