Please Note:
To use the resources of our website, you need to regist first.

Description of GCIP/GAPP Reprocessed Data at 1/2 degree

Table of Contents:

• File types
• Grid structure
• File structure
• File size
• File name
• Unit of measurement
• Special data value
• Sample Fortran read statements

File types

Instantaneous:	The instantaneous flux is the flux at the nominal time (UTC) of satellite observation. Observations are taken at Minute minutes after the hour. (In 1996, Minute=15 minutes.)
Hourly average:	The hourly average flux is an estimate of the flux averaged for an hour ending at the hour in local standard time. It is obtained from the instantaneous flux by normalizing it to the average solar zenith angle for the hour.
Daily average:	The daily average flux is the 24-hour average of the flux for a day. In calculating the daily average, missing day-time hours are filled in by replicating data from non-missing hours one hour before the missing ones. If the hour one before the missing one is also missing filling is done from the closest hour after the missing one. If all day-time hours are missing no filling is done. Fluxes at a missing hour are obtained by multiplying the non-missing fluxes by the ratio of the solar zenith angle cosine at the non-missing hour to that at the missing hour. In addition, the daily average flux, obtained from summing up the hourly values, is multiplied by the ratio of numerical-to-analytical integral of the daily top of the atmosphere downward flux. The latter procedure is intended to correct for potential inaccuracies in the numerical integration of hourly fluxes. Non-filled and non-corrected daily averages can be obtained from the hourly averages.
Monthly average:	The monthly average flux is the average of daily fluxes for a month. It is the arithmetic mean of daily averages for the non-missing days. Potential errors in this numerical "integral" are corrected by multiplying the arithmetic mean by the ratio of numerical-to-analytical integral of the monthly top of the atmosphere downward flux. Non-filled and non-corrected monthly averages can be obtained from the hourly averages. [Top]
In addidtion to fluxes, instantaneous, hourly, daily and monthly averages of the cloud cover fraction are provided.
Cloud cover:	The cloud cover fraction is calculated as the ratio of the number of cloudy pixels within a grid cell to the total number of pixels (cloudy+clear) in that cell. [Top]

 

Grid structure

Each file has information for 5661 (0.5 by 0.5 degree, equal-angle) cells (51 latitudes and 111 longitudes). All cells are present, even when data are missing. The cells are written sequentially, proceeding eastward through the latitude zone, then northward to the next latitude zone. The center lat/lon coordinates of the first cell are 25 N, 125 W, and those of the last cell are 50 N, 70 W. The following Fortran statements could be used to calculate the center coordinates of cells:       INTEGER   Nlat, Nlon, Lat, Lon       REAL  Lat1, Lon1, Reslat, Reslon       PARAMETER  ( Nlat=51, Nlon=111 )       REAL  Glat( Nlat ), Glon( Nlon )       C            ** Calculate latitude and longitude coordinates of C            ** cell centers       Reslat = 0.5       Reslon = 0.5       Lat1   = 25.0       Lon1   = -125.0       DO 20 Lat = 1, Nlat          Glat( Lat ) = ( Lat-1 ) * Reslat + Lat1 20    CONTINUE       DO 30 Lon = 1, Nlon          Glon( Lon ) = ( Lon-1 ) * Reslon + Lon1 30    CONTINUE Here Glat and Glon are the latitude and longitude coordinates of cell-centers, Reslat and Reslon are the distances between cell-centers in latitude and longitude (resolution), Lat1 and Lon1 are the latitude and longitude coordinates of the center of first cell , respectively. All quantities are in degrees. [Top]

File structure

Data for a particular parameter and time resolution for a month are contained in a single file. Instantaneous and hourly average files contain all 24 hours for a day, and all days for a month, even if data are missing. For the instantaneous files the hours are UTC (0-23), while those for the hourly average files are local standard time (1-24). Daily average files also include all days for a month, even if data are missing. In the instantaneous and hourly average files, all cells are written for a given hour of a given day. Writing of data proceeds from the first hour to the last of a day, and is repeated for all days in the month. That is: (lon1,lat1,h1,d1)    (lon2,lat1,h1,d1)     ... (lon111,lat1,h1,d1)        ... (lon1,lat51,h1,d1)   (lon2,lat51,h1,d1)   ... (lon111,lat51,h1,d1) (lon1,lat1,h2,d1)    (lon2,lat1,h2,d1)    ... (lon111,lat1,h2,d1)        ...        ... (lon1,lat51,h24,d1)  (lon2,lat51,h24,d1)  ... (lon111,lat51,h24,d1)        ...        ...        ... (lon1,lat51,h24,d31) (lon2,lat51,h24,d31) ... (lon111,lat51,h24,d31) n the daily average files, days are written sequentially, that is: (lon1,lat1,d1)   (lon2,lat1,d1)   ... (lon111,lat1,d1)        ... (lon1,lat51,d1)  (lon2,lat51,d1)  ... (lon111,lat51,d1)        ...        ... (lon1,lat51,d31) (lon2,lat51,d31) ... (lon111,lat51,d31) The structure of the monthly average file is as follows: (lon1,lat1)  (lon2,lat1)  ... (lon111,lat1)      ... (lon1,lat51) (lon2,lat51) ... (lon111,lat51) where lon=longitude, lat=latitude, h=hour, and d=day. All files are written unformatted with direct access. The logical record length is 444 bytes (111 columns times 4 bytes). The floating point binary data conform to the little endian specification, and were written on a DEC workstation. Here is a sample Fortran statement to open a file  (e.g., 9606sda.m): OPEN ( Unit=1, File='9606sda.m', Status='OLD', & Form='UNFORMATTED', Access='DIRECT', Recl=Lrec ) where Lrec is the record length (444 for compilers that require bytes for Recl, or 111 for those that need words for Recl). Please Note: If your machine uses Big endian representation of binary data (e.g. SGI, IBM), the instructions of reading data can be found at: http://www.atmos.umd.edu/~srb/gcip/faq.htm   [Top]

