Notes from Fourth LDAS meeting: November 23, 1998, 9:30 am, NCEP/EMC
Attendees:
NCEP/EMC - K. Mitchell, C. Marshall
NASA/GSFC - P. Houser, B. Cosgrove, R. Koster
NWS/OH - J. Schaake, Q. Duan,
NESDIS/ORA - D. Tarpley
Princeton - E. Wood
LDAS Progress, Decisions, and Discussion Summary:
1) Progress Milestones
-- NCEP/EMC has finished the software to spatially and temporally
interpolate 3-hourly, 40-km EDAS surface forcing fields to
hourly fields on the 1/8-th degree LDAS grid (C. Marshall)
--- a one-day snapshot (24 files, one per hour) was provided
on the LDAS server for development use by the LDAS parties
(NASA/GSFC was able to successfully read the above demo files)
--- example plots were presented of diurnal time series at a
point and spatial domain fields at a fixed time
--- next NCEP steps will focus on adding observed hourly GOES
surface insolation from NESDIS and observed hourly precipitation
(followed by terrain height adjustments to temperature etc)
-- NASA/GSFC has finished deriving on the LDAS grid the vegetation
classes (and fractional coverage for each) and the land/sea mask
for both the IGBP and SiB classification schemes (using the 1-km
EROS data sets as inputs)
--- next steps will focus on adding the BATS classification scheme
as a third choice and coming up with a single land/sea mask
(see more land mask and vegetation class discussion below)
-- The recently delivered NCEP LDAS server is now up and running
(4-processor SGI Origin 2000 with 2 GB of RAM)
--- NCEP has established external access via guest ids/passwords
for NASA/GSFC and Office of Hydrology (Princeton U. and
U. Washington are imminent, pending choice of support persons)
--- an LDAS and user directory structure has been established
--- FORTRAN and C compilers are up
--- W3 (GRIB) subroutines library is established
2) Decisions:
-- The group decided that during roughly the first year of parallel
LDAS runs of the several LSMs (land-surface models), we would NOT
try to use a unified vegetation classification scheme
--- use of a unified vegetation classification scheme (IGBP)
is a firm goal for year 2
--- currently 2 of the 3 LSMs use the SiB classification and
one LSM uses the BATS classification
--- switching to a unified vegetation classification too soon
may hamper the early performance of one or more of the LSMs
whose historical tuning is based on a different classification
--- tables of the necessary physical parameters for the IGBP
classification do not seem to be readily available as yet
-- A common LDAS land mask will be used by all participating LSMs
--- the unified mask will be employed as the mask in a) the common
GRIB forcing files, b) the physical model computations, and
c) the GRIB output of each LSM
--- NASA/GSFC will derive the single unified common land mask
from the land mask provided in the 1-km EROS vegetation classes,
rectifying (if necessary) modest differences in the land mask
in each vegetation classification
--- a 50 percent threshold will be used to identify land points
when deriving the 1/8-th degree mask from the 1-km EROS data
--- the LSMs will not perform physical computations over water
points in the mask, but may choose to model fractional coverage
of subgrid "lakes" at points designated "land" by the mask
-- In the hourly LDAS surface forcing files, NCEP/EMC will maintain
separate EDAS and observed forcing fields for solar insolation
and precipitation
--- each LSM will itself perform the necessary overlay of EDAS
and observed insolation and precipitation forcing
--- inter-LSM checks on the above overlays will be tracked by
requiring each LSM to chosen forcing summations
-- The support personnel for each participating LSM will learn to
read and write GRIB-formatted input/output files
--- NCEP/EMC will provide tutorial examples of read/write GRIB
subroutine calls, GRIB inventory utilities, and tables of
GRIB grid and product definition protocols (GDS and PDS)
--- NCEP/EMC will provide and maintain the NCEP "W3" subroutine
library (contains the GRIB library) on the LDAS workstation
-- The LDAS parties agreed that the GRIB spatial resolution limit
of .001 degree (about .11 km) does not require us to redefine
our chosen LDAS grid specifications
3) Other Discussion/Presentation:
-- Dan Tarpley reviewed the objectives and progress to date of the
GCIP-funded work of Kevin Gallo and Garik Gutman to derive a
U.S. database of the fractions of the top three most dominant
vegetation types (plus fraction of lakes and urban) per 20-km
grid box, and most importantly, a distinct annual cycle of greenness
for EACH of those 3 predominant vegetation types
-- Fulfilling an earlier Eric Wood request, Ken Mitchell reviewed the
NESDIS global 0.14-deg 5-year monthly climatology of green vegetation
--- provided the NESDIS journal publication by Garik Gutman documenting
the meaning, derivation, and intended application of the above
--- example point-wise time series and spatial maps were presented
--- algorithmic details of the application of the above in the
Eta LSM surface evaporation formulation were presented
--- debate emerged as to whether one can derive both greenness
fraction and LAI from NDVI
-- The length of the archives of the NESDIS GOES hourly insolation
and NCEP/CPC hourly precipitation products was reviewed