NSIDC global navigation NSIDC homeDataProjectsResearchCryosphereNewsSite map
Data section navigation NSIDC Data CatalogSearch the Data CatalogData Help CenterData ToolsData Features
NSIDC DAAC logo

Cold Land Processes Experiment (CLPX)

   Home  |   CLPX Abstract  |   Data  |   Submit Data  |  

CLPX MODIS Grids V0.3

Contents

Version History
Purpose
LRSA Grids
MODIS Data Products
MODIS Bands
Metadata and Attributes
Resampling Tools and Methods
File Naming Convention
Data Formats
Directory Structure
Sample Dataset
Sample Images

Version History

The current version of the CLPX MODIS Grids document is V0.3 released June 17, 2002.

V0.2 release June 1, 2002.

V0.1 released May 30, 2002.

V0.0 released May 20, 2002.

Purpose

The purpose of this document is to describe a plan for providing Moderate Resolution Imaging Spectroradiometer (MODIS) data for the Cold Land Processes Field Experiment (CLPX). These MODIS data are to be delivered in a variety of grids covering the Large-Regional Study Area (LRSA).

The intended audience for this document is anyone interested in using MODIS data in conjunction with CLPX. It may also prove useful for those providing data for CLPX from other satellite sensors or from airborne sensors, particularly for raster datasets that cover a large regional area

LRSA Grids

The Cold Land Processes Field Experiment (CLPX) Large-Regional Study Area (LRSA) is defined by a latitude-longitude bounding box as follows:
Upper Left: 42 N, 108.5 W
Lower Right: 38.5 N, 104 W
MODIS data will be resampled into two families of grids, Geographic (GEO) and Universal Tranverse Mercator (UTM), that cover the LRSA and span a range of resolutions. Each grid at a particular resolution is intended to be a relatively close match to a set of MODIS (or other sensor) products at a similar resolution.

Note that in the following descriptions, Upper Left refers to the latitude-longitude coordinates of the upper left corner of the upper left pixel, and Lower Right refers to the latitude-longitude coordinates of the lower right corner of the lower right pixel. When defined in this way, all grids in a family have the same Upper Left and Lower Right latitude-longitude coordinates.

Grid names marked with * indicate grids that will not be used for MODIS data but may be suitable for other sensors such as SSM/I or Landsat.

The formulas below for calculating Column, Row, Latitude, Longitude, X, and Y follow the ENVI convention of specifying the upper left corner of the upper left pixel in a grid as having a row value of 1 and a column value of 1.

Geographic (GEO) Grid

Upper Left: 42.05 N, 108.55 W
Lower Right: 38.45N 103.95 W
Datum: WGS84
Approximate Coverage: 4.6 degrees x 3.6 degrees =~ 400 km x 400 km
 
Name MODIS Products Resolution
arcseconds		 Resolution
degrees		 X
Resolution
meters
(appx.)		 Y
Resolution
meters
(appx.)		 Cols Rows
lrsa_geo720.0* None 720 0.2 16974 22239 23 18
lrsa_geo360.0* None 360 0.1 8487 11120 46 36
lrsa_geo180.0 MOD10C1
MOD10C2		 180 0.05 4243 5560 92 72
lrsa_geo030.0 MOD021KM
MOD03
MOD11_A1
MOD35_L2		 30 0.00833333 707 927 552 432
lrsa_geo015.0 MOD02HKM
MOD09A1
MOD10_L2
MOD10A1
MOD10A2		 15 0.00416667 354 463 1104 864
lrsa_geo007.5 MOD02QKM
MOD09Q1
MOD13Q1		 7.5 0.00208333 177 232 2208 1728
lrsa_geo001.0* None 1.0 0.000277778 23.6 30.9 16560 12960
lrsa_geo000.5* None 0.5 0.000138889 11.8 15.4 33120 25920

Row = (42.05 - Latitude) / Resolution_in_degrees + 1
Column = (Longitude + 108.55) / Resolution_in_degrees + 1

Latitude = 42.05 - (Row - 1) * Resolution_in_degrees
Longitude = (Column - 1) * Resolution_in_degrees - 108.55

Universal Transverse Mercator (UTM) Grids

Zone: 13
Semi-Major Axis of Ellipsoid:     6378137.000000 meters
Semi-Minor Axis of Ellipsoid:     6356752.314245 meters
Scale Factor at C. Meridian:      0.999600
Longitude of Central Meridian:     -105.000000 degrees
Upper Left XY: 175000.0, 4675000.0 = 42.159677085 N, 108.933826235 W
Lower Right XY: 600000.0, 4250000.0 = 38.392627781 N, 103.854898585 W
Datum: WGS84
Approximate Coverage: 5.1 degrees x 3.8 degrees =~ 425 km x 425 km
 
