The Channel 1 and Channel 2 values are original count values truncated to 8-bit precision. The Channel 4 and Channel 5 values are then converted to "GOES counts" (intended originally for use with geostationary satellites). The GOES counts are 8-bit binary numbers corresponding to temperature values according to a standard look-up table (Appendix B). The solar zenith angle and the scan angle values are scaled binary values to the nearest half of a degree. It should be noted that the solar zenith angle in the GVI increases with time. This is because the equator crossing time of the polar orbiter becomes later and later in the afternoon over the life of the spacecraft. Thus, the polar orbiters are not truly sun-synchronous.

The Channel 4 and Channel 5 binary counts are converted to "GOES" counts before they are mapped. This involves using the channel calibration coefficients to convert the binary counts to energy values, then using the inverse of Planck's radiation formula, convert the energy values to absolute temperatures, and finally scaling the temperature values to 8-bit binary values to be placed in one byte as GOES counts. In practice, the computation is performed once for each possible data value, for each of the two IR channels, to create a set of look-up tables which will be used for the actual data processing.

The basic composite arrays, each of which have 2500 by 904 elements, include the six items listed above plus a Normalized Difference Vegetation Index array (NDVI) (see Section 3.1.2.1 for definition).

Each week, these seven basic composite arrays are used to produce the Mercator maps and the Polar Stereographic arrays.

The resolution selected for the Plate Carrée arrays is 16 km at the equator, which results in the 2500 x 904 array to map the region between 75N and 55S.

All seven weekly composite Plate Carrée arrays are converted to the corresponding Mercator arrays; each Mercator array consisting of 2048 x 1038 elements, corresponds to a resolution of 19.5 km at the equator.

All seven weekly composite Plate Carrée arrays are converted to Polar Stereographic arrays. Each hemisphere is a 1024 by 1024 element array, consistent with the Polar Stereographic First Generation GVI product. The Polar Stereographic arrays have a better resolution of 13 km at the equator compared to the corresponding resolution of 16 km for the master Plate Carrée arrays. To avoid "holes" in the Polar Stereographic arrays near the equator (within 15 degrees latitude north or south), the mapping process fetches a pixel from the Plate Carrée arrays for each cell in the Polar Stereographic array, which results in repetition of some pixels (where otherwise holes would have occurred). As a result of the latitude limits (75N to 55S), selected to include significant land areas but also to minimize processing, Polar Stereographic arrays produced from the Plate Carrée arrays have conspicuous "holes" around the poles.

Beginning July 9, 1990, the Calibrated Vegetation Index (CVI) product was generated and added to the Plate Carrée tape as file #9. This product is run at the end of the normal seven day series of GVI processing. The weekly composited Channel 1 and Channel 2 maps are converted to albedos to compute the CVI. The Calibrated Vegetation Index is defined by the equation:

where A1 and A2 are the albedos which are defined as:

where S1, S2, I1, and I2 are the calibration coefficients (slope and intercept) shown in Table 4.1-1, and C1 and C2 are the 8-bit counts of the Channel 1 and Channel 2 maps. The slopes must be multiplied by four to give the correct albedo since the counts were truncated to 8-bit numbers.

The CVI is then scaled according to the following criteria:

Then, the scaled CVI is truncated to an 8-bit integer. The scaled CVI is only present on the weekly Plate Carrée tape.

Table 4.1-1. Pre-launch slopes and intercepts for AVHRR Channels 1 and 2.
Satellite	S1	I1	S2	I2
NOAA-7	0.1068	-3.4400	0.1069	-3.488
NOAA-9	0.1063	-3.8464	0.1075	-3.8770
NOAA-11	0.0906	-3.730	0.0900	-3.390
NOAA-14*	0.1081	-3.8648	0.1090	-3.6749
*Beginning in November 1996, the slopes and intercepts for NOAA-14 AVHRR Channels 1 and 2 were computed monthly, incorporated into the Level 1b datasets and posted on the ORA Home page (URL: http://orbit-net.nesdis.noaa.gov/ora).

