This section describes the data characteristics and magnetic tape format of Global Area Coverage (GAC) data. Section 3.1.1 contains the data characteristics and Section 3.1.2 contains the tape formats available. The tape formats include the full data set copy, the 16-bit unpacked format, and the selective extract subsets. NESDIS has enhanced the Level 1b format for GAC and the most recent format (for data collected after November 15, 1994) is included in this section. Previous formats are included in Appendices K and L.

Data Characteristics

The processor on board the satellite samples the real-time AVHRR data to produce reduced resolution GAC data. Four out of every five samples along the scan line are used to compute one average value, and the data from only every third scan line are processed. As a result, the spatial resolution of GAC data near the subpoint is actually 1.1 km by 4 km with a 3 km gap between pixels across the scan line, although generally treated as 4 km resolution. All of the GAC data computed during a complete pass are recorded on board the satellite for transmission to Earth on command. The 10-bit precision of the AVHRR data is retained.

Magnetic Tape Formats

The data set format for full data set copies (all channels) is different from the format for selective extract subsets (selected channels). Sections 3.1.2.1 and 3.1.2.2 contain formats for GAC full data set copies and GAC selective extract subsets, respectively. Section 3.1.2.1 also includes an explanation of the 16-bit unpacked data format.

Full Data Set Copies

Each GAC data set contains an individual satellite recorder playback (or a portion of a playback if there is an interruption in the data due to noise, etc., in which case a single playback may be fragmented into a number of data sets). Data within each GAC data set is in chronological order with one logical record per scan. Two logical records are written per 6440-byte physical record. Each logical record contains 3220 bytes written in binary format as shown in Table 3.1.2.1-1. This table contains the format of the GAC data record which was implemented on November 15, 1994.

The time code consists of the year, Julian day, and UTC time of day in milliseconds. The year is contained in the first 7 bits of the first two bytes, the 9-bit day of year is right-justified in the first two bytes and the 27-bit millisecond UTC time of day is right-justified in the last four bytes. All other bits are zero. The time code will always have the same format for all Level 1b data sets.

The quality indicators are contained in four bytes. The first byte contains processing detected conditions and the last three bytes contain DACS quality indicators. If the bit is one (on), then the condition is true. The format of the quality bytes is contained in Table 3.1.2.1-2.

The calibration coefficients consist of slope and intercept values for each of the five channels. The use of these coefficients is described in Section 3.3. Each value is stored in four bytes in the following order:

Channel 1 slope coefficient
Channel 1 intercept coefficient
Channel 2 slope coefficient
Channel 2 intercept coefficient
Channel 3 slope coefficient
Channel 3 intercept coefficient
Channel 4 slope coefficient
Channel 4 intercept coefficient
Channel 5 slope coefficient
Channel 5 intercept coefficient

The telemetry data contain information which may be used to compute calibration coefficients when these are not included in the data. The telemetry data are stored in 140 bytes. The first 103 (10 bit) words are packed three (10 bit) words in four bytes, right justified. The last four byte group contains one (10 bit) word with 20 trailing bits. All unused bits are set to zero. The contents of these 103 words are contained in Table 3.1.2.1-4, which is the entire HRPT minor frame format. For more information, refer to NOAA Technical Memorandum NESS 107 entitled, Data Extraction and Calibration of TIROS-N/NOAA Radiometers.

The GAC video data consist of five readings (one for each channel) for each of the 409 points in a scan. They are packed as three (10-bit) samples in four bytes, right-justified. The last four-byte group contains two (10-bit) samples with 10 trailing zero bits. The first two bits of each four-byte group are zero. The 2045 samples (409 points x 5 channels) are ordered scan point 1 (Channel 1, 2, 3, 4, 5), scan point 2 (Channel 1, 2, 3, 4, 5), etc., which is also known as Band Interleaved by Pixel (BIP). Note for TIROS-N, NOAA-6, NOAA-8, and NOAA-10, there is no sensor for Channel 5 so Channel 4 data is repeated in the Channel 5 position. The video data are stored in binary. See Figure 2.2.1-1 for the arrangement of the GAC channels (packed format).

Enhancements implemented on September 7, 1994, included clock drift corrections and orbit parameters format change from IBM real numbers to scaled integers. These enhancements were made to both the TOVS and AVHRR Level 1b data. However, the new system did not function properly for AVHRR and was removed the same day with the old process resumed. Problems were encountered with the data record time codes in the GAC Level 1b data and NESDIS was forced to remove the updates from the AVHRR processing. Initially, NESDIS tried turning off the clock drift corrections, but time codes were still incorrect so the old process was resumed. The AVHRR data were processed with the old on-line earth location software until updates could be reinstalled. The following list of orbits were affected for the AVHRR:

On November 15, 1994, NESDIS re-implemented the following changes to the Level 1b GAC datastream: 1) Current orbital parameters in the header were switched to four byte integers rather than 8 byte floating point numbers. 2) Calibration algorithms were updated to remove the effects of sunlight impinging on the internal calibration target (Solar Blackbody Contamination - SBBC). SBBC detection indicators were added to the AVHRR data record quality indicators. 3) Navigation software was updated to remove satellite clock errors from the data time codes. 4) Navigation quality control parameters were provided in the header record.