Name MODIS Products Resolution Cols Rows
lrsa_utm25000* None 25000 meters 17 17
lrsa_utm12500* None 12500 meters 34 34
lrsa_utm05000 MOD10C1
MOD10C2		 5000 meters 85 85
lrsa_utm01000 MOD021KM
MOD03
MOD11_A1
MOD35_L2		 1000 meters 425 425
lrsa_utm00500 MOD02HKM
MOD09A1
MOD10_L2
MOD10A1
MOD10A2		 500 meters 850 850
lrsa_utm00250 MOD02QKM
MOD09Q1
MOD13Q1		 250 meters 1700 1700
lrsa_utm00025* None 25 meters 17000 17000
lrsa_utm00012* None 12.5 meters 34000 34000

Row = (4675000.0 - Y) / Resolution_in_meters + 1
Column = (X - 175000.0) / Resolution_in_meters + 1

Y = 4675000.0 - (Row - 1) * Resolution_in_meters
X = 175000.0  +  (Column - 1) * Resolution_in_meters

MODIS Data Products

Data from the following MODIS data products will be provided covering the time period Feb 15 (day 046) through May 15 (day 135), a total of 90 days, for both study years (2002-2003). Some data products may be missing for certain time periods due to intermittent problems with the satellite, sensor, or ground systems. A comprehensive list of these data gaps will be documented in a future version of this document after the data products are resampled. Here is a tentative list of data gaps for 2002: Data will be resampled from the original HDF-EOS product files containing either swath data or Integerized Sinusoidal (ISIN) grid data. Usually one, but sometimes two, swath product files are required to produce a single corresponding LRSA grid file. Four ISIN gridded tile files are required to create each corresponding LRSA grid file.

In the table below, the Size column includes the total volume for both the LRSA GEO and UTM grids in both raw binary and GeoTiff formats (4 grids total). It is assumed that only daytime swath data will be used. The Frequency column (Freq.) indicates how often the corresponding LRSA grids will be produced

 
Name Description Level Source Source
meters/
pixel
(appx.)		 Bands Bytes/
Band
(avg.)		 Size
(MB)		 Freq.
MOD021KM Calibrated Radiances L1B Swath 1000 36 2.0 60.3 1/day
MOD02HKM Calibrated Radiances L1B Swath 500 7 2.0 46.9 1/day
MOD02QKM Calibrated Radiances L1B Swath 250 2 2.0 53.6 1/day
MOD03 Geolocation Fields L1A Swath 1000 8 1.75 11.7 1/day
MOD09A1 Surface Reflectance L2G ISIN Grid 500 13 2.15385 93.9 1/8 days
MOD09Q1 Surface Reflectance L2G ISIN Grid 250 3 2.0 80.5 1/8 days
MOD10_L2 Snow Cover L2 Swath 500 2 1.0 6.7 1/day
MOD10A1 Snow Cover L3 ISIN Grid 500 2 1.0 6.7 1/day
MOD10A2 Snow Cover L3 ISIN Grid 500 2 1.0 6.7 1/8 days
MOD10C1 Snow Cover L3 CMG 0.05
degrees		 4 1.0 0.1 1/day
MOD10C2 Snow Cover L3 CMG 0.05
degrees		 4 1.0 0.1 1/8
days
MOD11A1 Land Surface Temperature/Emissivity L3 ISIN Grid 1000 12 1.33333 13.4 1/day
MOD13Q1 Vegetation Indices L3 ISIN Grid 250 11 2.0 295.0 1/16 days
MOD35_L2 Cloud Mask and
Spectral Test Results		 L2 Swath 1000 14 1.0 11.7 1/day
The expected average daily volume (the sum of (Size*Freq.) above) is about 252.2 MB. The expected maximum daily volume (the sum of Size above) is about 687.3 MB. The above totals do not include Metadata and Attribute files (described below) which add a total of about 4 MB to the daily totals.

MODIS Bands

The following list shows the band names and data types included in each MODIS product for which grids will be produced. The band names are grouped according to the type of resampling that was done: Data types include the following:

Metadata and Attributes

In addition to gridded data, for each MODIS HDF-EOS product file, two ASCII text files will be provided containing MODIS metadata and attributes, respectively.