4.2 PLATE CARRÉE PROJECTION

Several map projections were considered for the daily master arrays: Mercator, Equal Area, and Plate Carrée. Plate Carrée was selected because it is the simplest for data processing, and it provides a reasonable representation of the problematic polar regions, where both the Mercator and the Equal Area projections would be more distorted.

The Plate Carrée projection is not conformal, but has the following simple form:

where and are the longitude and the latitude, respectively, measured in degrees, while is the central longitude. The value for is 0 degrees, corresponding to the Greenwich Meridian, which will be at the center of the array, with the edges of the array being 180 degrees East and West longitude. IN is the number of non-duplicated pixels in the array in the East-West direction. IN is set to 2500, corresponding to a resolution of 16 km at the equator. The I and J values on the map are given by:

I = X + 1250

J = -Y + 522

where I ranges from 1 to 2500, and J ranges from 1 to 904, covering the latitude range of 75N to -55S.

Longitudes are measured from 0 degrees to +180 degrees eastward and from 0 degrees to -180 degrees westward, which is the same convention used to store the values within the GAC data sets.

4.3 DAILY PROCESSING

This section describes the daily processing of GAC data sets and updating of the basic composite

4.3.1 PROCESSING OF AVHRR GAC DATA SETS

A given GAC data set is read record by record, and for each record of daytime data one scan line is processed. The Ch1, Ch2, and Ch4 data are extracted and mapped to the corresponding master daily arrays and held in computer memory. The Ch5, SZA, and SCA data are held in a temporary disk file. Once the entire master arrays are written to disk, the Ch5, SZA, and SCA arrays are read in, and updated from the temporary file. When this process is completed for a GAC data set, another is processed, until the full day's worth of GAC data sets have been processed.

4.3.1.1 Processing of GAC Scan Lines

The first of two scan lines packed in each GAC record is examined. If the scan line contains night time data, or if earth location or calibration information is missing, the scan line is skipped. If the latitudes are outside the range of 75N through 55S, the scan line is skipped. For a good scan line, the latitude and longitude values for each of the 409 data points are interpolated from the 51 benchmark latitude/longitude pairs. These earth locations are then converted into 409 pairs of (I,J) values on the Plate Carrée array, indexing the positions to which each GAC pixel will be mapped. Then the Solar Zenith (SZA) angles are interpolated for the 409 data points from the 51 benchmark values.

The five channel count values for each data point are then unpacked from the video data and truncated to 8-bit precision. The Ch4 and Ch5 count values are converted to GOES counts using a pre-computed look-up table (refer to Appendix B). This processing yields 10 bytes of information to be mapped for each GAC data point:

These data are then used to update the daily master arrays as outlined above. Since the resolution of the GAC data (4 km) is much finer than the resolution of the Plate Carrée arrays, a number of GAC pixels will be mapped to each location in the Plate Carrée array. Each value will replace the one previously written. Thus, a selection of one GAC sample is made to represent the population of GAC samples to be mapped to that location. Because the selection does not depend upon the value of the sample, this selection process is regarded as random (as opposed to basing the selection on the value of the Ch2-Ch1 difference, as was experimented with for a time while the First Generation GVI product was being produced).

4.3.1.2 Scan Angle Processing

The scan angles for one orbital swath range from -55.4 to +55.4 degrees relative to nadir, a total span of 110.8 degrees. Scan angles to the nearest half-degree corresponding to the 409 sample positions on a scan line are calculated as a binary count of half-degree steps from 0, for sample 1; to 222, for sample 409. The nadir point will therefore correspond to 110.8 half-degree steps, rounded to 111.

The scan angle SCA in half-degree steps for the Nth sample position is given by:

where 1 < N < 409. The scan angle values are stored in one byte as binary integers.

4.3.1.3 IR Temperature Look-up Table Generation

Look-up tables are used to convert Ch4 and Ch5 values to GOES counts. The first step is the conversion of the possible range of count values to radiances, using the Channel 4 and Channel 5 calibration coefficients present in the GAC scan line:

where E is the channel radiance (energy) in mW/(m²-sr-cm^-1); M is the slope parameter for the channel; N is the raw channel count; and S is the intercept parameter for the channel.

