| FTP Site | CASPR User's Guide (no longer available) |
Retrieval results are available as twice-daily images, monthly mean images, and daily and monthly mean values covering the entire Arctic. The pixel size is 25 x 25 km. The period covered by this data set is all of 1997 through all of 1998. Some days may be missing because of missing or invalid AVHRR data.
The ftp site contains the following: twice-daily image results files, monthly mean image results files, one land mask image, one latitude image, and one longitude image. In addition cloud mask images were also created. All files are stored in netCDF format. The file naming conventions will be described in data format section below. Images are 263 x 263 pixels in size, with contents and data types as described in the Image Parameter Files section below. The means files are described in the Mean Parameter Values section.
The land mask is a 263 x 263 pixel byte array, where a value of 254 indicates land and a value of 0 indicates non-land (ocean or lake). The latitude and longitude files are 263 x 263 pixel short integer (two byte) values in degrees times 100. Latitude are positive north of the equator and negative to the south. Longitudes are positive east of the Prime Meridian and negative to the west (0 to +-180 degrees times 100). These files are also stored in netCDF format.
The latitude, longitude, and landmask files are in the latlon directory; daily and monthly parameter results are in the params directory, and mean parameter files are in the means directory.
The AVHRR sensor is a five-channel system. The scan mirror collects earth observation data during a discrete part of the scan cycle. The scan mirror observes the scene below the spacecraft in a continuous line from horizon to horizon as it rotates. Energy from the scene is collected by a telescope and separated according to wavelength by beam splitters. Signals are amplified, filtered, and applied to the 10-bit analog/digital converter, which samples all five channels simultaneously. The five channels, numbered sequentially 1..5, cover the following spectral ranges: 0.58 - 0.68 mm (visible), 0.725 - 1.05 mm (reflected infrared), 3.55 - 3.92 mm (reflected/thermal infared), 10.3 - 11.3 mm (thermal infrared), 11.5 - 12.5 mm (thermal infrared).
The 10-bit resolution digital data is processed to create direct readout of High Resolution Picture Transmission (HRPT) data, Automatic Picture Transmission (APT) data, 4 km Global Area Coverage (GAC) data, and 1 km Local Area Coverage (LAC) data, to ground stations throughout the world. The AVHRR instrument scans in the across-track direction with a continuously rotating scan mirror, viewing a swath of over 100 degrees and up to 55 degrees off-nadir. Spatial resolution is approximately 1.1 km when the view is at nadir. Scanning to 55 degrees (68 degrees satellite zenith angle relative to the earth's surface) off nadir results in a ground resolution of over 2.4 km by 6.5 km at the maximum off-nadir position. The all of five channel signals received will be calibrated by calibration methods (Cracknell, 1997).
The APP data are twice-daily composites available at 5 km pixel size for June 1981-2000. The study period (September 1997 - August 1998) and the area (Figure 1) correspond to the Surface Heat Budget of the Arctic Ocean (SHEBA) field experiment, where an icebreaker drifted with the pack ice for one year . The APP standard products are clear sky surface temperature and broadband albedo, a cloud mask, sea ice motion, and the calibrated, geolocated channel data and viewing/illumination geometry (Maslanik et al., 2000; Maslanik et al., 1998; Meier 1997). We have extended this product set to include the all sky surface skin temperature and broadband albedo, cloud properties (particle phase, effective radius, optical depth, temperature, and pressure), and radiative fluxes using algorithms in the Cloud and Surface Parameter Retrieval (CASPR) system (Key, 2001). The extended APP data set is called APP-x data set. The calculation of cloudy sky surface skin temperature is based on an empirical relationship between the clear sky surface skin temperature, wind speed, and solar zenith angle (daytime). The cloudy sky broadband surface albedo is determined using the clear sky broadband albedo (interpolated from nearby pixels) adjusted by the cloud optical depth and the solar zenith angle. Radiative fluxes are computed in CASPR using FluxNet (Key and Schweiger, 1998). See Key (2001) and references therein for more information on the algorithms and their validation. The original APP data set resolution (5 km pixel size) has been sampled to a resolution of 25 km by selecting every central pixel in every 5 x 5 pixel box.
One netCDF output file is created corresponding to each input image. First is a list of the parameters that the file contains. The values are: 1=the parameter is present, 0=not present, as a 18-element short integer array. The parameters are given in Table 1 below. In this data set, all parameters are included in the file. Next are two short integer values that specify the size (columns by rows) of the data arrays that follow. The image size in this data set is 263 x 263 (columns by rows), with 25 km pixels. Then come the 18 parameters as short integer (2-byte) arrays, except for the cloud phase and the refined surface type mask which are short integer (2-byte) for daily images and float number for monthly mean images. Output images are band (parameter) sequential. Units for all values are listed in Table 1. After the image data comes the number of profile levels (short integer), followed by the temperature, humidity, and pressure profiles (float array) that are nearest the center of the image. Each profile has the specified number of levels.
Clear pixels in cloud parameter arrays and cloudy pixels in clear arrays are generally MISSING, but that is not guaranteed. MISSING is also used for any invalid pixel. MISSING values should, of course, be ignored. If no clear sky pixels are found for one or more surface types in the image, all results are set to MISSING. No cloud or flux calculations are done for solar zenith angles between DARKZEN (85 degrees) and 90 degrees, so the corresponding pixels values are MISSING. The value for MISSING is 9999.0.
Note: There is no cloud mask in the *.params.cdf files.
