A. The near real-time NLDAS Phase 2 (NLDAS-2) forcing and model output data is currently running operationally at NOAA/NCEP/EMC. Automated scripts and programs gather and process the data needed to generate the NLDAS-2 forcing datasets. These scripts/programs run once per day, and update one day's worth (typically 12Z to 12Z) of forcing. This forcing data is then used to drive the various land-surface models to produce the model output. Scripts archive the data at the NASA GES DISC.

The NLDAS-2 forcing currently is available around ~3-4 days behind the current date. Long-time users of NLDAS products will note that this is a slightly longer delay than the 2-3 day data latency that was typical from NLDAS Phase 1. The primary reason for this is that the NARR model data used in the generation of NLDAS-2 forcing is currently in a quasi-operational mode and not available as early as the model data used in the generation of NLDAS-1 forcing. Also, GDS data may be behind by another couple days, as the updated index file is generated.

New for 2019: NLDAS-2.5 is expected to be running operationally at NOAA by the end of the year.  NLDAS-2.5 will have the same LSMs and grid as NLDAS-2, with the only change being different forcing datasets used every day to fill the ~4-day latency gap to make NLDAS truly real-time. See the NULDAS white paper for details (especially Figure 1) on the forcing datasets that will be used.

A. Yes, monthly NLDAS datasets for NLDAS-1 forcing, NLDAS-2 forcing, NLDAS-2 Mosaic, NLDAS-2 Noah, and NLDAS-2 VIC are now available from the GES DISC. Please see the updated NLDAS-1 README and NLDAS-2 README files for details about the creation and content of these datasets. The latest month of data will typically be available around 10-15 days after the beginning of the following month. Monthly climatology datasets are also available.

A. Daily NLDAS datasets are not currently available, and there are no immediate plans to make them available. The NLDAS group has received many requests for these datasets; however, some users have requested 00Z-00Z averages, some requested 12Z-12Z averages, and some requested 00LST-00LST averages. If you need daily averages, please download the hourly NLDAS datasets and create your own daily datasets.
 

A. This is the most popular NLDAS question. Hopefully, below you can find an appropriately short yet fully detailed answer!

For further information, please refer to the NLDAS-1 Forcing, NLDAS-1 Model, NLDAS-2 Forcing, and NLDAS-2 Model pages.

The most significant difference is the time frames of the datasets:

• NLDAS-1 is available from mid-1996 to the end of December 2007.
• NLDAS-2 is available from January 1979 to near real-time.

Another major difference between the two phases of NLDAS is the sources of observations and reanalyses used to create the respective forcing datasets. NLDAS-1 uses the 40-km NCEP Eta model-based Data Assimilation System (EDAS) for the surface meteorology, while NLDAS-2 uses the 32-km NARR system. For downward shortwave radiation at the surface, NLDAS-1 uses GOES-based satellite retrievals, with EDAS data used when/where not available; NLDAS-2 uses GOES data to bias-correct the NARR shortwave radiation.

Both NLDAS-1 and NLDAS-2 use a 1/8th-degree CPC daily gauge analysis as the source of the precipitation forcing. This analysis is PRISM-adjusted and was produced by the CPC using a least squares distance weighting scheme. NLDAS-2 uses this analysis over the entire data record, while NLDAS-1 used this analysis from 2002 onward. For the period 2002 and before (in the NLDAS-1 data only), the other daily gauge analysis used was a daily 1/4th-degree CPC product, which was generated using a Cressman analysis and interpolated by the NLDAS team to 1/8th-degree using the budget bilinear method.

During the period when 4-km hourly Doppler radar Stage II precipitation estimates are available (mid-1996 to present) over CONUS, both NLDAS-1 and NLDAS-2 first use these hourly estimates to temporally disaggregate the daily gauge analysis into hourly precipitation. If the radar estimates are not available (such as due to maintenance or coverage issues), the additional hourly datasets used to calculate the weights differ between NLDAS-1 and NLDAS-2. In NLDAS-1, the EDAS precipitation data is used over CONUS, Canada, and Mexico. In NLDAS-2 over CONUS or Mexico, 8-km half-hourly CMORPH satellite-retrieved estimates (2002 to present) are used if Stage II is not available. If CMORPH is unavailable, such as before 2002, the 2 X 2.5 degree CPC Hourly Precipitation Dataset (HPD) is used. If the HPD is also unavailable, then the NARR precipitation data is used. Over Canada, only NARR precipitation is used due to poor gauge coverage, with a 1-degree blending applied at the U.S.-Canada border.

The hourly estimates from the radar, CMORPH, HPD, EDAS, or NARR are only used to calculate the weights for the temporal disaggregation; the daily sum of the hourly NLDAS precipitation will be equal to the daily gauge analysis from the CPC. In addition, over CONUS, the hourly precipitation from both the NLDAS-1 and NLDAS-2 precipitation should be nearly identical when/where the hourly radar estimates are available. Small differences between NLDAS-1 and NLDAS-2 may still be present, however, owing to the 1/4th-degree daily gauge analysis used in NLDAS-1 before 2002 or to the nature of the amount/quality of precipitation gauge observations and radar estimates used between retrospective or real-time data productions. Further details about the NLDAS-2 precipitation can be found in the Appendix C describing the NLDAS-2 forcing data.

