LDAS Meeting Minutes
September 21st
2001
NWS/OHD
I. - STATUS OF REALTIME LDAS RUNS:
On the NCEP N-LDAS demonstration workstation (SGI108), the four modeling groups are executing in realtime the LSMs of NOAH, MOSAIC, VIC, and SAC on the common N-LDAS CONUS grid, using realtime surface forcing produced by NCEP. The meeting began with a status review of these realtime LDAS runs.
a) NOAH (NCEP/EMC): presently executing in realtime and current up to present day, having cycled continuously since 17 Apr 99, when initial land states were taken from NCEP operational EDAS on that date, including soil moisture taken as a soil moisture availability fraction, with respect to saturation, in top 1-m of EDAS soil column.
b) MOSAIC (NASA/GSFC): recently restarted from 17 Apr 99 (using a newer version of MOSAIC in which the soil types are now independent of the plant types) using initial land states from the operational EDAS on that date, including soil moisture taken as a soil moisture availability fraction, with respect to saturation, in top 1-m of EDAS soil column. So far, reruns have cycled forward only to Oct 99, with catch-up to present day expected by next meeting.
c) VIC (Princeton): presently executing in realtime up to present day, having cycled continuously from a date in Oct 2000, using default VIC initial land state. Reruns from 17 Apr 99 to present will start soon.
d) SAC (NWS/OHD): presently executing in realtime and current up to present day, having cycled continuously since 17 Apr 99, when initial land states were taken from NCEP operational EDAS on that date, including soil moisture taken as a soil moisture availability fraction, with respect to saturation, in top 1-m of EDAS soil column. Catch-up to present day just recently achieved. Results not yet examined. These latest runs used EP forcing derived from an EP climatology method of John Schaake's. The forerunner realtime SAC results were discarded, as they were forced by EP derived from an experimental Penman-type relationship, and these EP values turned out to be much too high. A third set of reruns to the present realtime will be carried out using EP output from the NCEP NOAH model runs.
e) Ken Mitchell reviewed the disk space presently being occupied by each of the four N-LDAS modeling groups on the SGI108 LDAS workstation platform at NCEP and he presented formal quota limits on disk space use that each of the LDAS modeling groups must adhere to. A DLT tape drive is connected to SGI108 for off-loading LDAS fields to archive tape.
ACTION: Each group is to set up an automatic ftp (at least daily) of their realtime model output to the N-LDAS output archive at NASA/GSFC.
II. STATUS OF RETROSPECTIVE RUNS:
Brian Cosgrove of NASA/GSFC has finished the large task of producing retrospective surface forcing for the period 30 Sep 96 through 27 Sep 99, using 1) EDAS surface fields and Stage IV precip fields from the GCIP archive at NCAR, 2) reprocessed hourly GOES solar insolation from the GCIP project of Prof Rachel Pinker at U.Maryland, and 3) reprocessed daily gage-only CONUS precipitation analyses from the GCIP project of Wayne Higgins of NCEP/CPC. The meeting continued with a status report from each modeling group on their retrospective LSM runs over the period of this retrospective forcing. For each model, continuous cycling began at the common initial start time of 12Z, 30 Sep 96, using zero initial snowpack, and initial soil temperature and soil moisture (in terms of soil moisture availability relative to saturation) provided by Dag Lohmann from the NCEP Global Reanalysis II land states. Each institution is executing their retrospective runs on their own local computational platforms.
a) NOAH: The retrospective runs for the given 3-year period are complete.
b) MOSAIC: 10 months of the 36 months are complete. Finish likely by next meeting.
c) VIC: Runs to near end of 1997 are done. Completion expected in 20-30 days.
d) SAC: Full 3-year period is complete, using EP forcing derived from Penman-type relation (which again yielded poor EP values). Reruns are planned using both climatological EP and NOAH-output EP.
ACTION: All four modeling groups will provide a) all model output from their 3-year retrospective runs to the N-LDAS output archive at NASA/GSFC, and b) the soil moisture and soil temperature from each retrospective model run to Rutgers University, for the validation by Lifeng Luo and Alan Robock against Oklahoma Mesonet observations.
