Sorry, you need to enable JavaScript to visit this website.

NLDAS-2 Forcing Dataset Information

Appendix A (GRIB-1 format info)            Appendix B (bias-corrected shortwave)            Appendix C (precipitation)

This dataset contains the forcing data for Phase 2 of the North American Land Data Assimilation System (NLDAS-2). The data are in 1/8th-degree grid spacing and range from 01 Jan 1979 to present. The temporal resolution is hourly. The file format is WMO GRIB-1.

The NLDAS-2 forcing datasets (hourly, monthly average, and monthly climatology) are available from the "Get the Data" link on the right or on the top menu.

The spatial domain, spatial resolution, computational grid, terrain height, and land mask of NLDAS-2 are identical to that in NLDAS-1, which is described in Section 2.1 of Mitchell et al. (2004).

Appendix A provides web addresses to NCEP online GRIB-1 documentation, source code, and parameter tables. In particular, readers of this document should read the discussion in Appendix A of GRIB-1 Parameter Tables.

The NLDAS-2 land-surface forcing files and land model output files use Parameter Table 130, which is oriented toward land/hydrology modeling.

The non-precipitation land-surface forcing fields for NLDAS-2 are derived from the analysis fields of the NCEP North American Regional Reanalysis (NARR).  NARR consists of: 1) a retrospective dataset starting from Jan 1979, and 2) a daily update execution at NCEP. The daily update provides a real-time NARR continuation known as the Regional Climate Data Assimilation System, or R-CDAS.

NARR analysis fields are 32-km spatial resolution and 3-hourly temporal frequency. Those NARR fields that are utilized to generate NLDAS-2 forcing fields are spatially interpolated to the finer resolution of the NLDAS 1/8th-degree grid and then temporally disaggregated to the NLDAS-2 hourly frequency. Additionally, the fields of surface pressure, surface downward longwave radiation, near-surface air temperature and near-surface specific humidity are adjusted vertically to account for the vertical difference between the NARR and NLDAS fields of terrain height. This vertical adjustment applies the traditional vertical lapse rate of 6.5 K/km for air temperature. The details of the spatial interpolation, temporal disaggregation, and vertical adjustment are those employed in NLDAS-1, as presented by Cosgrove et al. (2003).

The hourly land-surface forcing fields for NLDAS-2 are grouped into two GRIB files, "File A" and "File B". This is a change from NLDAS-1, which had only one hourly forcing file.


File A is the primary (default) forcing file. It will contain the following eleven fields (units are given in parentheses):

  1. U wind component (m/s) at 10 meters above the surface
  2. V wind component (m/s) at 10 meters above the surface
  3. air temperature (K) ** at 2 meters above the surface
  4. specific humidity (kg/kg) ** at 2 meters above the surface
  5. surface pressure (Pa) **
  6. surface downward longwave radiation (W/m^2) **
  7. surface downward shortwave radiation (W/m^2) -- bias-corrected (see Appendix B)
  8. precipitation hourly total (kg/m^2)
  9. fraction of total precipitation that is convective (no units): from NARR
  10. CAPE: Convective Available Potential Energy (J/kg): from NARR
  11. potential evaporation (kg/m^2): from NARR

** indicates a field to which the aforementioned vertical adjustment is applied.

The first eight fields above are the traditional land-surface forcing fields, such as in PILPS (Project for Intercomparison of Land-Surface Process Schemes) and GSWP (Global Soil Wetness Project).

The surface downward shortwave radiation field in File A is a bias-corrected field wherein a bias-correction algorithm was applied to the NARR surface downward shortwave radiation. This bias correction utilizes five years (1996-2000) of the hourly 1/8th-degree GOES-based surface downward shortwave radiation fields derived by Pinker et al. (2003). The bias-correction algorithm is described in Appendix B.

The precipitation field in File A is not the NARR precipitation forcing, but is rather a product of a temporal disaggregation of a gauge-only CPC analysis of daily precipitation, performed directly on the NLDAS grid and including an orographic adjustment based on the widely-applied PRISM climatology. The methodology and source datasets for producing this precipitation forcing, including details of the temporal disaggregation from the daily analysis to hourly intervals, are given in Appendix C.

The field in File A that gives the fraction of total precipitation that is convective is an estimate derived from the following two NARR precipitation fields (which are provided in File B): NARR total precipitation and NARR convective precipitation (the latter is less than or equal to the NARR total precipitation and can be zero). The convective fraction of total precipitation and/or the CAPE field in File A are used by some land models to estimate the subgrid spatial variability of the total precipitation.

The potential evaporation field in File A is that computed in NARR using the modified Penman scheme of Mahrt and Ek (1984). Potential evaporation is needed by some land models (such as the SAC model) that require potential evaporation as an input forcing.


