 |
|
CWF |
| Caribbean Regional Node | ||
Information | Data | Software | Sites | Education | Feedback | News
| |
|
CWF |
| Caribbean Regional Node | ||
Information | Data | Software | Sites | Education | Feedback | News
IntroductionThe CoastWatch binary file format, traditionally referred to as IMGMAP or "image map", is composed of two main parts -- the image header followed by an image data stream. The image header contains information about the data dimensions, channel number, time, location, calibration, etc. The image data stream contains the data and graphics pixel values, possibly in a compressed form. The data compression uses an algorithm that takes advantage of small data differences between adjacent pixels, where as the graphics are compressed with a simple run length encoding scheme. The process of reading visible or thermal data from a CWF file usually includes the following steps:
Image Header
- read header values to determine data size and type
- read and uncompress data values
- read and uncompress graphics values
- convert data values from integer to floating-point
The image header has a variable size, depending on the data compression and size of the image. In compressed files (header word 39 below), the header is always 1024 bytes long. In uncompressed files the header is two times the number of data columns in length, so for example an uncompressed file with 1200 data columns will have a header size of 2400 bytes.
The table below gives the header value locations and codes. Note that each header value is a two-byte word (16-bit) integer value.
Word 0 Satellite designator (two EBCDIC characters) Byte 1 = N Byte 2 = B (NOAA-6) C (NOAA-7) D (NOAA-8) E (NOAA-9) F (NOAA-10) G (NOAA-11) H (NOAA-12) J (NOAA-14) K (NOAA-15) L (NOAA-16) M (NOAA-17) ... etc Word 1 Satellite ID 0 = morning satellite (AM) 1 = afternoon satellite (PM) Word 2 Data set type 1 = LAC 2 = GAC 3 = HRPT Word 3 Projection type 0 = unmapped 1 = mercator 2 = polar stereographic 3 = linear lat/lon Word 4 Beginning latitude of image area * 128 (N = +, S = -) Word 5 Ending latitude of image area * 128 (N = +, S = -) Word 6 Beginning longitude of image area * 128 (E = +, W = -) Word 7 Ending longitude of image area * 128 (E = +, W = -) Word 8 Image map resolution * 100 kilometers/pixel (mercator, polar) degrees/pixel (linear) sampling interval (unmapped) Word 9 Spare Word 10 Spare Word 11 Polar grid size (polar projection) Word 12 Grid size (polar projection) Word 13 Image hemisphere 1 = northern hemisphere -1 = southern hemisphere Word 14 Prime longitude (E = +, W = -) (polar projection) Word 15 Horizontal image coordinate of NW corner of image (I offset) (not applicable to unmapped or linear) Word 16 Vertical image coordinate of NW corner of image (J offset) (not applicable to unmapped or linear) Word 17 Number of image columns (not applicable to unmapped) Word 18 Number of image rows (not applicable to unmapped) Word 19 Spare Word 20 Spare Word 21 Number of records written to unmapped image files Word 22 Calibration flag 0 = no calibration 1 = calibrated to albedos and temperatures 2 = calibrated to albedos and GOES counts Word 23 Fill option 0 = no fill 1 = fill with average pixel values 2 = fill with adjacent pixel values Word 24 Data type 1 through 5 = AVHRR channel number 6 = MCSST 7 = NDVI 101 = scan angle 102 = satellite zenith angle 103 = solar zenith angle 104 = relative azimuth angle 105 = scan time 201 = MCSST split window eqn 202 = MCSST dual window eqn 203 = MCSST triple window eqn 204 = CPSST split window eqn 205 = CPSST dual window eqn 206 = CPSST triple window eqn 207 = NLSST split window eqn 208 = NLSST dual window eqn 209 = NLSST triple window eqn 301 = Ocean reflectance 302 = Turbidity 401 = Cloud mask Word 25 Data ID 0 = visible 1 = IR 2 = ancillary 3 = cloud mask 4 = graphics Word 26 Sun normalization flag 0 = not performed 1 = performed Word 27 Limb correction flag 0 = not performed 1 = performed Word 28 Non-linearity correction flag 0 = not performed 1 = performed Word 29 Number of orbits processed (for composited images) Word 30 Number of channel images written to output file Word 31 Size of image pixel Word 32 Starting block number of image data Word 33 Ending block number of image data Word 34 Number of ancillary images written to output file Word 35 Size of ancillary pixel Word 36 Starting block number of ancillary data Word 37 Ending block number of ancillary data Word 38 Block size of output image files Word 39 Image file compression flag 0 = uncompressed file 2 = compressed file Word 40 ID of MCSST equation used to compute SST 0 = no sst 1 = NOAA-9 Day Split Eqn. 2 = NOAA-9 Night Split Eqn. 3 = NOAA-9 Night Triple Eqn. 4 = NOAA-10 Day Eqn. 5 = NOAA-10 