III. REVIEW OF N-LDAS PROJECT TERMS OF REFERENCE DOCUMENT
Ken Mitchell reviewed the "Terms of Reference Document" for the N-LDAS project, which is available from Ken on request. This document lays out group responsibilities and the guidelines and criteria for the realtime and retrospective sets of 1) LDAS forcing (data sources, frequency, producers), 2) LDAS LSM model runs (start dates, source of initial conditions, blending of various forcing sources). A new restriction was added to the document stating that no LDAS party would publicly present LDAS results from the LSM of another LDAS party until obtaining approval from the latter party. Additionally, for both the retrospective and realtime runs, each of the four modeling groups described to what extent they were able to adhere, in their respective LDAS LSM drivers, to the "blended forcing" criteria for precipitation and solar radiation, as laid out in the terms of reference. All four modeling groups felt their forcing blending was clean in their retrospective runs, but in the realtime runs, SAC has no precipitation blending and VIC may be lacking some aspects of the solar radiation blending.
ACTION: The SAC and VIC modeling groups need to modify their LDAS drivers to adhere to the precipitation and solar radiation blending criteria laid out in the LDAS Terms of Reference (i.e. when to substitute EDAS precipitation and solar insolation forcing in place of Higgins precipitation and GOES solar radiation respectively).
IV. REVIEW OF N-LDAS PROJECT RESULTS PRESENTD AT SEP 2001 GEWEX CONF.
Ken Mitchell presented figures and plots shown on the N-LDAS project poster at the International GEWEX Conference in Paris, France.
V. LDAS VALIDATION OVER OKLAHOMA MESONET BY RUTGERS UNIVERSITY
Lifeng Luo of Rutgers University, under the GCIP project there of Alan Robock, presented an extensive set of plots comparing LDAS surface forcing, such as precipitation, on the 0.125-degree LDAS grid (nominally 14-km resolution) to local station observations of the Oklahoma Mesonet (purchased by the GCIP program) at hourly, daily, 5-day, and monthly averaging periods. (See Lifeng's web site at http://climate.envsci.rutgers.edu/luo/research/ LDAS/validation.result.php) As one might expect, the point-wise precipitation obs and the gridded LDAS precipitation forcing sometimes showed large differences, especially over short time scales, since the LDAS precipitation analysis cannot include the Oklahoma Mesonet observations. Furthermore, as a pilot study of the LDAS evaluation that Rutgers will expand to all four LDAS LSMs, Lifeng showed comparisons of VIC-modeled LDAS soil moisture and skin temperature compared to OU Mesonet observations. The top layer VIC-modeled versus observed soil moisture were pleasingly similar, but at deep soil layers, the observations show a suspicious lack of temporal variability, especially at times near significant precipitation events. (See Lifeng's web site at http://climate.envsci.rutgers.edu/luo/research/LDAS/models.vic.php)
VI. UPDATE ON THE JOINT NASA/NCEP GLOBAL LDAS PROJECT (G-LDAS)
Members of the G-LDAS project at NASA/GSFC (Matt Rodell, Jon Gottschalck, others) presented recent results from their project. Matt presented comparisons for the month of June 2001 over Europe of three solar radiation products being evaluated in G-LDAS (GEOS, GDAS, and AGRMET). Other G-LDAS members showed examples of the global fields that the group has assembled over the 0.25-degree G-LDAS grid of soil type (from Reynolds, Jackson, and Rawls; 1999), surface slope, soil color, and monthly LAI. Additionally, Jon Gottschalck showed pilot results from his recent work to derive realtime weekly LAI for use in G-LDAS. Finally, Matt Rodell showed some preliminary sensitivity results of running the CLM "lake" model in G-LDAS executions of the CLM LSM. Kristi Arsenault recently began an effort to include the NOAH LSM into the G-LDAS driver, along with the already existing capability in G-LDAS to execute the LSMs of MOSAIC, CLM, and the Catchment Model. Matt Rodell showed some pilot G-LDAS soil moisture fields that were the result of cycling from a March 2001 start date to near-present realtime. More information on the global LDAS project can be found at GLDAS link at the NASA LDAS web site at http://ldas.gsfc.nasa.gov.