One fundamental physical process represented in land modeling is the surface aerodynamic conductance, which represents the intensity of the near-surface vertical turbulence that transports heat and moisture between the land-surface and the overlying atmosphere.

There are many approaches to modeling the surface aerodynamic conductance in boundary layer modeling in general, and in land-surface modeling. The results from NLDAS-1 showed a surprisingly large difference among four different land models in the simulated magnitude of the warm-season diurnal cycle of aerodynamic conductance.

The 2-meter temperature and specific humidity and 10-meter wind fields applied in continental-scale land-surface modeling studies are typically products of the data assimilation/analysis systems of mainstream NWP centers. The 2-meter and 10-meter levels in such analysis/assimilation systems are rarely explicit levels in the background assimilating NWP model of the given NWP center. Hence NWP centers diagnose these 2-meter and 10-meter fields from the lowest prognostic level of the assimilating model. This diagnostic derivation of 2-meter temperature and humidity fields and 10-meter wind fields is done on the basis of: A) those same fields at the model's lowest prognostic level, which is usually well above the 10-meter level (e.g., 20-200 meters above), and B) the given model's method for modeling the aerodynamic conductance of the "surface layer", also known as the "constant flux layer" -- i.e., applying the given model's particular approach to such physical entities as the surface roughness length for momentum, the surface roughness length for heat, surface layer stability functions (e.g., similarity profile functions) and surface layer parameters (e.g., mixing length).

NLDAS-2 is therefore providing a second forcing file, File B, in which the surface temperature, humidity, and wind fields are represented not at 2 meters and 10 meters above the height of the NLDAS terrain, but rather at the same height above the NLDAS terrain as the height above the NARR terrain of the lowest prognostic level of the NARR assimilation system (namely, the same height above the model terrain as the lowest prognostic level of the mesoscale Eta model, which is the assimilating model in NARR).  We shall denote the latter height as "H", and this height H varies spatially in the horizontal. The motivation for this approach is to allow land models in NLDAS-2 to calculate their aerodynamic conductance from surface forcing fields that are significantly more independent (albeit not fully independent) of the aerodynamic conductance approach applied in the assimilation/analysis system from which the surface forcing fields were derived.

Specifically then, forcing File B of NLDAS-2 will contain the following ten fields:

  1. U wind component (m/s) at H meters above the surface
  2. V wind component (m/s) at H meters above the surface
  3. air temperature (K) ** at H meters above the surface
  4. specific humidity (kg/kg) ** at H meters above the surface
  5. pressure (Pa) ** at H meters above the surface
  6. height H above the surface (m)
  7. NARR surface downward shortwave radiation (W/m^2) -- without bias-correction
  8. NARR precipitation hourly total (kg/m^2)
  9. NARR convective precipitation hourly total (kg/m^2)
  10. NARR aerodynamic conductance (m/s)

Fields 7-10 above are provided for additional reasons, as follows. Fields 7 and 8 are provided to permit land modeling sensitivity tests in which the solar radiation and precipitation fields of File A are replaced by their less accurate counterparts taken directly from NARR. Fields 8 and 9 are the NARR fields used to derive the field of "fraction of convective precipitation" in File A. Field 10 is the aerodynamic conductance obtained from NARR, to allow comparison with the aerodynamic conductance computed independently by each land model in NLDAS-2.

Appendix A: The GRIB-1 Data Format (Documentation, Code, Parameter Tables)

The NCEP online User's Manual for the GRIB-1 data format is available here.

The NCEP web site providing Fortran-90 source code and documentation of code, as well as the NCEP GRIB-1 User's Guide, is here.

For a given physical variable, the GRIB data convention assigns the following:

  1. a unique numeric ID known as the GRIB parameter ID (range 1-255)
  2. a unique alphanumeric abbreviation (max of 8 characters)
  3. required physical units.

The unique GRIB parameter IDs for given physical variables are provided in tables known as GRIB parameter tables. The GRIB-1 Parameter Tables formally recognized at NCEP are available online here.

At of Dec 2007, NCEP formally recognized five GRIB-1 Parameter Tables, namely: Table 2, Table 128, Table 129, Table 130, and Table 140. Each of these tables define up to a maximum of 255 physical variables and their corresponding unique Parameter IDs. Additionally, each table has a Part 1 and a Part 2. Part 1 is identical across all GRIB-1 Tables and provides a WMO mandated list of 128 physical parameters and their WMO-mandated unique Parameter IDs. Part 2 is defined locally by the originating center or agency and the list of physical parameters in Part 2 is often aligned with a given physical specialty. At NCEP for example, Part 2 of Table 128 is oriented somewhat toward ocean modeling and ocean physics, and Part 2 of Table 129 is oriented somewhat toward cloud microphysics.