Night Eqn. 6 = NOAA-11 Day Split Eqn. 7 = NOAA-11 Night Split Eqn. 8 = NOAA-11 Night Dual Eqn. 9 = NOAA-11 Night Triple Eqn. Word 41 Percentage of non-zero pixels Word 42 Horizontal shift offset Word 43 Vertical shift offset Word 44 Horizontal skew offset Word 45 Vertical skew offset Word 46 Spare Word 47 Spare Word 48 Spare Word 49 Spare Word 50 Orbital node over target area -1 = ascending node 1 = descending node 2 = both Word 51 Day/night flag 0 = day 1 = night Word 52 Image data start row Word 53 Image data start column Word 54 Image data end row Word 55 Image data end column Word 56 Start year of orbit 1 Word 57 Start Julian day of orbit 1 Word 58 Start month and date of orbit 1 (MMDD format) Word 59 Start hour and minute of orbit 1 (HHMM format) Word 60 Start seconds of orbit 1 Word 61 Start milliseconds of orbit 1 Word 62 End year of orbit 1 Word 63 End Julian day of orbit 1 Word 64 End month and date of orbit 1 (MMDD format) Word 65 End hour and minute of orbit 1 (HHMM format) Word 66 End seconds of orbit 1 Word 67 End milliseconds of orbit 1 Word 68 Processing block ID (Orbit number) Word 69 Ramp/Auto calibration of orbit 1 Word 70 Number of data gaps in orbit 1 Word 71 Sync errors in orbit 1 Word 72 TIP parity errors in orbit 1 Word 73 Auxiliary errors in orbit 1 Word 74 Calibration parameters ID of orbit 1 Word 75 DACS status of orbit 1 Word 76 Slope of channel 1 calibration (*10000) Word 77 Intercept of channel 1 calibration (*10000) Word 78 Slope of channel 2 calibration (*10000) Word 79 Intercept of channel 2 calibration (*10000) Word 80 Spare Word 81 Spare Word 82 Spare Word 83-115 Repetition of word 50-82 for orbit 2 Word 116-148 Repetition of word 50-82 for orbit 3, etc.Image DataThe format of the image data will depend on the data ID (header word 25) and compression type (header word 39). For ancillary data including angles and scan times, the data stream is written uncompressed as rows*cols 16-bit integer values. The integer values may be directly converted to ancillary values using standard conversions (see below). Similarly for cloud mask files, the data stream is written uncompressed as rows*cols byte values. Each cloud mask bit value represents a different cloud mask test, with 1 indicating the presence of cloud, and 0 indicating clear ocean. Visible and infrared data may be written either in compressed or uncompressed format. For visible and infrared data in uncompressed format, the image data consists of rows*cols 16-bit values arranged as follows:
byte 1 2 nibble 1 2 1 2 |-------| |-------| SIII IIII IIII GGGG bit 7654 3210 7654 3210 S: sign bit (always 0) I: image data bit G: graphics bitIn the case of a compressed data stream, the image data values are compressed and written first, then the graphics are compressed and written. Image data compression is based on the difference between previous and next pixels. The compressed image data is written as a string of bytes, according to the following scheme:
if ((first pixel of image) or (abs (this pixel - last pixel) > 63)) { write two bytes arranged as: byte 1, nibble 1 = 1000 byte 1, nibble 2 = same as byte 1, nibble 1 in uncompressed byte 2, nibble 1 = same as byte 1, nibble 2 in uncompressed byte 2, nibble 2 = same as byte 2, nibble 1 in uncompressed } else { write one byte arranged as: bit 8 = 0 bit 7 is the sign bit (0 = plus, 1 = minus) bits 6 through 0 = abs (this pixel - last pixel) }The graphics compression algorithm uses run length encoding. The graphics are specified as byte pairs, where the first byte is the value of the graphics pixel and the second byte is the number of consecutive pixels (from 0 to 255) of that pixel. The encoding algorithm is as follows:
n = 1 run length = 0 last pixel = graphics at pixel 1 while (n < rows*cols) { this pixel = graphics at pixel n if (this pixel = last pixel) { if (run length = 255) { write two bytes arranged as: byte 1 = this pixel byte 2 = run length run length = 0 } else { run length = run length + 1 } } else { write two bytes arranged as: byte 1 = last pixel byte 2 = run length last pixel = this pixel run length = 0 } n = n + 1 }In order to recover the actual floating-point data values from the 11-bit integer data values in either compressed or uncompressed format, the following conversions are needed (header word 25):Visible: albedo = pixel_value / 20.47 Infrared: if (0 < pixel_value <= 920) temp = (pixel_value - 1)*0.1 + 178.0 if (921 <= pixel_value <= 1720) temp = (pixel_value - 921)*0.05 + 270.0 if (1721 <= pixel_value <= 2047) temp = (pixel_value - 1721)*0.1 + 310.0 Ancillary: angle = pixel_value / 128 hour = pixel_value / 100 hour = hour + (pixel_value - hour) / 60
 |
Back to main page | USDOC | NOAA | NESDIS | CoastWatch |