In turn, the radiances are converted to absolute temperatures, by the following equation:

where T_abs is the absolute temperature in Kelvin (K); A = 1.438833 cm-K; B = 1.1910659 x 10^-5 mW/(m²-sr-cm^-4); is the central wave number which can be found for Channels 4 and 5 for each satellite in Table 4.3.1.3-1.

Table 4.3.1.3-1. Central wave numbers for Channels 4 and 5.
Satellite	Channel 4 (cm-1)	Channel 5 (cm-1)
NOAA-7	927.22	840.872
NOAA-9	929.46	845.19
NOAA-11	927.83	842.20
NOAA-14	929.3323	835.1647
These central wave numbers are valid for the temperature range: 275-320 K.

The temperatures are then rescaled to GOES count values (so they will fall in the range from 0 to 255) using the following equation:

where GOES is the GOES count value, and T_abs is the absolute temperature. Constants C and D have the values C = -1.006412 and D = 419.05128 for temperatures in the range 164K < T_abs < 242K (which are in one degree steps); and C = -2.0057142 and D = 661.88571 for temperatures in the range 242K < T_abs < 330K (which are in half degree steps).

The results are two look-up tables of 255 GOES count values: a Channel 4 table and a Channel 5 table. Appendix B contains a single GOES count to absolute temperature conversion table that may be used to recover the absolute temperature values.

4.3.2 UPDATING THE BASIC COMPOSITE ARRAYS

Each day the seven basic composite Plate Carrée arrays are updated from the six new daily master arrays. Updating of the composites is performed cell by cell. The difference of the Channel 2 and Channel 1 values (Ch2-Ch1) for the new data is computed, and compared to the Ch2-Ch1 value for the corresponding element of the basic composite array. If the new difference is greater (i.e., if the cell is "greener" or clearer) then the NDVI value for the new data point is computed and replaces the NDVI value in the basic composite array, and the values in the other six basic composite arrays for that cell location are replaced by the corresponding six values from the new daily master arrays. Thus, compositing is bypassed if the new cell is less "green" than the corresponding cell in the basic composite arrays. Compositing is also bypassed if the difference value for the new cell is 255. If compositing is bypassed, the cell values in all seven basic composite arrays are left unchanged.

When a new data point is to be composited (i.e. when its value replaces the previous cell value in the basic composite arrays), its NDVI value is computed from the data point's Ch1 and Ch2 values from the new daily master array, as follows:

where XVI is an unscaled NDVI value, and Ch1 and Ch2 are the Channel 1 and Channel 2 count values, respectively. Thus, NDVI is computed as follows:

The values of XVI range from -.05 (no green) to +.60 (greenest). These values are then scaled to an 8-bit binary integer, so that the greenest areas appear the darkest when an image is produced. Thus, a scaled NDVI value of 240 (white) corresponds to an unscaled XVI value of -.05, and a scaled NDVI value of 12 (black) corresponds to an unscaled value of +.60.

The process described above is performed for every data point (cell) in the daily master arrays.

4.3.3 WEEKLY COMPOSITE ARRAYS

Once a week, the seven basic composite arrays (which are in the Plate Carrée projection) are used to produce a set of Polar Stereographic and Mercator arrays. The mapping is done by sequentially filling each cell location in the Polar Stereographic or Mercator arrays with a value from a location on the Plate Carrée arrays to ensure that there are no "holes" (i.e. cell locations that do not receive values) resulting from the difference in resolution of the new arrays compared to the Plate Carrée arrays.

4.3.3.1 Weekly Mercator Arrays

Seven Mercator arrays are produced from the corresponding basic composite Plate Carrée arrays. Each array is 2048 x 1038 cells in size, affording a resolution of 19.5 km at the equator.