Several significant attributes are identical between the NLDAS-1 and NLDAS-2 datasets. They are both run on the 1/8th-degree NLDAS grid and are available hourly. The vegetation, soils, and elevation datasets used are also identical. The same four land-surface models (Mosaic, Noah, SAC, and VIC) are also used in both phases, although the models underwent some upgrades and corrections between the two phases of NLDAS. For more on these model improvements, please see Section 2.3 of Xia et al. (2012, Part 1). The spin-up procedures between NLDAS-1 and NLDAS-2 do differ, owing the different time periods between them.
 

A. NLDAS datasets are available at both the NASA GES DISC (for both retrospective and real-time) and from NOAA (currently real-time only). Datasets from either NASA or from NOAA/EMC can be used, and the fields in the datasets are identical, except where noted below. The file name conventions are also different between NASA and NOAA, as well as the directory structures. Within each year, NASA uses the ordinal date while NOAA/EMC uses YYYYMMDD. The real-time data from NOAA is GRIB-2.  All NLDAS datasets are the NASA GES DISC are GRIB-1. The GRIB tables and a few PDS GRIB values also differ between the NLDAS datasets at NASA and at NOAA. GRIB tables and documentation for NLDAS data as stored at the NASA GES DISC are found on their documentation website.

For NLDAS-2 Mosaic, the NOAA retrospective data provides the vegetation cover as a fraction (from 0 to 1) while the real-time data from NOAA is provided as a percentage (from 0 to 100). All NLDAS-2 Mosaic vegetation cover data provided at the NASA GES DISC is in units of fraction.

For NLDAS-2 Noah, NOAA provides the albedo and the moisture availability variables in units of fraction, while the NASA GES DISC provides these variables in units of percentage. NOAA/EMC provides the net shortwave, net longwave, and ground heat flux with signs reversed from the traditional ALMA direction for general energy balance components. For these variables at the NASA GES DISC, the signs are reversed from the NOAA data, and these three variables have the sign correct in the traditional direction. Finally, the units of the snow water-equivalent in the NOAA data are in meters, while in the NASA GES DISC data, the units are in [kg m-2] (or millimeters).

For NLDAS-1 Forcing, some PDS variable ID numbers were changed for the datasets at the NASA GES DISC to prevent confusion, as the data at NOAA/EMC uses identical PDS ID values for several variables.

A. There are a number of known issues with the NLDAS datasets:

• NLDAS-2 data from 2022-08-01 to 2024-03-01 were re-processed and made available from the NASA GES DISC on 7 June 2024: For details, please see this NLDAS News article.
• NLDAS-2 Forcing "A" APCPsfc and CONVfracsfc: Two days with zero precipitation over CONUS/Mexico (13Z 19 Feb 2018 to 12Z 21 Feb 2018); these days had heavy rain over the central U.S., which has resulted in drier-than-actual soil moisture in the region starting from 19 February 2018. (view1) (view2) (detail). UPDATE - 14 June 2018: NLDAS-2 Forcing "A" hourly files for these 48 times, and the February 2018 Forcing "A" monthly-average file were updated at the NASA GES DISC. No changes to the NLDAS-2 LSMs at this time. (detail).
• NLDAS-2 Forcing "A" APCPsfc: Dry spots of precipitation, resulting in very dry soil moistures in localized areas; start times range from mid-2010 to early 2015, and end on the analysis on 6 Apr 2016 (view1) (view2) (detail)
• NLDAS-2 Forcing "A" and Forcing "B" APCPsfc: Unrealistically high precipitation in Canada south of 46 North from mid-March 2012 to late-July 2014, with occasional recurrence to present (view1) (view2) (detail)
• NLDAS-2 Forcing "A" and Forcing "B" APCPsfc: Unrealistically low precipitation in Canada south of 50 North during 2002 (view) (detail)
• NLDAS-2 Forcing "A" APCPsfc: Unrealistically high precipitation in small localized regions in Texas and the Southeast U.S. for July and Aug 2008 (view) (detail)
• NLDAS-2 Forcing "A" APCPsfc: A band with lower values of the precipitation on the annual scale on the Canada side of the U.S.-Canada border for most years (view) (detail)
• NLDAS-2 Forcing "A" TMP2m: Warm bias in the 2-meter surface air temperature (small bias before 2008, becoming a more moderate bias at times from 2008 to present), especially in the SE U.S. during summer (detail)
• The issues with the forcing will then affect the soil moisture, runoff, fluxes, etc. in the model output in the regions listed.

A. Yes. The precipitation fields in the NLDAS-1 Forcing files and in the NLDAS-2 Forcing "A" and "B" files represent the total precipitation, including both rain and snow. The land-surface models each have their own individual method to determine how this total precipitation reaches the land-surface (as rain, snow, or some combination). Typically, the near-surface air temperature from the forcing at the same time interval as the precipitation is used to determine rain or snowfall. The NLDAS-2 model output files contain rainfall and snowfall fields as a result of their individual methods.

