Notes from Sixth LDAS meeting: February 8, 1999, 9:30 am, NCEP/EMC
Attendees:
NCEP/EMC - Ken Mitchell, Curtis Marshall, Dag Lohmann, Mike Ek
NCEP/CPC - Wayne Higgins
NASA/GSFC - Paul Houser, Brian Cosgrove
NWS/OH - Yun Duan
NESDIS/ORA - Dan Tarpley, Garik Gutman
LDAS PROGRESS, DECISIONS, and DISCUSSION SUMMARY:
I. Progress Milestones
-- With the exception of the hourly time weights for the precipitation forcing, NCEP/EMC has finished the development and testing of the initial version of software to generate the realtime, hourly, 1/8-degree national forcing fields. NCEP/EMC announced the target date of 01 April 99 for EMC to begin realtime production of all hourly LDAS forcing fields.
-- NASA/GSFC continues to expand the LDAS Web site, in particular a) the realtime image-product generator for a vast array of LDAS land-surface model output fields, and b) refinements to the realtime MOSAIC-based LDAS prototype that currently uses straight NCEP Eta/EDAS forcing.
-- OH has finished the initial mapping of Doug Miller's (Penn State) 1-km soils database and NESDIS ORA's monthly greenness fraction database to the LDAS grid. OH will undertake a refined second version of these that strictly applies the common LDAS land mask.
II. Decisions
-- The group formally assigned the OH LDAS team the responsibility of a) defining soils characteristics on the LDAS grid and b) recommending soil hydraulic functions/parameters. Major OH progress has been accomplished on a) as cited in above section. Paul Houser had visited with Doug Miller during the Dallas AMS meetings, where Doug volunteered to assist our LDAS group with applications of his soils database as needed.
-- In addition to generating all the realtime LDAS forcing fields, NCEP/EMC agreed to undertake the responsibility of 1) mapping the three realtime snow analysis products of NESDIS, USAF, and NWS/NOHRSC to the LDAS grid and 2) developing a common river discharge routing model for the LDAS grid.
-- The realtime LDAS model execution suite will embody a three-step structure, consisting of 1) a delayed "final" run that will execute around 13-28 hours after realtime and will be the source of the continuously cycled LDAS state variables, 2) a less-delayed near-realtime "early bridge" run that will utilize early-look forcing (e.g. NCEP national Stage I or Stage II hourly HRAP precipitation analyses), and 3) a "forecast" LDAS run that will utilize NCEP NWP model forecast forcing. During the first months of realtime LDAS model execution, only component 1 of 3 will execute, while LDAS principals learn to build-in increasing realtime robustness and reliability and develop and refine archive procedures. Components 2 and 3 will be added in later months. A goal is to eventually get the time-lag of the "final" run down to about 12-13 hours (see below). Component 1 is the backbone of the entire system and is also the main component on which model intercomparisons will be performed (especially regarding streamflow, surface fluxes, soil moisture, skin temperature, snowpack and snowmelt).
III. Presentations and Discussions
A) Precipitation Analysis
Wayne Higgins of NCEP/CPC gave an excellent presentation describing his daily, gauge-based, 0.25-degree national precipitation analysis, which will form the underpinnings of our LDAS precipitation forcing. Wayne described his 1) gauge data sources, 2) extensive observation quality control, 3) objective analysis method (and alternatives he has tested), 4) his web site, and 5) his production timelines. Considerable discussion ensued concerning the 28-hour lag in the availability of his product and the reasons for that lag. Responding to our strongly voiced LDAS needs, Wayne kindly volunteered to try and provide LDAS an "early" analysis, with about a 13-hour lag, that will miss about 300-400 observations out of about 5000-6000 daily. Additional discussion focused on the omission of about 2000-3000 realtime hourly "HADS" gauge observations from Wayne's product, despite this HADS data stream being used in NCEP/EMC Stage I and II precipitation analyses. Ken Mitchell agreed to meet with Wayne later to try to forge links between this HADS observation pool and Wayne's data stream. That meeting has been set up for March 12.
B) GOES Surface Solar Radiation
Dan Tarpley of NESDIS/ORA reviewed Rachel Pinker's plans at U. Maryland to provide a 2-3 year retrospective archive of highly quality-controlled, hourly, GOES-based surface solar insolation and related products, to serve as a retrospective complement to the realtime retrievals that NESDIS/ORA will provide to the LDAS executions. Lengthy discussions took place regarding what primary products and secondary complementary products to ask Rachel to include in her retrospective archive. A strawman prioritized product list was formulated by the LDAS group, and Dan Tarpley agreed to record that list and distribute it for comment. Additionally, Rachel Pinker will attend a near-future LDAS meeting to a) further discuss the above proposed product list, b) describe her enhanced quality control procedures, and c) address whether the enhanced quality-controlled products could be produced sufficiently near realtime to be used by the realtime LDAS.
C) LDAS Production Timelines
Paul Houser spearheaded a fundamental and crucial discussion of possible scenarios for the timelines of realtime LDAS productions. This extensive discussion by all was very useful and led to a group decision to pursue the 3-step LDAS production structure cited above in the "Decision" section.
D) GRIB Data Packing Training
Curtis Marshall introduced Step 1 of LDAS group training on GRIB packing conventions and subroutine libraries. Curtis passed out NCEP and personal GRIB documentation and asked that LDAS group members review it in preparation for some explicit example walk-throughs of GRIB calling sequences at the next LDAS meeting.
E) Soils and Vegetation Greenness Database
Yun Duan of OH presented example displays and discussions of his LDAS work to map Doug Miller's 1-km soils database and NESDIS' global monthly greenness fraction database to the common LDAS grid. Early prototype versions can be obtained from the OH ftp site of ftp://www.nws.noaa.gov/oh/gcip/. Yun will add companion GIF displays soon and Paul or Brian will provide a link to those from the LDAS web site. Yun is preparing both 1) a "primary" 2-d soils database with just a handful of the fundamental soils parameters (such as dominant texture) and 2) a larger comprehensive 3-d soils database with much more extensive information and parameters, including information "layer by layer" in the soil column. Both databases will be depicted on the common LDAS grid. Follow-on group discussion focused on a refined strategy for applying the LDAS land mask to the above. Additional group discussion focused on coping with the dilemma wherein the original Miller soils database yields small regions with "undefined" soil properties or "bedrock".
F) Snow Analysis Products
Ken Mitchell provided 1) format/content/timeline/domain information and 2) web/ftp access sites for the three realtime snow cover and snow depth databases (NESDIS, USAF, NOHRSC) that the NCEP LDAS team will map in realtime to the LDAS grid for use by the LDAS project. The western U.S. NOHRSC snow-water equivalent analysis is a particularly critical LDAS input, since implicitly, the LDAS model simulations of snowpack will rely heavily on the NOHRSC product to offset the very poor quality and density of wintertime precipitation observations in western mountain states. Group discussion focused on the challenge of developing reasonable and workable algorithms to assimilate the above snow products into the continuously cycled snow water equivalent state variable of the LDAS models.
This site was constructed and is maintained by Brian A. Cosgrove:
