VEMAP Home

A USER'S GUIDE TO THE

VEMAP PHASE I DATABASE

November 1996 (revised September 2000)

---

Nan Rosenbloom
Timothy G.F. Kittel

To accompany CDROM and World Wide Web versions of:

THE VEMAP PHASE I DATABASE:

AN INTEGRATED INPUT DATASET FOR ECOSYSTEM AND VEGETATION MODELING
FOR THE CONTERMINOUS UNITED STATES

T.G.F. Kittel, N.A. Rosenbloom, T.H. Painter, D.S. Schimel, H.H. Fisher,
A. Grimsdell, VEMAP Participants, C. Daly, and E.R. Hunt, Jr.

Ecosystem Dynamics and the Atmosphere Section
Climate and Global Dynamics Division
National Center for Atmospheric Research

and

Climate System Modeling Program
University Corporation for Atmospheric Research

---

List of Tables and Figures

Table of Contents

---

1 INTRODUCTION

1.1 VEMAP and the VEMAP Database

The Vegetation/Ecosystem Modeling and Analysis Project (VEMAP) is an ongoing multi-institutional, international effort addressing the response of biogeography and biogeochemistry to environmental variability in climate and other drivers in both space and time domains. The objectives of VEMAP are the intercomparison of biogeochemistry models and vegetation-type distribution models (biogeography models) and determination of their sensitivity to changing climate, elevated atmospheric carbon dioxide concentrations, and other sources of altered forcing. The VEMAP exercise allows us to identify important commonalties and differences among model controls and responses. Where the models differ, the comparison highlights areas of uncertainty or error and identifies problems for future research. Inter-model differences also help to quantify the uncertainty in modeled responses to changing climate and other drivers.

The completed Phase I of the project was structured as a sensitivity analysis, with factorial combinations of climate (current and projected under doubled CO₂), atmospheric CO₂, and mapped and model-generated vegetation distributions. The highly structured nature of the intercomparison allowed rigorous analysis of results, while constraining the range of questions explored. Maps of climate, climate change scenarios, soil properties, and potential natural vegetation were prepared as common boundary conditions and driving variables for the models (Kittel et al. 1995). As a consequence, differences in model results arose only from differences among model algorithms and their implementation rather than from differences in inputs. Results from VEMAP I are reported in VEMAP Members (1995) and selected files are available through UCAR's anonymous FTP server (see Section 2.3). Abstracts describing the six modeling groups participating in VEMAP Phase I can be found under the subdirectory /docs.

The VEMAP input database for the Phase I model intercomparison is documented in this Technical Note. It includes compiled and model-generated datasets of long-term mean climate, soils, vegetation, and climate change scenarios for the conterminous United States. The data are on a 0.5deg. latitude/longitude grid. There are both daily and monthly representations of the mean climate. The climate data and climate change scenarios are presented in both gridded and time-sequential format. We developed the time-sequential, "site" file format to facilitate extractions of information for individual grid cells (Sections 4.3 and 12).

1.2 Citations and User Access Acknowledgments

The citations for the VEMAP database are:

Kittel, T.G.F., N.A. Rosenbloom, T.H. Painter, D.S. Schimel, H.H. Fisher, A. Grimsdell, VEMAP Participants[1], C. Daly, and E.R. Hunt, Jr. (1996) The VEMAP Phase I Database: An Integrated Input Dataset for Ecosystem and Vegetation Modeling for the Conterminous United States. CDROM and World Wide Web (URL=http://www.cgd.ucar.edu/vemap/).

Kittel, T.G.F., N.A. Rosenbloom, T.H. Painter, D.S. Schimel, and VEMAP Modeling Participants[2] (1995) The VEMAP integrated database for modeling United States ecosystem/vegetation sensitivity to climate change. Journal of Biogeography 22(4-5):857-862.

An additional reference for VEMAP is:

VEMAP Members[3] (1995) Vegetation/Ecosystem Modeling and Analysis Project (VEMAP): Comparing biogeography and biogeochemistry models in a continental-scale study of terrestrial ecosystem responses to climate change and CO₂ doubling. Global Biogeochemical Cycles 9(4):407-437.

Users are requested to acknowledge that access to the dataset was provided by the Climate System Modeling Program, University Corporation for Atmospheric Research, and the Ecosystem Dynamics and the Atmosphere Section, Climate and Global Dynamics Division, National Center for Atmospheric Research.

Development of the VEMAP database was supported by NASA Mission to Planet Earth, Electric Power Research Institute (EPRI), USDA Forest Service Southern Region Global Change Research Program, and NSF-ATM Climate Dynamics Program through the University Corporation for Atmospheric Research's Climate System Modeling Program.

1.3 VEMAP Mailing List

We have instituted an email list to keep VEMAP users informed of updates, future releases, and other information related to the VEMAP database. Our intent is to use this service as an electronic message board to quickly and easily disseminate pertinent database information. Archived list messages are available using the get command described in Appendix 5.3. For information on how to subscribe to the VEMAP mailing list, see Appendix 5.

---

2 DATA FILE ACCESS

The VEMAP database is available on CDROM, through the Internet on the VEMAP Web site, or via anonymous FTP from the UCAR anonymous FTP server.

2.1 CDROM

The VEMAP database CDROM contains the complete set of input files used for the VEMAP model intercomparison. We present the monthly data files in both gridded format (SVF) and time-sequential columnar format (site files), and the daily variables in binary format. Table 1 contains a list of the files available on the CDROM and Appendix A1 provides an overview of the CDROM directory structure. A README file resides under each subdirectory describing the files within that directory. We have included example postscript images under the subdirectory /images. Updated documentation and files for the CDROM can be found on the UCAR World Wide Web site:

or obtained through the anonymous FTP site (see Section 2.3), under the /vUPDATES subdirectory.