The NLDAS-2 land-surface forcing files and land model output files will utilize GRIB-1 Parameter Table 130, which is oriented toward land/hydrology modeling and land/hydrology physics.

The parameter IDs for Part 2 of Table 130 are available online here.

The parameter IDs for Part 1 of Table 130 are identical to those of Table 2, online here.

The parameter IDs GRIB tables specific to the NLDAS-2 data as stored on the NASA Hydrology DISC can be found here.

Appendix B: Bias Correction of Downward Shortwave Radiation for NLDAS-2

The NARR downward shortwave radiation field in the NLDAS-2 forcing files ("A" files) is bias-corrected to University of Maryland Surface Radiation Budget (SRB) dataset produced under the auspices of the GEWEX Continental Scale International Project (GCIP) and GEWEX Americas Prediction Project (GAPP) (Pinker et al., 2003). Data from the GOES-8 satellite is processed using an inference model to produce hourly estimates of downward shortwave radiation fluxes. This dataset is produced on the native 1/8th-degree NLDAS grid and no interpolation is necessary. A ratio-based (Berg et al., 2003) bias correction to the reanalysis downward shortwave radiation field was completed.

Appendix C: Generation of hourly precipitation forcing for NLDAS-2

The total precipitation field contained in File A is derived from CPC daily CONUS gauge data (Higgins et al. 2000; Chen et al. 2008 - with the PRISM topographical adjustment, Daly et al. 1994), CPC hourly CONUS/Mexico gauge data (HPD, Higgins et al. 1996), hourly Doppler Stage II radar precipitation data, half-hourly CMORPH data, and 3-hourly NARR precipitation data. Reflecting the strengths of each dataset, hourly NLDAS-2 precipitation is derived by using the Doppler radar, CMORPH products, or HPD data to temporally disaggregate the daily gauge products. This process, described in detail below, capitalizes on the accuracy of the daily gauge product, and on the temporal and spatial resolutions of the Doppler radar and CMORPH products.

Over CONUS, CPC PRISM-adjusted 1/8th-degree daily gauge analyses serve as the backbone of the NLDAS-2 hourly precipitation forcing. In Mexico, where this dataset is unavailable, CPC's 1-degree (1/4th-degree after 2001) North American daily gauge product is used instead. In NLDAS-2, these gauge-only daily precipitation analyses are first processed to fill in any missing values, and then are temporally disaggregated into hourly fields. This is accomplished by deriving hourly disaggregation weights from NWS real-time, 4-km Stage II and 8-km CMORPH hourly precipitation analyses. Stage II data is available from 1996 to the present, while CMORPH data is available from 2002 to the present. The Stage II product consists of WSR-88D Doppler radar-based precipitation estimates that have been bias-corrected using hourly multi-agency gauge data (Fulton et al. 1998), and mosaicked into a national product over the contiguous United States (CONUS) by NCEP/EMC (Baldwin and Mitchell, 1997). This CONUS mosaic of the Stage II product is interpolated to 1/8th-degree and any gaps in radar coverage (which total on average 13% of the area of the CONUS and are due to lack of radar coverage or equipment maintenance) are filled in with nearest neighbor Stage II data from within a 2-degree radius. If no Stage II data are available, then CMORPH data are used instead. CMORPH data is also used over the Mexican portion of the NLDAS domain which is outside of the Stage II's region of coverage. When CMORPH data is unavailable, such as before 2002, the 2 X 2.5 degree CPC Hourly Precipitation Dataset (HPD) is used. If HPD is also unavailable, NARR data is used instead. Details of the precipitation gauges used in the CPC analyses can be found here, which also links to a map of the gauges that ever provided data for these analyses, with about half of these gauges still active in the real-time data production.

The above hourly fields are then divided by fields of their respective daily totals to create hourly temporal disaggregation weights representing the proportion of the 24-hour total precipitation which fell in each hour. If the daily total is zero in an area of non-zero CPC precipitation, hourly weights are set to 1/24 to spread the precipitation evenly over the entire day. These hourly weights are then multiplied by the daily gauge-only CPC precipitation analysis to arrive at temporally disaggregated hourly NLDAS-2 fields. Since the Stage II, CMORPH, HPD, and NARR data are only used to derive the hourly disaggregation weights, a daily summation of these hourly NLDAS-2 precipitation fields will exactly reproduce the original CPC daily precipitation analyses. Since daily gauge and hourly precipitation data is sparse over Canada, NARR precipitation is used over all Canadian regions within the NLDAS domain. Rather than have an abrupt cutoff at the United States border, a 1-degree wide blending area is used. In this region, precipitation forcing consists of a weighted combination of the precipitation datasets discussed above.