For the Mercator arrays, the location of each cell on the Plate Carrée arrays (specified by I and J) is computed from the location of the cell on the Mercator arrays (specified by IM and JM) by the following equations:

where IM ranges from 1 to 2048 (with 0 and 2048 being equivalent), and JM ranges from 1 to 1038.

4.3.3.2 Weekly Polar Stereographic Arrays

Seven hemispheric pairs of Polar Stereographic arrays are produced from the corresponding basic composite Plate Carrée arrays. Each hemispheric array is 1024 x 1024 cells in size, affording a resolution of 13 km at the equator. Since the Plate Carrée arrays contain data from only 75N through 55S, the Polar Stereographic arrays will contain large holes around each pole.

The longitude and latitude of element I,J on the Plate Carrée array is given by the following:

the longitude,, is measured in degrees from the Greenwich Meridian, +180E to -180W. The latitude,, is measured in degrees from the equator, +90N to -90S.

Computation of the I, J location of a point on the Polar Stereographic arrays, given its latitude and longitude, is as follows:

The full earth is included in two hemispheric 1024 by 1024 arrays treated as a single 1024 x 2048 array during the mapping process. IPN and JPN are coordinates within the Northern hemisphere array, while IPS and JPS are coordinates within the Southern hemisphere array. IPN and JPN both range from 1 to 1024, with IPN = 0 and IPN = 1024 being equivalent. IPS also ranges from 1 to 1024, while JPS ranges from 1025 to 2048. , the prime longitude, is set to -80 degrees (80W) to center the map on the United States.

Because the Polar Stereographic arrays have a higher resolution than the Plate Carrée arrays (13 km at the equator compared to 16 km), "holes" (locations to which no cells are mapped) in the Polar Stereographic arrays would result from mapping cells from the Plate Carrée arrays into the Polar Stereographic arrays. Thus, the inverse is done in practice; the latitude and longitude of each cell in the Polar Stereographic array is computed, the corresponding I,J location in the Plate Carrée array computed, and the data value at that cell location in the Plate Carrée array is copied into the Polar Stereographic array.

For the Polar Stereographic projection, the latitude and longitude,and,of a given element (I,J) is computed by the following:

where JOF = 512 for the Northern hemisphere and JOF = 1536 for the Southern hemisphere.

4.4 DIGITAL GVI PRODUCTS

Daily and weekly composite digital products are available. The daily products are the six Plate Carrée master arrays, and the weekly products are seven composite Plate Carrée arrays (eight arrays after July 9, 1990), seven pairs of Polar Stereographic arrays, and seven Mercator arrays.

4.4.1 DAILY GVI PRODUCTS

Table 4.4.1-1 contains the file structure of the daily master array tapes (a physical file is defined as one or more physical records followed by an end-of-file mark):

Table 4.4.1-1. File structure of daily master array tapes.
File #	Contents
1	Documentation File
2	Channel 1 Plate Carrée array
3	Channel 2 Plate Carrée array
4	Channel 4 Plate Carrée array
5	Channel 5 Plate Carrée array
6	Solar Zenith Angle (SZA) Plate Carrée array
7	Scan Angle (SCA) Plate Carrée array

File 1, the Documentation File, contains a single record, which is 5000 bytes long. It documents the number of AVHRR GAC data sets used in the production of the day's master arrays, and identifies which GAC data sets were used. The format of the documentation record is contained in Table 4.4.1-2.

Table 4.4.1-2. Format of the documentation file for daily GVI Products.
Bytes	# Bytes	Contents
1-5	5	2-digit year and day of year (YYDDD) in ASCII
6	1	Number of GAC data sets (N) processed, in binary
7-11	5	Date processed (YYDDD), in ASCII
12	1	Unused, blank filled
13 - x	36N	Data set name of those GAC data sets used to create this day's file: 33 bytes of ASCII information and 3 bytes blank fill for each N data sets, where N is usually 14, occasionally 15, and less than 14 if some data were missing.
X - 4096	n/a	Blank filled.