A. There are four evaporation components of the total latent heat flux in the NLDAS-2 model datasets.

They are as follows (with PDS ID number, variable names, full description, and units):

   199:EVBSsfc:Direct evaporation from bare soil [W/m^2]
   200:EVCWsfc:Canopy water evaporation [W/m^2]
   198:SBSNOsfc:Sublimation (evaporation from snow) [W/m^2]
   210:TRANSsfc:Transpiration [W/m^2]

These variables correspond to the following evaporation components of the ALMA standard:

   NLDAS names     ALMA names     ALMA description/units
   -----------     ----------     ----------------------
   EVBSsfc         ESoil          Evaporation from bare soil [kg/m2s]
   EVCWsfc         ECanop         Evaporation from canopy interception [kg/m2s]
   SBSNOsfc        SubSnow        Total sublimation from the ground snow pack [kg/m2s]
   TRANSsfc        TVeg           Transpiration from canopy [kg/m2s]

Additionally, the total latent heat flux is:

   121:LHTFLsfc:Latent heat flux [W/m^2]

For NLDAS-2 Mosaic, these components are defined as positive in the downward direction, which is the opposite direction as the traditional ALMA standard for these variables. However, the latent heat flux for NLDAS-2 Mosaic is defined as positive in the upward direction, which means that the latent heat flux is roughly equal to the negative of the sum of these four components for NLDAS-2 Mosaic:

   LHTFLsfc ~= -(EVBSsfc + EVCWsfc + SBSNOsfc + TRANSsfc)

As an additional complication, the sign of the SBSNOsfc variable in NLDAS-2 Mosaic switched directions on and after 9 March 2008. Before this date, NLDAS-2 Mosaic SBSNOsfc was defined as positive in the downward direction, but after this date, SBSNOsfc is defined as positive in the upward direction. Thus, on and after 9 March 2008, the evaporation balance equation for NLDAS-2 Mosaic becomes:

   LHTFLsfc ~= -(EVBSsfc + EVCWsfc - SBSNOsfc + TRANSsfc)

For NLDAS-2 Noah and NLDAS-2 VIC, these components and the latent heat flux are all defined as positive in the upward direction. The latent heat flux is roughly equal to the sum of the four evaporation components:

   LHTFLsfc ~= EVBSsfc + EVCWsfc + SBSNOsfc + TRANSsfc

There is another energy term that describes snow phase-change heat flux:

   229:SNOHFsfc:Snow phase-change heat flux [W/m^2]

The SNOHFsfc can be thought of as roughly identical to the sum of the Qsm and Qfz terms (albeit, in different units) from the ALMA standard.

When comparing these four evaporation components and the SNOHFsfc variable between NLDAS-2 Mosaic and Noah/VIC, please reverse the sign in the NLDAS-2 Mosaic data. The exception to this is for SBSNOsfc, but only on or after 9 March 2008.

Additionally, another term for the total evapotranspiration (in mass units) can be found here:

   057:EVPsfc:Total evapotranspiration [kg/m^2]

The EVPsfc variable is in mass units and the LHTFLsfc variable is in energy units, but both represent the total upwards water flux. A unit conversion between the two variables (assuming a constant latent heat of vaporization value) shows that the values of these variables should be roughly equal.

A. NLDAS-2 Mosaic, Noah, and VIC provide the hourly instantaneous surface skin temperature (aka, the temperature at exactly 00 minutes of every hour - not the average temperature over the entire hour). The "average" here refers to the average over the entire grid box for all vegetation, bare soil, and snow skin temperatures. For more on the definition of the AVSFTsfc variable, see the "AvgSurfT" description at the ALMA standard pages.

A. On 9 March 2008, the NLDAS-2 Mosaic data was transferred from the retrospective stream (starting back from Jan 1979) to the real-time simulations. Unfortunately, the canopy conductance is not properly defined in the real-time NLDAS-2 production. There are no plans to correct this in the NLDAS-2 Mosaic hourly or monthly output data from after this date.

A. The value used for Mosaic in NLDAS-1 is 175,000 J m-2 K-1, while the value used for Mosaic in NLDAS-2 is 70,000 J m-2 K-1. The name of this variable in the Mosaic code is "CSOIL0". The reason for this difference is explained in detail in Section 5.3 within Robock et al. (2003).

A. No particular geographic coordinate system was consulted when the NLDAS grid was originally configured. The grid boxes are simply 1/8th-degree boxes with the center of the lower-left grid box at 25.0625 N and 124.9375 W. The details of the grid are available here and here.

A. Yes, a shapefile (and associated files) for GIS can be found here.

You can also subset by shapefiles through Giovanni.

Users can select a bounding box region or subset using shapefiles, including countries, U.S. states, and watersheds.

A. Please see the NASA GES DISC for Data Access page for instructions on how to access the NLDAS datasets. Users must first register with Earthdata Login and then authorize NASA GES DISC Data Access. Detailed instructions for various methods to access the data are provided.