Starting on 01 January 2012, NLDAS-2 transitioned from the unified CPC precipitation product to an operational CPC precipitation product, as the previous product was no longer being generated. The primary difference between the products is that the interpolation algorithm has changed (as the input precipitation gauge information is the same). For the unified CPC product from 1979 to 2011, the Inverse Distance method was used, with missing values replaced using a precipitation climatology. For the operational CPC product from 2012 to present, the Optimal Interpolation method (OI) is used, with missing values replaced using the value from the nearest station. This difference in the interpolation method has shown some differences in the behavior of precipitation, especially right on the U.S.-side of the border with Mexico, in western mountainous regions, and along coastlines.

The different precipitation products (by year and location) used for generation of hourly NLDAS-2 precipitation is summarized in the table below. Unfortunately, due to data availability and the quasi-operational nature of NLDAS, complete continuity of the data over all times/locations is not possible. Additionally, the number and location of the precipitation gauge observations stations has varied over the NLDAS-2 record. For details, please see Figure 2 in Mo et al. 2012 and this presentation by Chen and Xie. For known issues with the NLDAS-2 precipitation, please see this FAQ and answer.

*Table below scrolls to right

Dataset Years CONUS Mexico Canada
CPC daily gauge analysis (unified)
Daly et al. (1994)
Higgins et al. (2000)
1979 - 2011 1/8th-degree PRISM-adjusted analysis 1/4th-degree (before 2001, 1-degree) analysis Not used
CPC daily gauge analysis (operational)
Chen et al. (2008)
2012 - present 1/8th-degree PRISM-adjusted analysis 1/4th-degree analysis Not used
Stage II Doppler hourly 4-km radar data 1996 - present 1st choice to temporally disaggregate Not used Not used
CMORPH satellite-retrieved half-hourly 8-km analysis 2002 - present 2nd choice to temporally disaggregate 1st choice to temporally disaggregate Not used
CPC HPD 2x2.5-degree hourly gauge analysis 1979 - present 3rd choice to temporally disaggregate 2nd choice to temporally disaggregate Not used
NARR/R-CDAS 3-hourly 32-km model-simulated precipitation 1979 - present 4th choice to temporally disaggregate 3rd choice to temporally disaggregate Used for all precipitation with a 1-degree blend over U.S.-Canada border



Baldwin, M., and K.E. Mitchell, 1997: The NCEP hourly multi-sensor U.S. precipitation analysis for operations and GCIP research. Preprints, 13th AMS Conference on Hydrology, pp. 54-55, Am. Meteorol. Soc., Boston, Mass.

Berg, A.A., J.S. Famiglietti, J.P. Walker, and P.R. Houser, 2003: Impact of bias correction to reanalysis products on simulations of North American soil moisture and hydrological fluxes.  J. Geophys. Res., 108(D16), 4490, doi:10.1029/2002JD003334.

Cosgrove, B.A., et al., 2003: Real-time and retrospective forcing in the North American Land Data Assimilation System (NLDAS) project.  J. Geophys. Res., 108(D22), 8842, doi:10.1029/2002JD003118.

Daly, C., R.P. Neilson, and D.L. Phillips, 1994: A statistical-topographic model for mapping climatological precipitation over mountainous terrain.  J. Appl. Meteor., 33, 140-158, doi:10.1175/1520-0450(1994)033<0140:ASTMFM>2.0.CO;2

Fulton, R.A., J.P. Breidenbach, D.J. Seo, D.A. Miller, and T. O'Bannon, 1998: The WSR-88D rainfall algorithm.  Weather and Forecasting, 13, 377-395.

Higgins, R.W., J.E. Janowiak and Y. Yao, 1996: A gridded hourly precipitation data base for the United States (1963-1993). NCEP/Climate Prediction Center Atlas No. 1.

Higgins, R.W., W. Shi, E. Yarosh, and R. Joyce, 2000: Improved United States precipitation quality control system and analysis. NCEP/Climate Prediction Center Atlas No. 7.

Mitchell, K.E., et al., 2004: The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system.  J. Geophys. Res., 109, D07S90, doi:10.1029/2003JD003823.

Mo, K.C., L.-C. Chen, S. Shukla, T.J. Bohn, and D.P. Lettenmaier, 2012: Uncertainties in North American Land Data Assimilation Systems over the Contiguous United States.  J. Hydrometeor, 13, 996-1009, doi:10.1175/JHM-D-11-0132.1

Pinker, R.T., et al., 2003: Surface radiation budgets in support of the GEWEX Continental-Scale International Project (GCIP) and the GEWEX Americas Prediction Project (GAPP), including the North American Land Data Assimilation System (NLDAS) project.  J. Geophys. Res., 108(D22), 8844, doi:10.1029/2002JD003301.