In Table 4.4.1-1, the N groups of 36 bytes contain 33 bytes of data set information with 3 bytes of blank-fill trailing, in ASCII. Each 33-byte group contains a truncated form of a GAC data set name which was used to compute the daily GVI. The data set name consists of a set of alphanumeric qualifiers separated by periods (.) in the following format:

Spacecraft-ID.Year-Day.Start-time.Stop-time.Processing-block-ID.Source

Table 4.4.1-3 contains the data set name qualifiers.

Table 4.4.1-3. Data set name qualifiers.
Qualifier	Example
DATA-TYPE	Valid groups are: GHRR= GAC (recorded reduced resolution AVHRR)
SPACECRAFT-UNIQUE-ID	NOAA-C = NC = NOAA-7 NOAA-D = ND = NOAA-12 NOAA-F = NF = NOAA-9 NOAA-H = NH = NOAA-11 NOAA-J = NJ = NOAA-14
YEAR-DAY	D76104, where "D" identifies this group as a Julian day delimiter, "76" identifies the year in which the spacecraft began recording the data set and "104" identifies the Julian day on which the spacecraft began recording the data set.
START-TIME	S1355, where "S" identifies this group as a start time delimiter. "1355" denotes 13 hours 55 minutes UTC (to the nearest minute) and represents the time at which spacecraft recording began.
STOP-TIME	E1456, where "E" identifies this group as an end time delimiter. "1456" denotes 14 hours 56 minutes UTC (to the nearest minute) and represents the time of spacecraft recording of the last usable data in the data set.
PROCESSING-BLOCK-ID	B0016465, where "B" identifies this group as a processing block ID delimiter. "0016465" is a seven digit number identifying the spacecraft revolution in which recording of this data set began and the revolution in which the data was transmitted to ground (the first five digits identifying the beginning revolution and last two being the two least significant digits of the orbit number identifying the readout revolution). However, NESDIS does not necessarily guarantee that the Processing-Block-ID contains the correct beginning and ending orbit number. Frequently (especially with LAC data), the orbit numbers are 1 to 2 off the correct orbit; thus, it is always prudent when ordering data to include a time, if known.
SOURCE	Valid character groups are: Fairbanks, Alaska (formerly Gilmore Creek) = GC Western Europe CDA = WE SOCC (Satellite Operations Control Center) = SO Wallops Island, Virginia = WI

Files 2 through 7 each contain a Plate Carrée array, 2500 (West-East) cells by 904 (North-South) cells in size, each cell being an 8-bit byte. Files 2 through 7 each contain 452 physical records, each 5,000 bytes long. Each physical record contains two logical records, each 2,500 bytes long. Each logical record contains one scan of cells across the Plate Carrée array. Byte 1 of logical record 1 contains the upper left (North-West) corner of the array.

4.4.2 WEEKLY COMPOSITE GVI PRODUCTS

The weekly composite GVI product tapes contain the seven files listed above for the daily master array tape with the addition of an eighth file containing scaled NDVI data. Since July 9, 1990, a ninth file containing scaled Calibrated Vegetation Index (CVI) data has been included on the weekly Plate Carrée tape. Three types of weekly tapes are produced: a Plate Carrée tape, a Polar Stereographic tape, and a Mercator tape. In each case, the contents of File 1 is a single documentation record. The number of days worth of data used to produce the composite, and identification of their dates, is also included. The format of the documentation record is contained in Table 4.4.2-1.

Table 4.4.2-1. Format of the Documentation record for the Weekly Composite GVI products.
Bytes	# Bytes	Contents
1	1	Number of days processed for the composite, in binary
2	1	Unused, blank filled.
3-8	6	First day processed (YYDDDb) b=blank, in ASCII
9-14	6	Second day processed, as above.
...	...	...
39-44	6	Seventh day processed, as above.
45-4096	4052	Blank filled.

In Table 4.4.2-1, YYDDDb indicates the date as the last two digits of the year (YY), day of year (DDD), and one blank character (b).

Files 2 through 9 of each type of tape contain data arrays (in the same order as the daily master array tape with the addition of the NDVI file as file #8 and CVI as file #9) formatted as follows.

