Open-File Report 98-484
Project Overview | Technical Information | Satellite Imagery | Contact Information | About the USGS | Order Information | Disclaimer | References | Acknowledgements
The imagery for the study area covers the period from June, 1993 to May, 1997. The set was produced from data collected by the Advanced Very High Resolution Radiometer (AVHRR) on the NOAA polar-orbiting environmental satellites, NOAA-11 (in 1993-1994), and NOAA-14 (in 1995-1997) with overpass times of 1500 and 1400 local standard time, respectively. The data sets were acquired from the Louisiana State University Earth Scan Laboratory and the University of South Florida/Florida Dept. of Environmental Protection Joint Use Facility. The satellites carry several sensors for meteorological studies. More details on the AVHRR and the satellites can be found in Kidwell (1991).
The NOAA polar-orbiters are in a sun-synchronous orbit, meaning that they will pass over a point at about the same time each day. The satellite has a nominal elevation of about 830 km and takes about 100 minutes to complete an orbit. The nadir point shifts by about 3 degrees of longitude each day. About every 9 days an area of interest is at nadir, centered in the satellite view area. Because of the movement in nadir position, during any nine-day period an area is within the scanned region for 6 days.
The AVHRR is a multiple-band scanning radiometer. The instrument scans one line at a time, with the movement of the satellite producing an offset between lines, resulting in the development of an image. The instantaneous field of view (IFOV) is 1.4 milliradians, corresponding to a ground resolution of 1.1 km at the nadir point. Because of along-scan overlap of the pixels, the nadir pixel area is nominally 0.8 km by 1.1 km. The instrument scans 2048 pixels per line to a maximum scan angle from vertical of 55 degrees, covering a distance on the ground of about 2500 km. Because of the extreme angle at the ends of the scanlines, geometric and atmospheric distortion become substantial for areas imaged at the limits of the scanlines. Scenes that fell at the edges of the satellite viewing area were generally not used because of the poor image quality.
The AVHRR has 5 channels on the current satellites, 2 reflected light, and three thermal-infrared. The two reflected light channels (580-680 nm and 720-1000 nm) are used to derive water reflectance images for this study. These channels are used for a variety of purposes not addressed here, including mapping of vegetation, snow, ice, and clouds. The thermal infrared images are used to derive the sea surface temperature imagery.
The imagery set contained here is presented in a Mercator projection with a scale (pixel width) of 1.0845 km at the central latitude of 29° 45' N. This corresponds to a map scale of 1.25 km per pixel at the equator (the Mercator projection has a scale which varies with latitude). The images are 512 pixels across and 256 lines high and extend from 28° 30' N to 30° 59.5' N and from 86° 06' W to 91° 51 W.
The remapping process used the nearest-neighbor technique. In this technique, the latitude and longitude is determined for each point in the raw data. The values (SST, reflectance, etc.) for this point are then assigned to the nearest pixel in the mapped image that has the same latitude and longitude. Not all locations or pixels in the mapped image received data. Pixels in the mapped image that did not receive values from the raw data were filled using an average of the values in the surrounding 3x3 block of pixels. About 10% of the mapped pixels were not assigned values from the raw data and were filled using the local average.
The World DataBank 2 was used for overlays to these images. This database includes coastlines, state boundaries, and major rivers. A land mask was derived from the database shoreline.
The clock on board the spacecraft has a drift that leads to errors of 3-8 km in the calculated positions of the pixels. To correct for these errors, the output images were manually shifted to conform to a shoreline to within 1 pixel. This shift concentrated on Mobile Bay, which is at the center of the mapped image. Slight nonlinearities in the georeferencing may still lead to errors of >1 pixel (1 km) in the pixel position near the left and right edges of the mapped image.
The images are found in three major formats: GIF, geoTIFF 5.0, and PCIDSK 5.0. The geoTIFF format should be readable by most programs, and the geoTIFF files maintain the data in a form that can be analyzed. The GIF files are used in the imagery index as thumbnail images. The PCIDSK files are in the format used by PCI Remote Sensing Corporation. These require a program that can read PCIDSK format (PCI, EASI/PACE, or other). They contain mapping information, channel 1 (580-680 nm) and channel 2 (720-1000 nm) raw count values, and a separate cloud mask.
| Equation | Unit |
| SST = (count/6) -5 | °C |
| Reflectance = count/20 | unitless |
The PCIDSK (version 5.0) files are in a format used for the original analysis of the imagery. The PCIDSK format was developed by PCI Remote Sensing Corporation. In addition to the PCI software, PV-Wave includes modules that can read imagery from PCIDSK (v. 5.0) files. The files are located on both CDs: the first CD contains years 1993-1994, and can be found in the "pix" directory. The remainder of the files are located on the second CD. The filenames have the construction Ayymmdd_hh.pix, where yy is the 2-digit year, mm is the month, dd is the day, and hh is the hour in local (Central) standard time. For example, an image taken April 4, 1990 at 1326 CST has a file name A900404_13.pix. The PCIDSK files contain images and segments described in Table 2.
The conversions of the imagey types are shown below. For the raw channel 1 and 2 data, counts 0 to 200 match the original "raw" data 1:1; and from 200-255, one count corresponds to 10 counts of the raw data. Thus a value of 249 would correspond to a raw count of 690 (200 + 49*10). In most cases only clouds produce raw counts above 200 in channels 1 and 2. Where no measurable radiance is detected in channel 1 and channel 2, the minimum raw count has a value of about 36 rather than 0. The maximum raw count that can be saved with the scaling applied here is 755. This handles virtually all conditions of clouds or snow.
The cloud image in the file has a thematic code indicating whether temperature or reflectance identified the cloud (Table 3).
PCIDSK format includes ancillary information that is included in auxiliary segments. The cloud mask has been converted to a bit map for overlays, and georeferencing and other information are found in other segments as noted in Table 2.
| Image # | Image Type |
| 1 | *SST |
| 2 | channel 1 (580-680 nm) raw counts |
| 3 | channel 2 (720-1000 nm) raw counts |
| 4 | *water reflectance |
| 5 | *channel 1 reflectance corrected for Rayleigh radiance |
| 6 | *channel 2 reflectance corrected for Rayleigh radiance |
| 7 | *seston |
| 8 | cloud mask image |
| 9 | *attenuation coefficient |
| Segment # | Data Type |
| 1 | georeferencing segment |
| 2 | text segment with date, time, and satellite information |
| 3 | cloud mask |
*data sets contained in the regional view (both geoTIFF and GIF)
Table 3
| Cloud Image Identifiers |
| 0 - no clouds |
| 100 - cloud identified by temperature only |
| 200 - cloud identified by channel 2 only |
| 255 - cloud identified by both temperature and channel 2 |
U.S. Department of the Interior |
U.S. Geological Survey
URL: http://pubsdata.usgs.gov/pubs/of/1998/0484/technical.html
Page Contact Information: GS Pubs Web Contact
Page Last Modified: Thursday, 17-Apr-2025 12:58:20 EDT