Metadata Files

Metadata files are created along with the HDF-EOS product files. The metadata file contains some of the same metadata as in the product file, but also includes other information regarding archiving, user support, and post production quality assurance (QA) relative to the granule ordered.

Attribute Files

Attribute files contain the following information extracted from the corresponding product file: Of particular importance are local attributes that pertain to individual data arrays in the HDF-EOS product files. Some of these local attributes include scale and offset values that can be used to convert the integer values contained in the corresponding grids into meaningful scientific quantities such as radiances, reflectances, and emissivities.

Resampling Tools and Methods

Two different tools are used to resample the data contained in the MODIS HDF-EOS files listed above into the LRSA grids.

The MODIS Swath-to-Grid Toolbox (MS2GT)

The MODIS Swath-to-Grid Toolbox (MS2GT) is used to resample MODIS swath data contained in HDF-EOS files into an intermediate Lambert Azimuthal Equal-Area grid stored in a raw binary format. Discretely valued data (e.g., Snow Cover or QC flags) are resampled using nearest neighbor resampling; continuously valued data (e.g., radiances or temperatures) are resampled using an elliptical weighted averaging technique.

Since the UTM projection and GeoTiff output are not yet supported in MS2GT, all MODIS swath data are first put into an Lambert Azimuthal Equal-Area grid (not documented here), and then run through MRT to produce the final output (see below). It is hoped that a future version of MS2GT or some other tool will allow the production of UTM gridded output and GeoTiff output from MODIS swath data so that resampling to this intermediate grid will no longer be performed.

The MODIS Reprojection Tool (MRT)

The MODIS Reprojection Tool (MRT) is used to resample gridded MODIS data into both the LRSA GEO and UTM grids. It is used to resample MODIS gridded data contained in HDF-EOS files into the raw binary and GeoTiff formats described below. It is also used to resample the Lambert Azimuthal Equal-Area gridded output of MS2GT into the LRSA GEO and UTM grids in both the raw binary and GeoTiff formats described below. Discretely valued data are resampled using nearest neighbor resampling; continuously valued data are resampled using cubic convolution resampling.

A third tool called ncdump is used to extract attributes from each MODIS HDF-EOS product file.

File Naming Convention

Files Containing Gridded Data

Files containing MODIS data resampled to the above grids have names in the following form:

<dataset>.a<year><day>[.<time>].<grid><resample>[.<band>].<extension>

where

<dataset> is the name (mapped to lowercase) of the MODIS dataset corresponding to the granule from which the data were extracted and (re)gridded, e.g., "mod09q1" or "mod10_l2".
<year> is the four digit year the data were acquired, e.g., "2000".
<day> is the three digit day of year the data were acquired, e.g., "052".
[.<time>] is the four digit UTC time of the start of acquisition for the five minute swath granule from which the data were gridded, e.g., "1725". For Level 3 datasets (i.e. daily and multiple day datasets), this field, including the leading period, is omitted.
<grid> is the name of the grid to which the data have been resampled, e.g., "lrsa_geo007.5".
<resample> is the resampling method and is one of:
"nn" for nearest neighbor resampling used in MRT.
"cc" for cubic convolution resampling used in MRT.
[.<band>] is the name of an individual band in the dataset, e.g., "sur_refl_b01" or "Snow_Cover". For header files (see below), this field, including the leading period, is omitted since the header contains information for all bands for a particular dataset granule for a particular resampling method.
<extension> is one of:
"dat" for raw binary data file.
"hdr" for an ASCII header file.
"tif" for a GeoTiff file.

Examples:

mod021km.a2002052.1725.lrsa_geo030.0cc.hdr
mod021km.a2002052.1725.lrsa_geo030.0cc.ch01.dat
mod021km.a2002052.1725.lrsa_geo030.0cc.ch01.tif

mod10_l2.a2002052.lrsa_utm00500nn.hdr
mod10_l2.a2002052.lrsa_utm00500nn.Snow_Cover.dat
mod10_l2.a2002052.lrsa_utm00500nn.Snow_Cover.tif

Files Containing MODIS Metadata

Files containing MODIS metadata are ASCII text files having names in the following form:

<granule>.met

where <granule> is the name of the MODIS HDF-EOS file described by the metadata file.

Examples:

MOD09Q1.A2002049.h09v04.003.2002062162758.hdf.met
MOD10_L2.A2002052.1725.003.2002055102744.hdf.met

Files Containing MODIS Attributes

Files containing MODIS attributes extracted from corresponding MODIS HDF-EOS product files are ASCII text files having names in the following form:

<granule>.atr

where <granule> is the name of the MODIS HDF-EOS file from which the attributes were extracted using the ncdump tool.