4.4.2.1 Plate Carrée Tape

The format of files two through eight (nine since July 9, 1990) of the Plate Carrée weekly composite product tape is the same as the daily master array tape format described in Section 4.4.1, with the addition of the NDVI file as file #8, and since July 9, 1990, the CVI file as file #9.

4.4.2.2 Polar Stereographic Tape

The Polar Stereographic array size is 1024 x 2048 elements. Each file contains 512 physical records, each 4096 bytes long. Each physical record contains four logical records, each 1024 bytes long. Each logical record contains one scan across the Polar Stereographic array. The first group of 1024 scans (or logical records) contain the Northern Hemisphere, and the second group of 1024 scans (or logical records) contain the Southern Hemisphere.

The first byte in logical record 1 contains the upper left corner of the Northern Hemisphere portion of the array, while byte 1 of logical record 1025 (in physical record 257) contains the upper left corner of the Southern hemisphere portion of the array.

4.4.2.3 Mercator Tape

The Mercator array size is 2048 (West-East) elements by 1038 (North-South) elements. Each file contains 1038 physical records, each 2048 bytes long. Each physical record contains one logical record which is one scan across the Mercator array. The first byte of record 1 is the upper left (North-West) corner of the Mercator array.

4.4.3 WEEKLY COMPOSITE GVI PRODUCT - STACKED

The weekly composite GVI product tapes are stacked for each of the three map projections. Approximately five weeks of stacked weekly composite data are contained on each IBM 3480 cartridge. Refer to Appendix F for the composition of each cartridge. The format is identical to the normal weekly composite GVI product tapes described in Section 4.4.2.

4.4.4 WEEKLY COMPOSITE NDVI PRODUCT - STACKED

Another form of weekly NDVI product is available to the user: the stacked product. This product offers only the Documentation and scaled NDVI files stacked on IBM 3480 cartridges for each type of map projection. Appendix H lists the composition and SSB tape numbers for each type of stacked weekly composite NDVI tape. Plate Carrée and Polar Stereographic stacked NDVI tapes can contain as much as 50 weeks of data, while Mercator stacked NDVI tapes hold approximately 35 weeks of data.

4.4.5 CONTINENTAL WEEKLY COMPOSITE DATA SET

In 1994, NOAA/NESDIS completed development of the Continental Weekly Composite data set. This data set covers more than six years of weekly composite mapped time series data over the land masses of six continents. There are 403 weeks of weekly composite data, covering the period from April 20, 1985 to December 29, 1992. Each composite consists of AVHRR Channels 1, 2, 4 and 5, and the corresponding solar zenith (SZA) and scan angles (SCA) for the day of the week which exhibited maximum differences between channels 1 and 2.

To conserve space, the GVI data were further compressed by retaining one map cell out of a 2 x 2 array of map cells, giving a final resolution of 32 km. From this data, NOAA/NESDIS constructed the Continental Weekly Composite data set. Table 4.4.5-1 shows the geographical areas covered by this time series on a continent by continent basis.

Table 4.4.5-1. Geographical areas covered by Continental Weekly composite data set.
Continent	Areal Coverage	Number of 32 km map cells
North America	Lat: 75N - 15N Lon: 168W - 50W	28,011
South America	Lat: 15N - 55S Lon: 168W - 30W	14,808
Europe	Lat: 75N - 25N Lon: 50W - 60E	24,906
Africa	Lat: 39.5N - 55S Lon: 30W - 60E	28,683
Asia	Lat: 75N - 5N Lon: 60E - 168W	41,820
Australia	Lat: 5N - 55S Lon: 70E - 168W	9,729

Data for each continent were generated on an IBM mainframe computer and stored on separate non-labeled IBM 3480 cartridges. North America, Africa and Asia were stored on two cartridges each because of the large data volumes. Table 4.4.5-2 gives a breakdown of the continental time series by cartridge, while Table 4.4.5-3 shows the general structure of the cartridges.

