---

HIRS CLOUD-CLEARED Version 1 PRODUCTS

The HIRS cloud-cleared products were developed from the HIRS 1B data set.   The International TOVS Processing Package (ITPP) was used to calibrate the raw counts into brightness temperature for TIROS-N, NOAA-6, -7,-8,-9,-10,-11,-12,-14 and ATOVS and AVHRR processing package (AAPP) was used for processing NOAA-15..  A modified ISCCP approach was used for cloud-clearing and multiple channel regression was developed for limb correction.  We currently have data available for each satellite starting in November 1978 and ending December 1999.

The first processing stage conducted in the spring 2000 used a cloud detection method and limb correction algorithm so to produce a clear-sky scan line and gridded data sets.

Details of the products derived from our process are given below.  File name formats are given for each product and the file name key is provided.  Software for reading the satellite-view and gridded products can accessed via anonymous ftp.  Data and software requests can be made to Darren.L.Jackson@noaa.gov.

Orbital statistics

These files containing statistical information about each HIRS orbit, satellite and channel.  The parameters include mean, maximum, minimum and standard deviation of brightness temperature, number of observations, number of ITPP bad data, and number of bad data as determined by additional quality control.  There are approximately 1000 orbital log files using a total of 250 Mbytes.

Scan line products

File format

These data give all the clear-sky HIRS observations at the resolution of the satellite observations.  Data are stored in daily files that consume about 4-5 Mbytes each.  The entire HIRS clear-sky satellite view archive is approximately 60 Gbytes.  These pixel data have a 56 byte record structure.  We have Fortran and IDL code that reads these data and applies limb correction.  Addition coefficient files are need for the limb correction.
 

VARIABLE	DATA TYPE	DESCRIPTION	SCALE & OFFSET
itime	int*4	GMT time in seconds from start of day	itime=time*100
ilon	int*2	longitude in degrees (0  to 359.99; GM eastward)	ilon=(lon-180)*100
ilat	int*2	latitude in degrees (-90 to 90)	ilat=lat*100
iline	int*2	line number(1-1100)	NA
isp	byte	scan position (1-56)	NA
iszen	int*2	solar zenith angle in degrees (0-90)	iszen=angle*100
ialt	int*2	satellite altitude in km	ialt=altitude*10
iqc	byte	spare	NA
isf	byte	spare	NA
iref	byte	reflectance	iref=reflectance*100
itb	int*2(19)	channels 1-19 brightness temperatures (Kelvin)	itb=(TB-100.)*100

I/O information

Tape format

Grid products

File format

All gridded data are INT*2 with array arrangement of  LON x LAT X CHN.   LON and LAT are determined by resolution.  Latitudes only range from 70N-70S.  Channel array element is 20 representing each HIRS channel.  Data were created on PC, so byte swapping might be an issue.  The mean fields are scaled and offset (MEAN = INT2/100. + 100.) while the standard deviation fields were only scaled (STD = INT2/100.).  Count values do not need modification.  Missing values are -99 AFTER offset and scale are applied. First grid box (1,1) is located at (0.5E,69.5S).  Longitudes move eastward.  All data values for channels 1-19 are brightness temperatures expressed in Kelvin.  Channel 20 is reflectance with values between 0 and 1.

I/O information

Tape format

The following table identifies all of the available grid products. (Monthly and Pentad grid products table)

Data Problems

• An error in the limb correction algorithm applied to Channel 8 data prior to cloud detection caused the removal of too many far-limb observations.
• Some cloud observations escaped detection particularly over land regions in the Tropics.
• Standard deviation field reveals higher than normal values for the December data of any leap year. Likely a bug in the cloud detection algorithm for these months.
• Artificial features in the standard deviation file in the TIROS-N data after Jan 1980. Features found primarily over Saudi Arabia and Northern Africa.
• NOAA-6 has a low number of observations for March 1981. Unknown reason why.
• Large number of missing values identified in orbit statistics are due to problem in ITPP code used to process the 1b data into brightness temperatures. Most of the missing values originate from missing data indentified in one of the three calibration lines. These values should not have been identified as missing.