Examples:

MOD09Q1.A2002049.h09v04.003.2002062162758.hdf.atr
MOD10_L2.A2002052.1725.003.2002055102744.hdf.atr

Data Formats

Gridded data are provided in two formats: raw binary (including an ASCII header) and GeoTiff. Both formats are provided as output from the MODIS Reprojection Tool which is used to regrid MODIS gridded data.

Raw Binary Format

Raw binary data are stored in individual data files, with one file per band. Within each file, data are stored in row-major order, starting at the upper-left corner of the image. The data type may be 8-bit integer, 16-bit integer, 32-bit integer, or 32-bit float. Integer values may be signed or unsigned. Two-byte and four-byte data types are stored in big-endian order (high byte followed by low byte).

Header Format

The header file contains information produced by the MODIS Reprojection Tool for processing data in the raw binary file format. An ODL-like format is used, as illustrated by the following file named mod09q1.a2002049.lrsa_geo007.5cc.hdr for the 7.5 arcsecond GEO grid:

PROJECTION_TYPE = GEOGRAPHIC

PROJECTION_PARAMETERS = (
             0.000000000             0.000000000             0.000000000
             0.000000000             0.000000000             0.000000000
             0.000000000             0.000000000             0.000000000
             0.000000000             0.000000000             0.000000000
             0.000000000             0.000000000             0.000000000 )

# COORDINATE_ORIGIN = UL

UL_CORNER_LATLON = ( 42.050000000 -108.550000000 )
UR_CORNER_LATLON = ( 42.050000000 -103.950000000 )
LL_CORNER_LATLON = ( 38.450000000 -108.550000000 )
LR_CORNER_LATLON = ( 38.450000000 -103.950000000 )

# UL_CORNER_XY = ( -108.550000000 42.050000000 )
# UR_CORNER_XY = ( -103.950000000 42.050000000 )
# LL_CORNER_XY = ( -108.550000000 38.450000000 )
# LR_CORNER_XY = ( -103.950000000 38.450000000 )

NBANDS = 2
BANDNAMES = ( sur_refl_b01 sur_refl_b02 )
DATA_TYPE = ( INT16 INT16 )
NLINES = ( 1728 1728 )
NSAMPLES = ( 2208 2208 )
PIXEL_SIZE = ( 0.002083 0.002083 )
MIN_VALUE = ( -100 -100 )
MAX_VALUE = ( 16000 16000 )
BACKGROUND_FILL = ( -28672 -28672 )

DATUM = WGS84

The corresponding raw binary files corresponding to the above header file are named:
mod09q1.a2002049.lrsa_geo007.5cc.sur_refl_b01.dat
mod09q1.a2002049.lrsa_geo007.5cc.sur_refl_b02.dat
For a full description of the header file format, see Appendices B and C of the MODIS Tool User Guide v2.2.

GeoTiff Format

GeoTiff data are stored in individual data files, one file per band. Included within each file is metadata that allows GeoTiff-enabled software (such as ENVI) to determine the pixel width, byte-ordering, and geolocation of the data. See the GeoTiff home page for a further explanation of the GeoTiff data format.

Directory Structure

Data will be made available via ftp in a nested directory structure. The following list is intended to represent the proposed directory structure. Each element in the list represents a directory. The data for a particular type of data for a particular day are stored in the particular daily directory. There are 90 daily directories for each type of data. Depending on the availablility of the 8-day and 16-day MODIS products for a particular day, there are between 15 and 31 files in each METADATA and ATTRIBUTES daily directory, and between 93 and 129 files in each of the other daily directories.

MODIS_LRSA_2002
    METADATA
        2002046
        2002047
              .
              .
        2002135
    ATTRIBUTES
        2002046
        2002047
              .
              .
        2002135
    BINARY
        GEO
             2002046
             2002047
                   .
                   .
             2002135
       UTM
             2002046
             2002047
                   .
                   .
             2002135
    GEOTIFF
        GEO
              2002046
              2002047
                   .
                   .
             2002135
       UTM
             2002046
             2002047
                   .
                   .
             2002135

Sample Dataset

A sample dataset having the above directory structure is available for perusal and/or download at ftp://sidads.colorado.edu/pub/incoming/tharan/clpx/MODIS_LRSA_2002. Included are all 8-day and 16-day data for 2002049 (2002/02/18) and all 1-day data for 2002052 (2002/02/21). The 1-day data for 2002049 is not included, so the total sample represents the maximum amount of data expected for a single day. Total number of bytes for the entire sample dataset is about 693 MB.