Table 1. Datasets available on the VEMAP CDROM, Web site (www), and FTP site.

*SVF = ASCII gridded "Single Variable Format" (see Section 4.1)

 

2.2 UCAR World Wide Web Site

The VEMAP files are accessible from the Internet. Using an Internet browser (e.g., Mosiac, Netscape, etc.), enter the URL:

Note for Macintosh Users: If you are using an early version of Netscape on a Macintosh, you may have difficulty downloading files. In this case, it is advisable to obtain the VEMAP files via anonymous FTP from ftp.ucar.edu (see Section 2.3).

The VEMAP home page contains a short description of the VEMAP project and a directory to additional pages. To gain access to the VEMAP dataset from the VEMAP home page, click on "Access to the VEMAP Dataset" under the heading, "The VEMAP Dataset". A listing of available data files by dataset category is given in Table 1.

 

2.3 UCAR Anonymous FTP Server - ftp.ucar.edu

To gain access to the VEMAP files via anonymous FTP from the UCAR FTP site, type:

> ftp ftp.ucar.edu
Name: anonymous
Password: <your_login>
ftp> cd edas/vemap
ftp> cd <subdirectory>
ftp> get <filename>

Available datasets are listed in Table 1 and the FTP directory structure is presented in Appendix 1.

 

2.4 GNU Compressed UNIX Tarfiles (Web Site and FTP Site Only)

We have archived multiple files into compressed UNIX tarfiles for more efficient data transfer. For example, all 12 gridded monthly maximum temperature files are stored in TX_MON.tar.gz (a .gz file suffix indicates that the tarfile has been compressed using the GNU software utility gzip, see below).

Tarfiles can be found on the FTP site under the subdirectory /tarFiles. To use these files, first FTP the desired tarfile to your home machine (remember to set the transfer mode to binary before FTP'ing the tarfiles). Then, decompress the .gz file and extract the archived files using the GNU gzip and tar commands appropriate to your machine. For example, on a UNIX system, type:

> gunzip <filename.gz>
> tar xvf <filename>

This process will create a complete set of files in your current local directory. The tarfiles frequently contain a considerable number of files and require sufficient space in your current directory. For a complete listing of the space required for the contents of each tarfile, download the file:

         /tarFiles/README.tarFiles

from ftp.ucar.edu.

Characteristic decompressed and/or extracted individual file sizes are:

         Daily binary: ~4700 kBytes/file
         Gridded SVF: ~33 kBytes/file
         Columnar Site: ~370 kBytes/file

 

2.4.2 How to Obtain Free GNU Software

The gzip utility is provided by the Free Software Foundation GNU project. It is available for multiple system platforms, and may be freely downloaded from the Internet. The Macintosh gzip version is available at:

http://persephone.cps.unizar.es/general/gente/spd/gzip/gzip.html

UNIX and MS-DOS versions, along with other GNU software, can be found at:

ftp://prep.ai.mit.edu/pub/gnu/

 

2.4.3 Hints for Using gzip on Selected Systems

UNIX: To compress files, use the command gzip; gunzip will decompress a .gz file. The command man gzip will provide more information on how to use gzip.

MS-DOS: To compress files, use the command gzip; gzip -d will decompress a .gz file. The command gzip -h will provide more information on how to use gzip.

Macintosh: The Macintosh web site provides full instructions on how to use MacGzip.

---

3 THE VEMAP GRID

The grid used for the VEMAP coverage is a 0.5deg. latitude x 0.5deg. longitude grid covering the conterminous U.S. Grid edges are aligned with 1.0deg. and 0.5deg. latitude-longitude lines; grid centers are located at 0.25deg. and 0.75deg. latitude-longitude intersections. Latitude and longitude for each cell are included in the VEMAP dataset (Section 7.3).

The grid's minimum bounding rectangle (MBR) is defined by grid domain corners given in Table 2. The full 0.5deg. VEMAP grid contains 5520 cells, with 115 columns and 48 rows.

Table 2. VEMAP grid corners defining the minimum bounding rectangle (MBR).

*Negative longitudes are degrees West.

---

4 FILE STRUCTURE

We use three file formats throughout the VEMAP dataset:

(1) ASCII SVF format for gridded monthly current climate, climate change scenarios, soils, vegetation, and georeferencing files.

(2) Binary time-sequential format for daily climate data. Each record contains the "characteristic year" of daily data for a grid cell (Section 8.3). Records are indexed by grid cell.

(3) ASCII column format for sequential monthly climate data and climate change scenarios. Each line presents 12 monthly values for a single grid cell. Records are indexed by grid cell (site files, Section 12).

 

4.1 SVF File Format

All gridded VEMAP data files are in an ASCII format based on, but not identical to, the SVF format specified by the GENAMAP Geographic Information System (GIS). Typical SVF files have 2 header lines followed by a 6-digit integer array. In contrast, VEMAP files have 5 header lines.

The first 2 lines are a VEMAP data access policy statement, followed by a blank line. These first 3 lines must be removed in order to convert the file to standard SVF format.

The 4th header line is a title line identifying the gridded variable and its units. For continuous data (i.e., non-categorical datasets), we also include the scale factor used to convert values to stored integers (Section 5.1). Division by this factor will restore the original value. Version number and revision date are also in the title line.

The 5th header line gives the gridded array's column and row indices (as four 6-digit integers): 1, 115 and 1, 48.

The header lines are followed by the gridded VEMAP integer array, which is dimensioned 115 columns x 48 rows (Section 3). The 6-digit integers in the VEMAP array include at least one blank space so that values in the file are space delimited. The array starts in the northwest corner of the grid, with the column index running west to east and row index running north to south (Fig. 1).

Figure 1. Layout of the VEMAP gridded array, with grid cell ID numbers.

The full grid contains 5520 grid cells, 3261 of which are within the boundaries of the conterminous U.S. and predominantly covered by land (see Section 5.2). Background cells (ocean and inland water cells) are assigned the value of -9999.

 

4.2 Binary File Format

All daily variables are stored in IEEE binary format. We have provided a FORTRAN program to read the binary files:

Each binary file contains 365 days of data for the 3261 grid cells with landcover in the U.S. Background grid cells are not included.

The files begin with two header lines containing information about the data. The first lists the variable name, units, scaling factor, and version number. The second describes the content of each data record with the following string:

The two header lines are followed by 3261 data records. Each record includes the grid point identifying number (ID), longitude, latitude, and a year's worth of scaled daily values. Only one year of data is given per grid cell, representing a "characteristic" year (see Section 8.3). Grid cell ID numbers begin at the top left corner of the grid and proceed left to right, top to bottom (Fig. 1).

Daily files on the NCAR FTP site or the WWW are available in GNU compressed format (Section 2.4) to speed FTP transmission.

 

4.3 ASCII Column Format for Site Files

Site files contain the monthly climate data and scenarios in column format, with each record containing 12 monthly values for a single variable. This time-sequential format was developed to facilitate data extraction for individual grid cells. Site files contain 8 header lines, beginning with a 2 line VEMAP data policy statement, followed by a blank line. The 4th header line is a title line identifying the gridded variable and its units. For continuous data (i.e., non-categorical datasets), we also include the scale factor used to convert values to stored integers (see Section 5.1). Division by this factor will restore the original value. Version number and revision date are also in the title line. The next 4 lines provide column headings for the data records.

In addition to the 12 monthly data values, each record contains auxiliary geographic information for each cell: grid cell ID, latitude, longitude, elevation, VEMAP vegetation type (Section 10 and Table 9), Küchler vegetation type (Appendix Table A3.2), and state identification number. The format of each data record is given in Table 3. VEMAP vegetation types listed in the site files are from version 2 of vveg (vveg.v2, Table 9). State identifying codes are listed in Appendix 4. Order of monthly values is January to December.

Table 3. Format of data records in site files.

*Scaling factor for monthly values is stored in the 4th header line.

Site files include only non-background grid cells, so that there are 3261 data records per file.

---

5 SCALING FACTORS, BACKGROUND VALUES, AND LANDCOVER MASK

5.1 Scaling Factors

In SVF, site, and daily binary files, data values are represented by scaled integers. We produced these integers by multiplying the original data values by a scaling factor (e.g., 100.0, 0.001) which is included in the fourth header line of the SVF and site files and first header of the daily binary files. Division by this scaling factor will restore the original value. For example, if the listed value equals 297 and the scaling factor equals 10.0, the actual value equals: (297/10.0 = 29.7).

5.2 Background Values

Data files contain roughly 2200 cells that are outside the physical or political boundaries of the conterminous U.S. (i.e., outside the VEMAP domain). In the SVF gridded files, these cells are set to the stored background value of -9999. In addition, cells dominated by large inland water bodies (e.g., Lake Michigan, Great Salt Lake) are also set to -9999.

In some files, generally those containing VEMAP Phase I model results, cells classified as wetlands (in vveg.vx, Table 9) are also set to background. Counts of background cells and data cells, either including or excluding wetlands, are given in Tables 4a and 4b, respectively.

Table 4a. Number of background cells and cells within the VEMAP domain
with land cover (including wetlands).

Most VEMAP data files contain data for wetland cells, such that the non-background cell count for data files is 3261 cells (Table 4a). Exceptions are latitude and longitude files (Section 7) in which all cells are filled with data (non-background cell count = 5520), and elevation and vegetation files (non-background cell count = 3385). Typical non-background grid cell counts for VEMAP Phase I results files are 3168 cells because most of the models were not run for the wetland cells (Table 4b).

 

5.3 Landcover Mask

Any soil or gridded climate file can be used as a VEMAP domain land mask. In these files, background values (-9999) indicate cells outside the domain or over inland water bodies, and all other values identify non-water cells within the domain.

---

6 ROAD MAP TO FILE AND VARIABLE DESCRIPTIONS

In the following sections (Sections 7 - 11), we describe VEMAP database variables and associated files. Each of these sections follows this general outline:

* Summary of available variables
* File naming protocol
* Derivation of variables (for most sections)
* Description of individual variables

For subsections that describe individual variables (e.g., Section 7.3.1), subsection headings include the variable name code used in filenames (in parentheses) and units (in square brackets).

Descriptions include data sources and derivations where appropriate. At the end of each subsection, names of gridded SVF files, daily binary files (when present), and scaling factors are listed. We list a background cell value of "N/A" if all cells are filled with data.

---

7 GEOREFERENCING VARIABLES AND CELL AREAS

7.1 Summary of Geographic Variables

The VEMAP dataset includes three georeferencing and three cell area variables (Table 5). These are described in more detail in Section 7.3. On the CDROM and FTP site, these data files are located under the subdirectory /geog. Note that the area variables are related:

Table 5. Geographic variables. Variable name codes are those used in filenames
(Section 7.2).

7.2 Filename Protocol

The filename protocol for area and georeferencing files is:

VAR

where:

7.3 Variable Descriptions

 

7.3.1 Elevation (VAR = elev) [m]

Elevation was aggregated from 10-minute Navy Fleet Numeric Oceanographic Center (NFNOC 1985) data (C. Vörösmarty, personal communication). Aggregated elevation for each 0.5deg. cell was computed as a simple mean of nine 10-minute grid cell modal values. Elevations for inland water bodies are included; non-background cell count = 3385 (see Section 5.2).

Gridded SVF file: elev
Scaling factor: 1.0

 

7.3.2 Latitude (VAR = lat) [degrees and decimal degrees]

Latitude of grid cell center. Positive for North latitudes. All cells are filled with latitude values; there are no background cells.

Gridded SVF file: lat
Scaling factor: 100.0

7.3.3 Longitude (VAR = lon) [degrees and decimal degrees]

Longitude of cell center. Scaling factor gives negative degrees for West longitudes. All cells are filled with longitude values; there are no background cells.

Gridded SVF file: lon
Scaling factor: -100.0

7.3.4 Area (VAR = area) [km²]

Absolute area of a grid cell. Determined by coordinate geometry.

Gridded SVF files: area
Scaling factor: 1.0

 

7.3.5 Percent Land Area (VAR = areap) [%]

Percent of the area of a 0.5deg. latitude/longitude grid cell that is covered by land and within the VEMAP domain (the conterminous U.S.). Derived from the Kern U.S. EPA 10-km gridded soil coverage (Section 9), this is the number of non-zero 10-km pixels relative to the total number of pixels in a 0.5deg. cell.

Gridded SVF files: areap
Scaling factor: 1.0

 

7.3.6 Absolute Land Area (VAR = varea) [km²]

Absolute area of a grid cell that is covered by land and within the VEMAP domain (the conterminous U.S.). Absolute land area is determined as:

varea = (area) x (areap/100)

Gridded SVF files: varea
Scaling factor: 1.0

---

8 DAILY, MONTHLY, AND ANNUAL CLIMATE DATASETS

8.1 Summary of Climate Variables

The database includes 21 climate variables (Table 6), which are described in Sections 8.4 - 8.8, and are presented in daily, monthly, and annual files. Section 8.3 discusses development of the daily and monthly versions. On the CDROM and FTP site, these

data are in the subdirectories /daily and /monthly. Selected climate variables are also available in site file format (Sections 4.3 and 12).

Table 6. Climate variables. Variable name codes are those used in filenames (Section 8.2).

8.2 Climate Filename Protocol

8.2.1 Gridded Monthly and Annual SVF Files

The filename protocol for gridded monthly and annual climate SVF files, with the exception of gridded absolute temperature files (Section 8.2.2), is:

VAR.MMM

where:

8.2.2 Gridded Absolute Temperature SVF Files

Protocol for naming gridded absolute temperature and month of occurrence of absolute temperature files is:

P_VAR

where:

8.2.3 Binary Daily Files

Filenames for binary daily files follow the form:

VAR. BI, or

 

where:

and where variable name codes are the same as for monthly and annual SVF files (Section 8.2.1).

 

8.3 Creation of Climate Variables

The VEMAP dataset includes daily, monthly, and annual climate data for the conterminous U.S. including maximum, minimum, and mean temperature, precipitation, solar radiation, and humidity. Seasonal mean surface wind speed is also provided. The monthly, seasonal, and annual data are long-term climatological means and are on the CDROM and FTP site under the subdirectory /monthly. Annual averages are simple means of the 12 monthly fluxes. The daily set presents a "characteristic year" in which monthly averages or accumulations of the daily values match the long-term monthly climatology but where the daily series has variances and covariances characteristic of a station's weather record. The daily data are on the CDROM and FTP site in the subdirectory /daily.

We used two processes to create the daily climate data (Kittel et al. 1995):

(1) statistical simulation of daily temperature and precipitation records, and
(2) empirical estimation of corresponding daily radiation and humidity records.

 

8.3.1 Temperature and Precipitation Records

In the first process, we generated one year of daily precipitation and maximum and minimum temperature for each VEMAP grid cell. These records were produced using a stochastic daily weather generator, WGEN (Richardson 1981, Richardson and Wright 1984), which we modified to better utilize temporal statistics created by its accompanying parameterization program, WGENPAR. Parameterization of WGEN was based on daily records from 870 stations. WGEN was run for each grid cell with parameters assigned from the closest station. Climate records created by WGEN have realistic daily variances and temporal autocorrelations (e.g., persistence of wet and dry days) and maintain physical relationships between daily precipitation and temperature. For example, in the WGEN records, days with precipitation tend to have lower maximum temperatures than days with no precipitation.

To obtain the one year daily series, we first produced a 20-year weather record using WGEN. From this 20-year record, we derived the VEMAP characteristic year by choosing 12 individual months whose monthly means most closely matched the corresponding long-term historical monthly means (e.g., January from year 5, February from year 2, etc.). Daily values of the selected months were adjusted so that their monthly sum (for precipitation) or mean (for temperature) exactly matched the historical long-term monthly means.

As the final step in this process, we determined the absolute maximum and minimum temperatures and their month of occurrence for the characteristic year (c_atmax, c_mtmax, c_atmin, c_mtmin). We also saved "record" absolute maximum and minimum temperatures and their month of occurrence (r_atmax, r_mtmax, r_atmin, r_mtmin) from the full 20-year WGEN simulation which includes interannual variation about the long-term mean.

 

8.3.2 Solar Radiation and Humidity Records

We used CLIMSIM (Running et al. 1987) to generate daily records of solar radiation and surface air humidity from daily maximum and minimum temperatures and precipitation. We produced 6 solar radiation variables: total incident solar radiation at the surface (sr), sr as a fraction of potential total solar radiation at the top of the atmosphere (fsr) and at the surface (fsr_sfc), potential total solar radiation at the top of the atmosphere (psr) and at the surface (psr_sfc), and mean daily irradiance at the surface (irr). Humidity variables generated were vapor pressure (vp) and mean daylight relative humidity (rh). Because of biases in the method used in CLIMSIM to generate humidities from daily minimum temperature (Kimball et al. 1996), daily vapor pressure values were adjusted so that monthly means match the long-term means of Marks (1990). More details on this adjustment are given in Section 8.7.1. Monthly means of solar radiation and humidity variables were created from the daily CLIMSIM output. Because the solar radiation and humidity data are based on temperatures and precipitation that are constrained to match their long-term means and because the humidity data are additionally constrained by the Marks (1990) means, monthly means of the solar radiation and humidity dailies are taken to represent the climatological means of these variables.

For radiation variables, monthly and annual files contain either averages or totals of daily values. To distinguish between these, refer to units and file descriptions (e.g., "Average monthly file" vs. "Total monthly file") in Section 8.6.

 

8.4 Maximum, Minimum, and Mean Temperature [deg.C]

8.4.1 Maximum, Minimum, and Mean Temperature (VAR = tx, tn tm) [deg.C]

Long-term monthly mean daily maximum and minimum temperatures were interpolated to the VEMAP grid from 4613 station 1961-1990 normals (NCDC 1992, dataset TD-9641). Station values were adiabatically lowered to sea level (Marks and Dozier 1992), interpolated to the 0.5deg. VEMAP grid, and then re-adjusted to the new grid elevation. Mean temperatures were computed as a simple average of the gridded maximum and minimum monthly temperatures. We then generated daily maximum and minimum temperatures for each grid point, as described in Section 8.3.1. Daily temperatures were constrained in the generation process so that their monthly means matched the interpolated long-term monthly normals. Daily mean temperatures are not provided.

Daily binary files: tx.BI, tn.BI

Average monthly SVF files: tx.MMM, tn.MMM, tm.MMM
Average annual SVF files: tx.ann, tn.ann, tm.ann
Scaling factor: 10.0

8.4.2 Record Absolute Maximum and Minimum Temperature (VAR = r_atmax, r_atmin) [deg.C]

Absolute daily maximum and minimum temperature in the 20-yr WGEN record.

SVF file: r_atmax, r_atmin
Scaling factor: 10.0

 

8.4.3 Month of Occurrence of Record Absolute Maximum and Minimum Temperature (VAR = r_mtmax, r_mtmin) [month id: 1-12]

The month of occurrence of absolute maximum and minimum temperature in the 20-yr WGEN record. Month identifier runs from 1 to 12, corresponding to months January through December.

SVF file: r_mtmax, r_mtmin
Scaling factor: 1.0

8.4.4 Characteristic Year Absolute Maximum and Minimum Temperature (VAR = c_atmax, c_atmin) [deg.C]

Absolute maximum and minimum temperature found in the VEMAP characteristic year.

SVF file: c_atmax, c_atmin
Scaling factor: 10.0

8.4.5 Month of Occurrence of Characteristic Year Absolute Maximum and Minimum Temperature (VAR = c_mtmax, c_mtmin) [month id: 1-12]

The month of occurrence of absolute maximum and minimum temperature found in the VEMAP characteristic year. Month identifier runs from 1 to 12, corresponding to months January through December.

SVF format: c_mtmax, c_mtmin
Scaling factor: 1.0

8.5 Precipitation (VAR=p) [mm/day, month, or year]

Long-term mean monthly precipitation was spatially aggregated from a 10-km gridded U.S. dataset developed using PRISM by Daly et al. (1994). PRISM models precipitation distribution by (1) dividing the terrain into topographic facets of similar aspect, (2) developing precipitation-elevation regressions for each facet type for a given region based on station data, and (3) using these regressions to spatially extrapolate station precipitation to 10-km cells that are on similar facets.

We generated daily precipitation for each grid point using WGEN, as described in Section 8.3. Daily values were constrained such that monthly rainfall accumulations for each grid point matched the long-term monthly means.

Note: Units and scaling factors differ for daily, monthly, and annual files.

Daily binary files: p.BI [mm/day]
Scaling factor: 10.0

Total monthly SVF files: p.MMM [mm/month]
Total annual SVF file: p.ann [mm/year]
Scaling factor: 1.0

 

8.6 Solar Radiation

8.6.1 Relationship Among Solar Radiation Variables

Six solar radiation variables are included in the climate dataset (Table 6). These variables are either measures of solar radiation inputs at the top of the atmosphere (psr) and the surface (psr_sfc, sr, and irr) or of cloud and total transmissivity (fsr_sfc and fsr, respectively). Relationships among these variables on a daily basis are illustrated in Fig. 2 and are as follows.

(1) Potential total incident solar radiation at the surface (psr_sfc) is the potential at the top of the atmosphere (psr) reduced by clear sky effects on transmissivity, such that:

psr_sfc = psr x (clear sky transmissivity)

(2) Total incident solar radiation at the surface (sr) is derived from potential solar radiation at the top of the atmosphere (psr) diminished by total atmospheric (clear sky and cloud) effects on transmissivity (fsr, ranging from 0 to 1), so that:

sr = psr x fsr

(3) Total incident solar radiation at the surface (sr) is also related to potential at the surface (psr_sfc) (which accounts for only clear sky effects on transmissivity), by further reducing psr_sfc by cloud effects:

sr = psr_sfc x fsr_sfc

where fsr_sfc is cloud transmissivity (0 - 1).

(4) Atmospheric transmissivity variables are related to each other, such that total atmospheric transmissivity (fsr) is the product of cloud (fsr_sfc) and clear sky transmissivities:

fsr = fsr_sfc x (clear sky transmissivity)

(5) Daily mean surface irradiance for daylight hours (irr) is derived from sr and day length, such that, with unit conversion:

irr = sr x (1 day/day length) x (1000J/1kJ)

where day length is in seconds.

Note that because radiation variables were determined on a daily basis, these relationships do not precisely hold for monthly averages or accumulations (see notes in Sections 8.6.5 and 8.6.6).

8.6.2 Total Incident Solar Radiation (VAR = sr) [kJ m^-2 day^-1 or kJ m^-2 yr^-1]

Total incident solar radiation at the surface. Generated by CLIMSIM, sr is based on daily potential solar radiation at the top of the atmosphere (psr) and an estimate of daily atmospheric transmissivity (reported in this dataset as "fraction potential total solar radiation", fsr), such that:

We report sr as daily (sr.BI) and monthly (sr.MMM) average values, and as an annual summation of daily values (sr.ann).

Daily binary files: sr.BI [kJ m^-2 day^-1]
Average monthly SVF files: sr.MMM [kJ m^-2 day^-1]
Total annual SVF file: sr.ann [kJ m^-2 yr^-1]
Scaling factor: 1.0

 

8.6.3 Surface Total Solar Radiation as Fraction of Top of Atmosphere Potential Total Solar Radiation (VAR = fsr) [fraction, 0-1]

Ratio of total incident solar radiation at the surface (sr) to potential total solar radiation at the top of the atmosphere (psr), or total atmospheric transmissivity. CLIMSIM generates fsr as an estimate of atmospheric transmissivity (reported as "trans" in CLIMSIM). In CLIMSIM, atmospheric transmissivity is estimated first from clear sky transmissivity, which is a function of elevation. Clear sky transmissivity is then diminished by a surrogate for cloudiness, based on the occurrence of precipitation and the diurnal temperature range using the method of Bristow and Campbell (1984). Daily temperatures and precipitation used in these calculations are from the WGEN-generated record (tx.BI, tn.BI, p.BI).

We report fsr as daily (fsr.BI) and monthly (fsr.MMM) average values, and as an average of the 12 monthly mean values (fsr.ann).

Daily binary files: fsrday.BI
Average monthly SVF files: fsr.MMM
Average annual SVF file: fsr.ann
Scaling factor: 1000.0

8.6.4 Surface Total Solar Radiation as Fraction of Surface Potential Total Solar Radiation (VAR_sfc = fsr_sfc) [fraction, 0-1]

Ratio of total incident solar radiation at the surface (sr) to potential solar radiation at the surface (psr_sfc), or cloud transmissivity. Because psr_sfc already accounts for clear sky transmissivity, fsr_sfc represents a further reduction in transmissivity due to cloud cover. (See discussion of transmissivity calculations in the subsection on fsr, Section 8.6.3.)

Therefore, fsr_sfc can be used as a surrogate for percent possible hours of sunshine or for (1 - % cloudiness). However, these 3 variables are not strictly the same. Percent hours of sunshine is determined at meteorological stations by a sunshine switch, and percent cloudiness by hourly observations of fractional cloud cover.

We report fsr_sfc as daily (fsr_sfc.BI) and monthly (fsr_sfc.MMM) average values, and as an annual average of the monthly means (fsr_sfc.ann).

Daily binary files: fsr_sfc.BI
Average monthly SVF files: fsr_sfc.MMM
Average annual SVF file: fsr_sfc.ann
Scaling factor: 1000.0

8.6.5 Potential Total Solar Radiation at the Top of the Atmosphere (VAR = psr)

[kJ m^-2 day ^-1, mo^-1, or yr^-1]

Monthly and annual accumulated potential total incident solar radiation generated by CLIMSIM. Potential total solar radiation is based on latitude and solar geometry using the method outlined by Gates (1981).

We report psr as daily (psr.BI), monthly (psr.MMM), and annual (psr.ann) accumulations of daily values.

Note: Because psr, fsr, and sr were determined on a daily basis, it is not possible to reproduce the monthly sr value based on monthly accumulated psr and mean monthly fsr values (i.e., [sr.MMM] x [days/month] != [psr.MMM] x [fsr.MMM]).

Daily binary files: psr.BI [kJ m^-2 day^-1]
Total monthly SVF files: psr.MMM [kJ m^-2 mo^-1]
Total annual SVF file: psr.ann [kJ m^-2 yr^-1]
Scaling factor: 0.01

8.6.6 Potential Total Solar Radiation at the Surface (VAR_sfc = psr_sfc)
[kJ m^-2 day ^-1]

Monthly and annual averaged daily potential total incident solar radiation at the surface generated by CLIMSIM. Potential total solar radiation at the top of the atmosphere (based on latitude and solar geometry, Gates 1981) is modified by clear sky transmissivity to estimate potential solar radiation at the surface.

We report psr_sfc as daily values (psr_sfc.BI) and monthly (psr_sfc.MMM) and annual (psr_sfc.ann) averages.

Note: Because psr_sfc, fsr_sfc, and sr were determined on a daily basis, it is not possible to reproduce the monthly sr value based on the mean monthly psr_sfc and the mean monthly fsr_sfc values (i.e., [sr.MMM] != [psr_sfc.MMM] x [fsr_sfc.MMM]).

Daily binary files: psr_sfc.BI [kJ m^-2 day^-1]
Average monthly SVF files: psr_sfc.MMM [kJ m^-2 day^-1]
Average annual SVF file: psr_sfc.ann [kJ m^-2 day^-1]
Scaling factor: 0.01

8.6.7 Daily Mean Irradiance (VAR = irr) [W m^-2]

Daily mean irradiance for daylight hours, derived from CLIMSIM calculations of total incident solar radiation (sr.BI) and day length.

Daily binary files: irr.BI [W/m²]
Average monthly SVF files: irr.MMM [W/m²]
Average annual SVF file: irr.ann [W/m²]
Scaling factor: 100.0

8.7 Humidity

8.7.1 Vapor Pressure (VAR = vp) [mb]

Vapor pressures were generated by CLIMSIM using WGEN-produced daily minimum temperature. CLIMSIM estimates surface air humidity by assuming that dew point temperature is equal to daily minimum temperature.

To account for arid regions where the minimum temperature may not be an adequate estimate of dew point temperature, we modified vapor pressure and relative humidity values to more closely match long-term monthly means calculated by Marks (1990) (Kittel et al. 1995). If the Marks vapor pressure was less than CLIMSIM monthly mean vapor pressure, daily vapor pressures were adjusted by the corresponding monthly ratio:

If the Marks vapor pressure was equal to or higher than CLIMSIM (ratio >= 1.0), no adjustment was made to daily vp and rh. New monthly mean vapor pressures were calculated from the adjusted values.

Daily binary files: vp.BI
Average monthly SVF files: vp.MMM
Average annual SVF file: vp.ann
Scaling factor: 100.0

8.7.2 Mean Daylight Relative Humidity (VAR = rh) [%]

Generated by CLIMSIM with WGEN-generated temperature input (see Section 8.7.1). The mean is for daylight hours, as CLIMSIM calculates relative humidity relative to the saturated vapor pressure for a computed daylight-period temperature mean. If daily vapor pressures were adjusted (see Section 8.7.1), relative humidities were modified accordingly.

Daily binary files: rh.BI
Average monthly SVF files: rh.MMM
Average annual SVF file: rh.ann
Scaling factor: 10.0

8.8 Surface Wind Speed (VAR=w) [m/s]

Grid-averaged seasonal wind speed at 10-meter height. These data are based on a 10-km EPA dataset (Marks 1990), which is in turn based on DOE seasonal (3-month) mean wind speeds with some topographic adjustment (Elliott et al. 1986). Wind speeds reported here in monthly files are the same within each season (e.g., winter = January, February, March).

Average monthly SVF files: w.MMM
Average annual SVF file: w.ann
Scaling factor: 10.0

---

9 SOILS

9.1 Summary of Soil Variables

The soils dataset includes 18 variables (Table 7). These are described in more detail in Sections 9.3 - 9.5. For most variables, soil data are provided for 2 layers:

(1) 0 -> 50 cm
(2) 50 -> 150 cm

Relationships among area variables (ma, oa, map, tap) are presented in Appendix 2.1. On the CDROM and FTP site, soil data can be found in the /soil subdirectory.

Table 7. Soil variables. Variable name codes and layer codes (L) are those used in filenames
(Section 9.2).

*layer ID code, L: 1 = 0 - 50 cm, 2 = 50 - 150 cm
**moc is not available for layer 2

 

9.2 Soil Filename Protocol

File names for modal soils follow the form:

CVARL_mM

and for average soils:

CVARL_ave

where:

9.3 Creation of the VEMAP Soils Dataset

9.3.1 Source Data

Soil properties were based on a 10-km gridded EPA soil database developed by Kern (1994, 1995). Two soil coverages are provided in the Kern dataset: one from the USDA Soil Conservation Service (SCS) national soil database (NATSGO) and the other from the United Nations Food and Agriculture Organization soil database (FAO 1974-78). Only the SCS NATSGO soils are included in the VEMAP set.

 

9.3.2 Modal Representation of Cell Soil Data

Physical consistency in soils data was incorporated by representing a grid cell's soil by a set of dominant (modal) soil profiles, rather than by a simple average of soil properties. Because soil processes, such as soil organic matter turnover and water balance, are non-linearly related to soil texture and other soil parameters, simulations based on dominant soil profiles and their frequency distribution can account for soil dynamics that would be lost if averaged soil properties were used.

To spatially aggregate Kern data to the 0.5deg. grid, we used cluster analysis to group the subgrid 10-km elements into up to 4 modal soil categories (Kittel et al. 1995). In this statistical approach, cell soil properties are represented by the set of modal soils, rather than by an "average soil." We also provide cell-averaged soil data.

See Appendix 2 for determination of the absolute area represented by the entire cell (or by each modal soil within a cell) and the application of these quantities to model results. These areal values are included in the database as variables ma and oa (Section 9.5.4) and varea (Section 7.3.6).

