Land Data Assimilation Schemes (LDAS): Parameters

[an error occurred while processing this directive] Land Data Assimilation Schemes (LDAS): Parameters

LDAS Logo

Land Surface Parameters

LDAS Vegetation Parameters Mapped To UMD Classification Scheme

The UMD vegetation classification scheme was chosen for use in the NLDAS project because of its accuracy in depicting the vegetation coverage of both the NLDAS domain and the rest of the world. While UMD vegetation coverage data sets are available, data sets containing associated UMD physical parameters are not. By contrast, physical paramater data sets based on other classification schemes (IGBP, BATS, NCAR LSM, SiB, SiB2 and Mosaic...hereafter refered to as secondary schemes) are readily available. As such, a multi-step mapping process was used to derive parameter values for UMD vegetation types from parameter values associated with secondary scheme types. Starting with vegetation maps from the UMD, IGBP, BATS, NCAR LSM, SiB, and SiB2 classification schemes, mapping fractions were derived which allowed secondary scheme vegetation classes (and thus parameters) to be mapped to UMD vegetation classes. Mapped secondary scheme parameter values were then averaged together to form a generic parameter value. These generic values (as well as pre-averaged secondary scheme values) are featured below as are values based on observations and mapping fractions.

Mapping Process:

1. Vegetation maps (UMD, IGBP, BATS, SiB, SiB 2) covering the NLDAS domain are statistically compared (using Fortran). A count is kept of how often each UMD vegetation category overlaps with each category from the secondary schemes. This count is then used to derive raw mapping fractions. For example, suppose that when the UMD map is compared to the SiB map, instances of UMD vegetation type 1 are colocated 20% of the time with SiB type 3 vegetation and 80% of the time with SiB type 5 vegetation. In this case the mapping fractions would be .8 and .2, the UMD LAI value would be arrived at as follows:

2. Subjective adjustments are made to the mapping fractions to insure that they make physical sense. For example, assume that SiB vegetation types are being mapped to UMD Category 1 (Evergreen Trees), and that the following raw mapping fractions are derived from a comparison of vegetation maps: UMD category 1 is colocated 80% of the time with SiB Category 4 (Evergreen Trees), and 20% of the time with SiB Category 12 (Broadleaf Trees With Winter Wheat). This result may be correct based on the straight comparison between vegetation maps, but it isn't logical. So in this case, the mapping fractions would be changed from .8 and .2, to 1, with a direct mapping between SiB and UMD Evergreen Tree types occuring as a result.

3. For each UMD vegetation type, the four largest mapping fractions of each secondary scheme vegetation type are kept, while the others are deleted. These four fractions are scaled up so that they total '1' and become 'final' mapping fractions. The lone exception to this step occurs with the UMD Urban vegetation class. In this case, all constitutent vegetation types are retained. This step acts as a quality control step and ensures that the mapping process is based upon well matched vegetation classes. For example, assume that the mapping fractions associated with a mapping of BATS vegetation types to the UMD Open Shrubland type were: .4, .2, .15, .15 and .1. In this case, the .1 would be deleted, and the other values would be scaled up such that they became .44, .22, .17 and .17.

4. Mosaic and NCAR LSM vegetation maps were unavailable, so final mapping fractions are created for these two schemes based on the final mapping fractions of the other secondary schemes.

5. Fractional mapping values are applied to vegetation parameters from the secondary schemes to yield mapped parameter values. For example, assume that we are interested in deriving UMD Category 1 LAI values from SiB data. Further, assume that the mapping fractions for this pairing of UMD and SiB data are .8 and .2, that is, UMD Category 1 is colocated with SiB Category 5 80% of the time and SiB Category 1 20% of the time. Therefore:
LAI Value For UMD Type 1 = ((.8*LAI Value Of SiB Type 5) + (.2*LAI Value Of SiB Type 1) )

6. Mapped parameters from the secondary schemes are merged with each other whenever possible as follows: When three or fewer versions of a mapped parameter are available, the mapped values are simply averaged together to form a merged parameter. When more than three versions are available, a particular version of a parameter was only included in the overall merging if it did not differ by more than 50% from the other versions of the parameter. For example, assume that mapped values of LAI for UMD Category 1 exist which are based on SiB, SiB2 and BATS (2.2, 2.6, 1.0). In this case, all three parameter values would be averaged together to form a 'generic' value. However, if LAI had existed based on SiB, SiB2, BATS and Mosaic ( 2.2, 2.6, 1.0, 3.0) then the BATS value would have been deleted and the remaining values from SiB, SiB2 and Mosaic would have been averaged together to form the generic value.

Static/Merged/Mapped Vegetation Parameters

--These tables feature static UMD vegetation parameter values. They are based on several vegetation data sets and have been derived using the six-step process described above.
--Variables in this table include: Height of Canopy Top, Height of Canopy Bottom, Roughness Length, Minimum Stomatal Resistance, Minimum Stomatal Conductance, Leaf Angle Distribution, Leaf Width, Leaf Length, Visible and Near IR Leaf Albedo, Visible and Near IR Stem/Dead Albedo, Visible and Near IR Leaf Transmittance, Visible and Near IR Stem/Dead Transmittance, Maximum Leaf Area Index, Minimum Leaf Area Index, Stem Area Index, Rooting Depth, Displacement Height, Maximum Fractional Vegetation Coverage, Difference Between Max. Frac. Coverage and Coverage at 269K, Canopy Cover Fraction, Visible and Near IR Soil Litter Reflectance and Overstory Flag

Table of LSM-based and observation-based vegetation parameters. Sources of data are listed in table.

web.veg.table.xls Actual Microsoft Excel file containing the information shown in the link above. Right click with mouse to download.

Monthly/Merged/Mapped Vegetation Parameters

--These tables feature monthly UMD vegetation parameter values. They are based on the Mosaic vegetation data set and have been derived using the six-step process described above.
--Variables in this table include: Roughness Length, Leaf Area Index, Greenness Fraction, Root Length Density, Zero Plane Displacement Height and Mosaic's Subcanopy Aerodynamic Resistance Parameter

Table of Mosaic-based monthly vegetation parameters which have been mapped to the UMD scheme.

web.veg.monthly.table.xls Actual Microsoft Excel file containing the information shown in the link above. Right click with mouse to download.

Static/Un-Merged/Mapped Vegetation Parameters

--These tables feature static UMD vegetation parameter values. They are based on several vegetation data sets and have been derived using steps 1-5 of the process described above. I.E. the parameter values have not been merged and so still exist as separate mapped secondary scheme values.

Table of LSM-based monthly vegetation parameters which have been mapped to the UMD scheme. Data from the different LSMs have not yet been merged into a single representative group of values.

web.veg.full2.xls Actual Microsoft Excel file containing the information shown in the link above. Right click with mouse to download.

Mapping Fractions

--These tables feature the mapping fractions that were used to map the IGBP, SiB, SiB2, BATS, Mosaic and NCAR LSM vegetation classes to the UMD vegetation classes.

HTML table of mapping fractions.

web.veg.fractions.xls Actual Microsoft Excel file containing the information shown in the link above. Right click with mouse to download.

NLDAS Web Page
Maintained and Updated at NOAA/NCEP/EMC

Kenneth Mitchell: Kenneth.Mitchell-at-noaa.gov
Youlong Xia: Youlong.Xia-at-noaa.gov