However,
one can be created directly from the cloud phase array or access the
cloud
mask data set to get them. Values of 0 and 1 indicate cloud; a value of
2 indicates either clear or invalid data for daily images. But for
monthly
mean cloud mask images the cloud phase array in *.params.cdf
files
are float-point number indicating the relative frequency of cloud phase
occurrence of water or ice phase with the value ranged from 0 ~1. The
monthly
cloud mask images are also the frequency of the cloud occurrence during
that month, which can be obtained in this data set too.
Another file type contains the mean value of each parameter for each image. This is a text file with the date, parameter mean values, and parameter standard deviations for each of the standard parameters (Table 1) plus cloud amount. If more than one image was processed then data from all images is present. The file is structured as follows. The first record is the list of binary values corresponding to each parameter, 1 if computed, 0 if not. The second record is the number of cases (equals the number of images processed) that the file contains. The third record contains the date, which are year, month, day and hour. The fourth record contains the mean values of the parameters (19 in total). The fifth record contains the standard deviations for all parameters. The records for the means and standard deviations will have values for every parameter, but those that were not computed will have values of MISSING. This includes the situation where no clear sky pixels were found for one or more of the surface types present. An example is:
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
263
1992 6 29 21.88
272.999 0.379594 16.0942 4.33651 1.11329 262.619
652.282 1.84806 38.2547 613.432 277.001 220.099
316.748 715.730 314.146 227.239 -10.222 30.123 0.602507
8.72553 0.148731 16.9871 9.16805 0.922641 12.7292
171.225 0.257467 87.5301 135.422 34.8783 84.7163
42.6045 308.334 102.159 22.3348 5.123 6.456 0.00000
...
There are means files for each daily image (in the means directories) and for the monthly mean images (in the monthly/means directories). In both cases the means are calulated over the entire image.
Data for 1997 and 1998 are in different directories, and for each there are separate directories for each of the two times of day (0400 and 1400 LOCAL SOLAR TIME, not GMT!).
The daily parameter image naming convention is app_n025_yyyyddd_tttt.params.cdf.gz, where app means the APP-x data set, yyyy is 4-digit year, ddd is 3-digit julian day, and tttt is 4-digit local solar time in 24-hour format with the last two digits representing the decimal hour. The params means parameter data file, cdf indicates netCDF format file, gz denotes it was gzipped by "gzip" command. For example, app_n025_1997183_1400.params.cdf.gz is a parameter image for julian day 183 of 1997 on the local solar time of 14:00, of which minute and second are not available. It was saved in netCDF file format, and compressed by gzip.
The daily cloud mask image naming convention is similar to that described above, but with "params" being replaced by "cmask".
The monthly mean parameter image naming convention is mean_yyyymmdd_tttt.params.cdf.gz,
where mean means it is monthly mean APP-x data set, yyyy
is 4-digit year, mm is 2-digit month, dd is
2-digit day, and tttt
is 4-digit local solar time in 24-hour format with the last two digits
representing the decimal hour. The params means
parameter data file,
cdf indicates netCDF format file, gz denotes it was
gzipped.
For example, mean_19970999_1400.params.cdf.gz is a monthly mean
parameter
image at 14:00 local solar time for September 1997 . It was saved in
netCDF file format, and compressed
by gzip. The 99 in the dd position means that day
information is not available.
The monthly mean cloud mask image naming convention is as above but with "cmask" in place of "params".
The file nlandmask25.img.cdf.gz contains the surface type mask value of each pixel in the entire study area. It is netCDF format file and compressed by gzip command. The files nlat25.img.cdf.gz and nlon25.img.cdf.gz contains latitude and longitude values, respectively, of each pixel in the entire study area. They are in the netCDF format file and compressed by gzip.
The routines provided for reading the netCDF files were written in IDL (Interactive Data Language from Research Systems, Inc., Boulder, Colorado). Of course, any computer programming language that has access to netCDF libraries can be used. The IDL procedures are in the file read_sheba_caspr.pro in the read directory.
| Name | Purpose |
| read_netcdf_landmask | Read land mask netCDF file. |
| read_netcdf_latlon | Read latitude and longitude netCDF files. |
| read_netcdf_cloudmask | Read cloud mask netCDF file. |
| read_netcdf_params | Read parameter image netCDF file. |
| read_netcdf_means | Read parameter statistical mean and standard deviation file. |
Additional documentation is provided in the individual procedures. NOTE: The netCDF files are compressed, and must be uncompressed before reading.
Key, J., 2001, The Cloud and Surface Parameter Retrieval (CASPR) System for Polar AVHRR Data User's Guide. Space Science and Engineering Center, University of Wisconsin, Madison, WI, 62 pp.
Key, J. and A.J. Schweiger, 1998, Tools for atmospheric radiative transfer: Streamer and FluxNet, Computers and Geosciences, 24(5), 443-451.
Maslanik, J.A., J. Key, C. Fowler, T. Nguyen, 2000. AVHRR-derived regional cloud and surface conditions during SHEBA and FIRE-ACE. J. Geophys.l Res., in press.
Maslanik, J., C. Fowler, J. Key, T. Scambos, T. Hutchinson, and W. Emery, 1998. AVHRR-based Polar Pathfinder products for modeling applications. Annals of Glaciology 25:388-392.
Meier, W.N., J.A. Maslanik, J.R. Key, and C.W. W. Fowler. 1997.
Multiparameter
AVHRR-derived products for Arctic climate studies, Earth
Interactions, Vol. 1.