Unit of measurement:

Fluxes are given in Wm-2. Albedo and cloud cover fraction are dimensionless, their values range from 0 to 1.  [Top]

Special data value

Missing data are represented by -999. [Top]

Filenames

instantaneous : yymmppp.i hourly : yymmppp.h daily : yymmppp.d monthly : yymmppp.m where yy : year (two digits), mm : month (two digits) ppp : parameter (three characters), their values are: sda: surface downward flux, par: photosynthetically active radiation, tda: top of atmosphere downward flux, tua: top of atmosphere upward flux, sal: surface albedo, ccf: cloud cover fraction. All files are compressed using the gzip utility, and all have the extension ".gz" attached to the above names. [Top]

File size

File type	Uncompressed size (bytes)	Compressed size (bytes) (varies, an example is given)
instantaneous:	15.5M	2.5M
hourly:	15.5M	2.5M
daily:	663K	196K
monthly:	22K	7K

[Top]

Sample Fortran Read Statements

• instantaneous files
• hourly files
• daily average files
• monthly average files

[Top]

Instantaneous files:

      INTEGER Day, Hour, Irec, Lat, Lon, Maxday, Ndays, Nhours, Nlat, 
     &        Nlon 
      PARAMETER ( Maxday=31, Nhours=23, Nlat=51, Nlon=111 )
      REAL  Fluxi( Nlon, Nlat, Nhours, Maxday ), Minute, Time

C              ** Ndays is the number of days in the month
C              ** open for read, Minute is the minutes after the
C              ** hour when satellite observation is taken
C
      Ndays  = 30
      Minute = 15.0

      Irec = 0
      DO 10 Day = 1, Ndays
         DO 20 Hour = 0, Nhours
C              ** Time is the nominal time of the satellite
C              ** observation
            Time = Hour + Minute / 60.
            DO 30 Lat = 1, Nlat
               Irec = Irec + 1
               READ ( 1, Rec=Irec ) ( Fluxi(Lon,Lat,Hour,Day), Lon=1,
     &                Nlon ) 
30          CONTINUE
20       CONTINUE
10    CONTINUE
    

Sample output and sample Fortran program are also available.

[Read statements] [Top]

Hourly files:

      INTEGER Day, Hour, Irec, Lat, Lon, Maxday, Ndays, Nhours, Nlat, 
     &        Nlon 
      PARAMETER ( Maxday=31, Nhours=24, Nlat=51, Nlon=111 )
      REAL  Fluxh( Nlon, Nlat, Nhours, Maxday )

C              ** Ndays is the number of days in the month
C              ** open for read
C
      Ndays = 30

      Irec = 0
      DO 10 Day = 1, Ndays
         DO 20 Hour = 1, Nhours
            DO 30 Lat = 1, Nlat
               Irec = Irec + 1
               READ ( 1, Rec=Irec ) ( Fluxh(Lon,Lat,Hour,Day), Lon=1,
     &                Nlon ) 
30          CONTINUE
20       CONTINUE
10    CONTINUE  

Sample output and sample Fortran program are also available.

[Read statements] [Top]

Daily files:

      INTEGER Day, Irec, Lat, Lon, Maxday, Ndays, Nlat, Nlon 
      PARAMETER ( Maxday=31, Nlat=51, Nlon=111 )
      REAL  Fluxd( Nlon, Nlat, Maxday )

       Ndays = 30

       Irec = 0
       DO 10 Day = 1, Ndays
          DO 20 Lat = 1, Nlat
             Irec = Irec + 1
             READ ( 1, Rec=Irec ) ( Fluxd(Lon,Lat,Day), Lon=1, Nlon ) 
20        CONTINUE
10     CONTINUE
       

Sample output and sample Fortran program are also available.

[Read statements] [Top]

Monthly files:

      INTEGER Nlat, Nlon
      PARAMETER ( Nlat=51, Nlon=111 )
      REAL  Fluxm( Nlon, Nlat )

       Irec = 0
       DO 10 Lat = 1, Nlat
          Irec = Irec + 1
          READ ( 1, Rec=Irec ) ( Fluxm(Lon,Lat), Lon=1, Nlon ) 
10     CONTINUE
        

Sample output and sample Fortran program are also available.

[Read statements] [Top]