Viewing Sample Data

Using a web browser to view sample data

The GeoTiff files in the sample dataset can be viewed using any web browser that has been enabled to read Tiff files using either a plugin or an image viewing application. Starting with the URL above, go to the GEOTIFF directory, select a grid family (e.g., UTM), select a day (e.g., 2002049), and then click on one of the tif files (e.g., mod09a1.a2002049.lrsa_utm00500cc.sur_refl_b01.tif). If you're on a Unix or Linux platform you shouldn't have any problems. However Windows users may not be able to view the two-byte-per-pixel files (such as mod09a1.a2002049.lrsa_utm00500cc.sur_refl_b01.tif) correctly. This is probably because the images were created on a big-endian machine and most freeware Windows image viewers apparently are not smart enough to interpret the big-endian tag in the Tiff file correctly. One-byte-per-pixel files (such as mod10a2.a2002049.lrsa_utm00500nn.Maximum_Snow_Extent.tif) should look ok on Windows platforms. On Macintosh platforms, the QuickTime viewer apparently can't view Tiff files at all.

None of the web-enabled image viewers that the author has used can geolocate the data in the GeoTiff files or read out individual pixel values.

Using a standalone program to view sample data

The best way the author has found to view the GeoTiff files in the sample dataset using a freeware program is to download the sample data files you wish to view to your computer using ftp, and then install a demo version of ENVI called Freelook. Of course, if you have ENVI, that's even better. All the above mentioned problems using web-enabled image viewers on Windows platforms go away, and you can even view the latitude-longitude and (for UTM grids) X-Y coordinates of any pixel together with the value of the pixel. However, Freelook is apparently not currently available on the Macintosh, despite what some of the documentation says.

If you want to download some sample data to view with a standalone program, and you don't want to download the entire sample dataset (about 693 MB), you might want to settle for just the 2002049 and 2002052 directories in GEOTIFF/UTM (about 149 MB).

Sample Images

The following four images were created with ENVI using data from the sample dataset. Each image shows 500 meter MODIS data resampled to the lrsa_utm00500 grid. UTM grid lines and lat-lon grid crosses have been drawn on the images. State and county boundaries and interstate highways have been overlaid as well.
  1. mod02hkm.a2002052.1725.lrsa_utm00500cc.ch143.jpg consists of MODIS channels 1, 4, and 3 mapped to red, green, and blue, respectively. It represents a portion of a single 5-minute swath of MOD02HKM Calibrated Radiances data acquired on Feb 21, 2002 (day 052) during day 3 of IOP-1 at about 10:25 MST, and resampled to the lrsa_utm00500 grid. A Gaussian stretch was applied to the image. The image is somewhat blurred due to the fact that the LRSA was considerably off nadir during this satellite pass, but the image is relatively cloud-free.
  2. mod10_l2.a2002052.1725.lrsa_utm00500nn.Snow_Cover.jpg consists of the MOD10_L2  Snow Cover band acquired at the same time as image #1. A linear 0-255 stretch was applied to the image and the ENVI "Volcano" palette was used to map the single byte Snow Cover coded values to colors. The colors are as follows:
    1. violet = 200 = snow
    2. blue = 50 = CloudObscured
    3. red = 100 = LakeIce
    4. green = 25 = NoSnow
  3. mod09a1.a2002049.lrsa_utm00500cc.sur_refl_b143.jpg consists of MODIS channels 1, 4, and 3 mapped to red, green, and blue respectively, just as in image #1 above; however this image is taken from MOD09A1 Surface Reflectance data and thus represents an eight-day composite of cloud-cleared reflectances starting on Feb 18, 2002 (day 049) through Feb 25 (day 56) so it includes all of IOP-1. Note that this image looks sharper than image #1. This is due in part to the use of data that were acquired closer to nadir.
  4. mod10a2.a2002049.lrsa_utm00500nn.Maximum_Snow_Extent.jpg consists of the MOD10A2 Maximum Snow Extent band for the same eight day period as in image #3 and using the same color coding as in image #2. Notice that there are fewer pixels marked as cloudy in this image compared to image #2 since it is less likely that a particular pixel would be truly cloudy for all eight days. However there are areas containing what is probably marginal snow (such as in North Park) that are marked as cloudy in this image. This is probably due to lingering problems in distinguishing marginal snow from clouds.
http://barrow.colorado.edu/~haran/

Last modified: June 17 15:20 MDT 2002