 

9.3.3 Missing Code

When a soil mode is not present, the cell value is set to a stored value of -98 (i.e., -98 is not scaled by the scaling factor). If a mode is absent, no additional soil profiles are present for the cell from that mode level on. For example, if mode 3 is not present, then all the soil information for that cell is contained in the previous modes (modes 1 and 2), and mode 4 will also be absent.

 

9.4 Hierarchical Division of Soils into Mineral and Organic Components and Texture Classes

Structure of the VEMAP soil dataset follows the hierarchical division of a cell's soil in the Kern sets. In the Kern SCS NATSGO database, each soil type is represented by 2 component soils: a mineral soil (C=m) and an organic soil (C=o), each with its own profile of soil properties. Both mineral and organic soils are further differentiated into rock fragments and finer elements. Rock fractions are presented as a percentage of the entire soil volume for each of these component soils. VEMAP Phase I simulations used the mineral soil component of mode 1 soils.

The finer elements of the mineral soil are defined texturally in terms of mineral (msa, msi, mcl) and organic (moc) content. Percent organic matter (by weight) is relative to the combined mineral and organic fractions of the mineral soil. Percent by weight of sand, silt, and clay are relative to the mineral portion only.

Values are averages for each modal (or cell average) soil. Percent sand, silt, and clay add up to 100% (+/- 1% due to rounding error).

 

9.5 Soil Files

9.5.1 Modes per Cell (VAR = modes) [units = number of modes]

Number of modal soils per cell. Number of modes range from 1 to 4.

Gridded SVF files:modes
Scaling factor: 1.0

 

9.5.2 Mineral Soil Percent Areal Coverage within a Given Modal (or Average) Soil

(CVAR = map) [% of modal soil area, or % of area of all modal soils]

Relative area covered by mineral soils as percent of the total area covered by a given modal soil (map_mM) or for all soils in a cell (map_ave). Note that the area covered by organic soils equals (1 - map) for the corresponding modal soil or cell average.

Gridded SVF files:

 

9.5.3 Modal Soil Percent Areal Coverage (CVAR = tap) [% of cell land area]

Relative area covered by modal soil M as percent of area covered by land (and within U.S. borders) for each 0.5deg. grid cell. Includes both mineral and organic components.

Gridded SVF files:tap_mM
Scaling factor:1.0

 

9.5.4 Absolute Areal Coverage of a Modal (or Average) Mineral or Organic Soil

(CVAR = ma, oa) [km²]

Areal coverage of the mineral (ma) or organic (oa) component soil in a cell for either a modal (_mM) or average (_ave) profile. (See Appendix 2.1 for calculation of this variable.)

Gridded SVF files:

 

9.5.5 Bulk Density (CVAR = mbd) [g/cm³]

Bulk density of the mineral soil component for layer L and soil mode M (or cell average).

Gridded SVF files:

 

9.5.6 Soil Depth (CVAR = mz, oz) [cm]

Soil depth for mineral and organic soils for soil mode M (or cell average).

Gridded SVF files:

 

9.5.7 Texture (CVAR = msa, msi, mcl, moc) [% by weight]

Percent sand, silt, and clay of mineral portion of mineral soil and percent organic content of entire mineral soil (see Section 9.4) for layer L and soil mode M (or cell average).

Note: This coverage is for mineral soils only.

sand/silt/clay -

organic matter - (for layer 1 only)

 

9.5.8 Rock Fragments (CVAR = mrf, orf) [% by volume]

Rock fragments for mineral and organic soils for layer L and soil mode M (or cell average).

Gridded SVF files:

 

9.5.9 Water Holding Capacity (CVAR = mwh, owh) [cm H₂O per unit area]

For mineral soils, Kern (1995) provides water holding capacity (WHC) based on Rawls et al. (1982) and Saxton et al. (1986). We used the Rawls et al. WHC for layer 1, because this method utilizes organic matter content, and the Saxton et al. WHC for layer 2, where organic content information is not present (Saxton et al. calculations are based only on % sand/silt/clay). For WHC of organic soils, Kern used Paivanen (1973) and Boelter (1969). There was an error in the original Kern WHC values for layer 1. This is corrected as per Kern (1996).

Gridded SVF files:

Soil organic carbon (SOC) for 0-20 cm and 0-100 cm layers for modal and average soils. Calculated from mean SOC values in the Kern EPA soil database, based on SCS NATSGO data. SOC is for both mineral and organic soil combined. SCS data are for current SOC levels, including for agricultural soils where present. Kern's SOC values are adjusted for rock fragment content and actual soil depth. SOC 0-20 cm was derived using mean SOC values for 4 soil layers in the Kern EPA database by (1) integrating between 15 and 20 cm along a spline function that was fit to values for 0-8, 8-15, 15-30, and 30-75 cm layers and (2) adding the integrated value to the sum of 0-8 and 8-15 cm SOC.

Gridded SVF files:

---

10 VEGETATION

10.1 Summary of Vegetation Variables

The vegetation dataset includes one variable: vegetation type (Table 8). This coverage is of potential natural vegetation under current conditions (see Section 10.2). We include the original coverage used in VEMAP Phase I simulations (vveg.v1), as well as a slightly modified version (vveg.v2). Vegetation files can be found in the subdirectory /geog on the CDROM and FTP site.

Table 8. Vegetation variable name code and description.

Table 9. VEMAP vegetation types: vveg identifying code and corresponding VEMAP vegetation type. Where type description differs between vveg versions, the version is identified in parentheses.