NSIDC was alerted to errors in the latitude, longitude and pixel area files supplied with the SSM/I polar stereographic data. This error affects the latitude, longitude and pixel area files supplied with the "DMSP SSM/I Daily Polar Gridded Brightness Temperatures," "DMSP SSM/I Daily and Monthly Sea Ice Concentration Grids for the Polar Regions," and "Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I Passive Microwave Data".
Errors impact the northern polar region files:
The location files from the "DMSP SSM/I Daily Polar Gridded Brightness Temperatures" and "DMSP SSM/I Daily and Monthly Sea Ice Concentration Grids for the Polar Regions" were found to be in error by up to 8 km, while the area file was in error by up to 0.5 sq. km.
On 22 June 1999, files in error were replaced on NSIDC's ftp site with corrected files. Users of this data should download the new latitude, longitude, and pixel area files via ftp.
The latitude and longitude files accompanying "Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I Passive Microwave Data" are in error for both polar regions. Below are the impacted files:
latitude.n longtude.n latitude.s longtude.s
Location files for the "Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I Passive Microwave Data" were found to be in error by 27.7 to 36.3 km.
Corrected latitude and longitude files were staged to NSIDC's FTP site on 25 June 1999.
The error in the northern grids was assessed using GRIDLOC, a program used to generate the SSM/I polar stereographic grid, with the following input:
Polar Stereographic Ellipsoid 90.0,-45.0, 70.0 lat0, lon0, lat1 0.0 rotation 100.0 scale (km/map unit) 90.00, 135.00 center: lat,lon 20.00 , 90.00 lat: min,max -180.00, 180.00 lon: min,max 10.00, 15.00 grid 00.00, 00.00 label: lat, lon 1, 0, 0 cil, bdy, riv 6378.273 Earth equatorial radius (km) 0.081816153 eccentricity
The output from the GRIDLOC program matches the output of the LOCATE.FOR program, available via ftp, to within 7 m. Differences between the program results are random noise with no spatial pattern. Both LOCATE.FOR and GRIDLOC match the brightness temperature processing system code, thereby indicating that these are the best references for the polar stereographic grid cell centers' true values.
Output comparisons between GRIDLOC, the NSIDC files (distributed with the "DMSP SSM/I Daily Polar Gridded Brightness Temperatures" data), and files provided by Goddard Space Flight Center (GSFC)(distributed with the "Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I Passive Microwave Data") result in the following values for the Northern Hemisphere:
Source lat min lat max lon min lon max NSIDC files 31.096709 89.836268 0.0000000 359.99998 GSFC files 30.858588 89.836817 0.0000000 359.80773 gridloc N3B 31.102669 89.836808 -180.00000 179.81398 locate.out 31.102659 89.836807 0.0000000 359.80899
Prior to correction, the NSIDC files differed from LOCATE values by up to 8km. The distance forms an annulus around 60N. The azimuth is randomly either -180 or 180 (i.e. always south), indicating that the latitude at which 25 km on the map equals 25 km on the Earth differs between the NSIDC and LOCATE files. Using the LOCATE program with the above input, 70 degrees is the latitude at which 25 km on the map equals 25 km on Earth.
The existing GSFC values differ from those calculated using LOCATE by 27.7 to 36.3 km. The distance forms a symmetric hump with the maximum at the North Pole. The azimuth swings right around the pole (-180 to 180) indicating a straight-line translation toward 0 degrees longitude. The distance error (dist=arclength(llat(0:302,0:446), llon(0:302,0:446), glat(1:303,1:447), glon(1:303,1:447))) is close to the distance defined by the grid cell hypotenuse. Shifting the GSFC grid up and left one cell causes the distance and azimuth to match LOCATE's values to within approximately 7 m and GRIDLOC's values to within approximately 2 m. Thus, it appears that the GSFC files were shifted by one pixel.
Additional note: Users should be aware that the NSIDC and GSFC pixel-area grids differ slightly (by no more that one percent) due to the use of different procedures to calculate areas.
Please contact NSIDC User Services with any questions.