Notes for April 3 LDAS Team
Meeting
Rutgers University
Notes by Alan Robock
Present: Robock, Houser, Mitchell, Lohmann, Duan, Wood, Luo, Cosgrove
1. We discussed and approved a proposal from Ken Mitchell that he take over responsibility for the agenda of the meetings, taking notes at the meetings, and overall scientific direction of the project. The aim is to focus ourselves on specific milestones and to hasten completion of scientific publications. We agreed to try this for a while to see how it works.
2. The next LDAS Team Meeting will be on Thursday, May 10, at the NOAA Office of Hydrology (OH) at 1 p.m. There will also be a Global LDAS Team Meeting at OH at 10 a.m. the same day. All are welcome.
3. We agreed on plans for the retrospective LDAS runs. Before they can begin, the retrospective forcing will have to be completed to minimize errors, such as the stripe in precipitation across the US on Dec. 21, 1998 at 20UT, as shown by Eric Wood. The following rules will be followed:
a. Because of gaps in the Stage IV precipitation data set in 1996 and 2000, and the Cray fire on September 30, 1999, the evaluation period will be from October 1, 1997 through September 30, 1999.
b. Model simulations will begin one year earlier, on September 30, 1996 at 12UT. The first one year of simulation will be considered as a spinup period, and will not be considered in the formal evaluation, although the results will be saved for future analysis, and possible comparison to results of future retrospective runs with different initialization.
c. Five models will participate, NOAH, VIC, MOSAIC, SAC-SMA, and CLM.
d. All models will be initialized at the same percent saturation in the top 1 m of soil, which will be provided from the ETA EDAS system at September 30, 1996 at 12UT. [What about those models with deeper layers?]
e. All models will be initialized at the same soil temperature, extrapolated to the layers in each model from those which will be provided from the ETA EDAS system at September 30, 1996 at 12UT, for the 0-10 cm, 10-40 cm, 40-100 cm, and 100-200 cm layers.
f. The MOSAIC and CLM models will be run at NASA Goddard, the NOAH and SAC models will be run at NCEP, and the VIC model will be run at Princeton.
g. The first paper will be on the results of these reruns, or possibly on an evaluation of real-time output from the models, available since April 15, 1999.
h. All LSMs will use the common reprocessed forcing from NASA Goddard.
i. The precipitation forcing will consist of EDAS output for Mexico and Canada, and Stage IV rain gauge output reprocessed by Wayne Higgins for the US. They will be blended over a 2 deg. band along the borders. The Stage IV precipitation uses rain gauges for the 24 hour total precipitation and uses radar data to distribute the precipitation on an hourly basis during the day, without changing the total precipitation or the spatial pattern.
j. The temperature criterion to distinguish rain from snow will remain at 0 deg. C.
k. There will be no assimilation of snow depth observations. Rather each model will calculate snow based on the precipitation and energy inputs.
l. All output will be in the common hourly GRIB format. [Output is about 50 GB per model per year.]
4. Eric Wood presented 21 basins that he chose arbitrarily, but with a uniform distribution across the country that he will use for model evaluation. These are a subset of Schaake's 188 basins. He will also compare the LDAS radiative forcing with Surfrad, and will look at snow cover from three sources, Air Force, IMS Daily snow cover (Tarpley), and the Minnesota data (Tom Carroll).
5. Alan Robock described his current data collection, which includes Nebraska soil moisture for the past couple years, and Oklahoma soil moisture data for 1997-1999. He will contact Ken Crawford to get the 2000 Oklahoma data to reprocess, as was promised by Rick Lawford. Because flux data are important, he will also try to obtain flux data from the 10 Oklahoma Mesonet OASIS supersites, Nebraska, CART/ARM (CASES, Peggy Lamone), Arizona (Soroosh), Utah (John Horel), Georgia (Christa Peters-Lidard), and Oklahoma/Illinois (Tilden Meyers).
6. Brian Cosgrove discussed the Cray fire. The ETA was not run for several weeks, so for September-November, 1999, Global Reanalysis or GDAS will be used for forcing.
7. Ken Mitchell and Dag Lohmann, pointed out that the SGI100 is no longer available and the SGI108 should be used.