An anomalous dip occurs in the AVHRR Channel 3 blackbody counts between 70 and 80S latitude shortly after the spacecraft comes out of eclipse but only for sun angles less than 25 degrees above the horizon. This dip, which produces an anomalous increase in the calibration gain coefficient, has been attributed to reflected sunlight impinging onto the internal calibration target (ICT). Evidence suggests that this dip may have occurred in all of the afternoon satellites in the current NOAA series.

When SBBC is detected, the slope and intercept are replaced by the running average of the twenty-four previous good values. If no checking is done, the value is set to zero. Currently, only NOAA-11 shows this contamination. On Dec. 1, 1993, AVHRR gain anomaly corrections were implemented for NOAA-11 beginning with dataset:
NSS.GHRR.NH.D93335.S1408.E1554.B2673132.GC. However, for 24 hours, the following datasets for NOAA-12 were affected:
NSS.GHRR.ND.D93335.S1303.E1457.B1324445.GC through
NSS.GHRR.ND.D93336.S0535.E0729.B1325455.WI.

The drift in the onboard TIP clock for each polar satellite introduces a timing error in the instrument data time codes. This error is reflected as an along-track Earth location error. SOCC measures this error and corrects it periodically to maintain a maximum of approximately ±0.5 seconds. Under Phase II of the navigation enhancements, SOCC error corrections will be used to adjust the instrument time codes and eliminate its Earth location error effect. The adjusted values will be stored in the scan data record and an indicator flag will be set for users. Consequently, users are apprised of the magnitude of the adjustment and also have the flexibility of removing the correction, if needed.

This enhancement also results in improved navigation parameters. The operational quality control (QC) of Earth location data will be improved. A nadir Earth location tolerance value will be used to indicate when the Earth location data are beyond an acceptable range. An independent calculation of the latitude and longitude values at nadir will be compared to the operational values. This comparison will be performed approximately four times per orbit. If differences are beyond the acceptable tolerance range, a "BAD EARTH LOCATION" flag is set, and a message identifying the scan is generated. The Earth location tolerance range will be provided in the header record as an integer, scaled to 0.1 km (ranges from 0.1 to 25.5). If no checking is done, the value is set to zero.

See Table 2.0.4-2 for the current format of the level 1b header record which was implemented November 15, 1994.

The 16-bit unpacked format for full copy GAC data has the same format as the "packed" data described above except for the video data. The video data values for each channel are contained in the 10 least significant bits and the 6 most significant bits are zero filled. The record length is 4540 bytes. Table 3.1.2.1-3 contains the format for the unpacked GAC video data (for all five channels).

Selective Extract Subsets

The following sections describe the data set format for GAC data sets which have been produced using channel selection (16-bit format) or by choosing the 8-bit format offered by SAA. Section 3.1.2.2.1 describes the 16-bit format and Section 3.1.2.2.2 describes the 8-bit format. SSB is aware of a problem which affects the format of both the 8-bit and the 16-bit (unpacked) data for the Level 1b AVHRR. The problem involves the omission of 22 bytes of extra precision information containing the Solar Zenith angle and clock drift delta in each record. These bytes were added to the Level 1b data record (on October 24, 1992) after the extraction software was written, and were therefore not included. If users of the 8- or 16-bit format data want this extra precision data, they are only available if the user requests packed format output from the SAA's online archive of Level 1b data sets.

16-bit format

When channels are selected for GAC data, the format is identical to that described in Section 3.1.2.1 except for the GAC video data. In the GAC video data, the selected channels are packed in consecutive half words (16 bits). The data values for each channel selected are contained in the ten least significant bits and the remaining six most significant bits are zero filled. The GAC video data have the format shown in Table 3.1.2.2.1-1 when two channels are selected.

The total record length for two channel selective GAC would be 4168 bytes (2 x 2084, since there are 2 scans/record) with no spare bytes trailing. Similarly, if one channel was selected the record would contain 820 bytes (409 points/scan + 1 (to make a whole word) x 2 bytes/point) for the GAC video, which would make the entire record 2536 bytes long (2 x 1268).

8-bit format

When the 8-bit format is chosen (from SAA), the normal 10-bit video data are truncated to 8-bits. (The 2 least significant bits of data are dropped.) The format is identical to that described in Section 3.1.2.1 except for the GAC video data. In the GAC video data, the data are packed in consecutive bytes (four data points per 32-bit word). Because there are an odd number of data points (409) in the video data, the last three bytes are zero filled in order to pad the data to a full word boundary. The GAC video data have the format shown in Table 3.1.2.2.2-1.

The record length for one channel selected GAC in the 8-bit format would be 1720 bytes (2 x 860, since there are 2 scans/record) with the last three trailing bytes zero filled.

Similarly, a user may select channels and request them in the 8-bit format. The output physical record length information for GAC is included in Table 3.1.2.2.2-2.

Source: http://www.ncdc.noaa.gov