Table 4.4.5-2. Contents of Continental Weekly Composite cartridges.
Tape #	Continent	# Weeks	Starting-ending weeks
1	North America	399	1-399
2	North America	4	400-403
3	South America	403	1-403
4	Europe	403	1-403
5	Africa	391	1-391
6	Africa	12	392-403
7	Asia	247	1-247
8	Asia	156	248-403
9	Australia	403	1-403

Table 4.4.5-3. General structure of Continental Weekly composite cartridges.
File #	Record #	Contents
1	1	200-byte EBCDIC header (see Table 4.4.5-4).
2	1	6,354-byte mixed mode record containing all the time series data by week for land map cell #1 (see Table 4.4.5-5).
2	2	Same as File 2, Record 1 for map cell #2.
...	...	...
2	m	Same as File 2, Record 1 for map cell #m.

Each cartridge contains two files. The first file has a 200-byte EBCDIC header which is described in Table 4.4.5-4.

Table 4.4.5-4. Contents of header file on Continental Weekly Composite data set (EBCDIC).
Bytes	Contents
1-6	Continental identification
7-11	Northern-most latitude (+N, -S)
12-16	Southern-most latitude
17-21	Western-most longitude (+E, -W)
22-26	Eastern-most longitude
27-31	Starting day of year in time series
32-36	Ending day of year in time series
37-39	Number of weeks in time series
40-47	Channel 1 slope for NOAA-9
48-55	Channel 1 intercept for NOAA-9
56-63	Channel 2 slope for NOAA-9
64-71	Channel 2 intercept for NOAA-9
72-79	Channel 1 slope for NOAA-11
80-87	Channel 1 intercept for NOAA-11
88-95	Channel 2 slope for NOAA-11
96-103	Channel 2 intercept for NOAA-11
104-195	Blank filled
196-200	Number of land map cells for the continent

The second file consists of a series of m records corresponding to m map cells. Each 6,354-byte record contains the time series data arranged by week for each map cell. This file is in mixed mode, meaning that it contains both EBCDIC characters and binary (8-bit) data values. Table 4.4.5-5 contains the format of the records within the second file. The second through last week n of every map cell contains the 2-digit year, day of year, satellite ID and data. Table 4.4.5-5 shows the format of the second file.

Table 4.4.5-5. Format of the second file in the Continental Weekly composite data set.
Variable	# Bytes	Bytes	Data type
Latitude (first week only)	6	1-6	EBCDIC
Longitude (first week only)	6	7-12	EBCDIC
Year and day of year (Week #1)	5	13-17	EBCDIC
Satellite ID for Week #1 (SATID=9 for NOAA-9, =11 for NOAA-11)	1	18	Binary
Data (Ch. 1, Ch. 2, Ch. 4, Ch. 5, SZA, SCA) for map cell #1, Week #1	1 each	19-24	Binary
Year and day of year (Week #2)	5	25-29	EBCDIC
SATID for Week #2	1	30	Binary
Data for map cell #1, Week #2	6	31-36	Binary
...	...	...	...
Year and day of year (Week #n)	5	12n+1 thru 12n+5	EBCDIC
SATID for Week #n	1	12n + 6	Binary
Data for map cell #1, Week #n	6	12n+7 thru 12n+12	Binary

The data stored on tape was designed to be read using an IBM mainframe computer. The file "tape.jcl" contains actual IBM Job Control Language (JCL) and should be used as the prototype for extracting the data from tape. The basic idea of file "tape.jcl" is to use a system routine called IEBGENER to extract the data from tape and copy it to disk. The amount of primary and secondary allocations of space that IEBGENER requires to copy the data to disk varies from continent to continent. The larger time series, Asia and Africa, could take upwards of 3500-4000 tracks of space! A copy of the IBM JCL code (tape.jcl) can be found in Appendix I.

To convert the one-byte image data (of the mixed-mode records) to IBM formatted data for browsing, the program "calibrd.for" may be used. A copy of the FORTRAN code and documentation for "calibrd.for" can also be found in Appendix I.

4.5 HARDCOPY IMAGE PRODUCTS

Hardcopy images of the Second Generation GVI products are no longer available due to funding and manpower limitations.