Background information about this type of data may be found in An Introduction to Atmospheric and Oceanographic Datasets.
As with any high volume operational model production system there were times when late file postings and network outages interrupted data acquisition and processing for the archive. The CISL/DSS made a "best" effort to obtain a complete archive time series.
CISL/DSS selected the FNL analysis over all others because it used the most complete set of observations and was run last in the sequence of operational global models. The FNL is likely the best option for a long-term operational model archive from NCEP. Note that NCEP also made forecasts using the same model, but DSS did not collect them.
Correspondence received from NCEP in June 1992 indicates that the production of the global analyses was given second priority. They considered it their most expendable product. When they experienced hardware failures and the like, they gave the MRF model priority, and sometimes (especially at 00Z) this meant skipping the final run which produced the global analyses. The topographic grids from NCEP, which correspond to the global grids, may be found in DS757.0..
For a 1987-1994 subset of this dataset, where gaps are filled with ECMWF data and monthly means are available, see this NCAR Climate Analysis (CAS) page.
INFORMATION ABOUT THE NCEP GDAS FNL 2.5 DEGREE 12-HOURLY ANALYSIS METHODOLOGY
The NCAR documents Information about NMC Analysis and Forecast Models and History of Changes in the NMC Analysis Techniques discuss model features and changes.The Hough analysis method was used through 1978Sep21. It forced a relationship between heights and winds which causes the zonal mean meridional wind at every level to be zero. This also caused other problems with winds in equatorial areas. The Global Data Assimilation System (GDAS) was started on 1978Sep22, using the optimal interpolation (OI) analysis method (9 layer). The quality of stratospheric global analyses are then very poor until about mid-1980, but better grids are in DS067.1.. On 1980May27, the GDAS was switched to a 12 layer (R24) model. Multivariate OI (R30) implemented 1982Aug14. On 1986May28, GDAS started using the Medium Range Forecast (MRF) model, with 18 layers and R40 resolution. On 1987Aug12, MRF87 implemented, raising resolution from R40 to T80. Throughout the period 1976-1987, and probably continuing, various changes made: in applications of satellite and buoy data; and in first guess fields and error-corrections.
Prior to 1981Feb22, the land temperature analyses were derived from the model surface energy budget. Then they were extrapolated from the lowest model layers. Surface temperatures over the oceans were at first derived from climatologies, then from analyses. Boundary layer grids, including wind, relative humidity, and potential temperature, start 1980Jun06.00Z. Sea level pressure starts 1977Dec08. Surface pressure starts 1978Sep21.00Z. Sea surface temperature starts 1979May16. Snow cover grids, about every 7 to 10 days, begin 1979Jul. Vertical motion not available, but could be calculated from the winds. No forecasts. See DS066.0. for forecasts, and vertical motion.
About 1991, NCEP acknowledged that the temperature is actually virtual temperature. I.e., in the LIST O (ON84 formatted) files, Q=16 means virtual temperature. In the LIST G (GRIB formatted) files, kpds=11 means virtual temperature (which would normally be kpds=12).
A humidity field problem is described here.
Some wind field problems are described here.
Sea surface temperatures, based on Navy analyses, begin 1979May16. For earlier analyses, see DS240.0.
Snow depth analyses begin 1979Jul25 and appear just once a week or so.
Surface elevation field from DS757.0.
1000mb 850mb 700mb 500mb 400mb 300mb 250mb 200mb 150mb 100mb 70mb 50mb 30mb 20mb 10mb
surface upper-air troposphere mandatory tropopause stratosphere sigma
A surface subset prepared by DSS is in DS082.1.
A tropopause subset has been built by DSS in DS082.2.
For higher levels (70 to 0.4mb), which are on a 65x65 grid, please see DS067.0.
FORMAT AND USAGE:
To see the WMO's GRIB Documentation click here.
To see NCEP's GRIB1 Documentation click here.
To see NCEP's ON84 Documentation click here.
To see FAQs about FNL and ADP data click here.
Tools to manipulate GRIB data
For support using NCAR/CISL's NCL click here.
For support using COLA's GrADS click here.
For support using NCAR/MMM's WRF click here.
For support using ATMET's RAMS click here.