General Description

The AORadFlux (Arctic Ocean Radiative Fluxes) data set contains gridded monthly fields of radiation balance components over the Arctic ocean for the period 1983-1990. This data set was designed to provide radiative forcing fields for sea ice modeling experiments.

Algorithm

Radiative fluxes in this data set were calculated using the Internationl Satellite Cloud Climatology Project (ISCCP) C2 data set as input. ISCCP C2 data were combined with other information as inputs for a radiative transfer model (STREAMER). Algorithm details may be found in Schweiger and Key (1994)

Grid

This data set uses uses a 100-km version of the Equal Area SSM/I Earth (EASE) grid, an emerging standard grid used for a number of polar data products at NSIDC. The same grid is used for the TOVS Pathfinder Path-P products. This grid uses an equal-area azimuthal projection centered at the North Pole. The grid coordinates, r and s, are defined with axes parallel to the rows (s) and columns (r) (Fig. 1) of the grid and units equal to the sampling interval. The grid sample locations (grid-cell centers) are then the integer coordinate points. The coordinate system starts at the top left corner with r increasing to the right, and s increasing downward. A grid cell (i,j) in the r,s coordinate system is defined as the area between grid coordinates i-0.5 and i+0.5, and j-0.5 and j+0.5. The lower bound is included in the grid cell, while the upper bound is not; i and j are zero-based array indices for this grid cell. This definition means that grid cells are referenced in r and s by their grid cell center coordinate. A grid spacing of 100.27 km is chosen to allow a direct overlay with the TOVS Path-P product (temperature and humidity profiles) and other products being developed at NSIDC. This grid spacing was created by oversampling the radiative fluxes modeled from the ISCCP-C2 produt at a resolution 250km

Figure 1: Coordinate system and layout of grid. Ncol and Nrow are the number of colums and rows in the grid. The grid center is at i=33, j=33. i and j are zero-based indices so that the first (top, left) gridbox is referenced by indices 0,0.

Geographic Transformation

Forward equations to convert from (,) to (r,s):

Where:

R (Earth radius) = 6371.228 km
C (Grid cell size) = 100.27 km
r0, s0 (grid origin in r,s) = 33.0, 33.0

Longitude

Latitude

Array indices i,j are found from r,s by rounding to the nearest integer

Inverse equations to convert from (r,s) to (,):

if :

where:

x = (r - r0) C
y = (s - s0) C

Note: In order to obtain geolocation information for this data set, the user may also choose access the scientific data sets contained in each HDF-formatted file that contain latitudes and longitudes for the center of each grid cell. These are included as separate scientific data sets in the HDF-formatted files.

Format

This data set is implemented in Hierachical Data Format (HDF) (National Center for Supercomputing Applications, 1994) files and followws standards and recommendations outlined in the EOSDIS Version 0 Data Product Implementation Guidelines (GSFC, 1994). This data set requires version HDF 3.3r3 or higher, because they are implemented using the Multifile SD interface for scientific data sets, which was not available prior to this release. Users familiar with HDF will find it easy to access the data using the interfaces provided by the HDF library. Users unfamiliar with HDF may resort to the sample programs and utilities provided, or consult the HDF users and reference guides (NCSA, 1994) that can be obtained via anonymous ftp from ftp.ncsa.uiuc.edu.

Each HDF-formatted file contains two major objects: metadata and scientific data sets. Metadata are implemented as "global attributes." These attributes contain information about the data set such as dates, times, etc. A complete list can be found in the data dictionary in file AORadFlux.data.dictionary.txt or Appendix A of this documents. Attributes can be obtained by the user via a call to the SDreadattr function as shown in the programs fortran_sample_readAORadFlux.f in Appendix B. Radiative fluxes and geolocation information are contained in 7 scientific data sets. Scientific data sets can be accessed via calls to the SDreaddata function from the HDF library. Example programs on how to access the data are included in Appendix B. Users unfamiliar with the HDF interface may choose to modify the example programs fortran_sample.f or the function read_sub.f for their particular application.

Scientific data sets in the HDF formatted file AORadFlux.hdf are three or two-dimensional. (In HDF nomenclature the quantity "rank" is equivalent to the number of dimensions in a data set). HDF stores multidimensional arrays in the same order they were stored in memory by the program that wrote the data. It is important to note the difference between accessing the data via a C or Fortran program. All scientific data sets in the AORadFlux product were written by a C-program with the dimensions defined as:

array[NMONTH][NROW][NCOL]]

for three-dimensional data and:

array[NROW][NCOL]

for two-dimensional data (latitude and longitude grids only).

Note that the the NROW and NCOL dimension correspond to the previously defined s (j) and r (i) coordinates. The C-language stores multidimensional arrays in row major order. This means that the last dimension varies fastest. Since the data were written in C, users accessing the data via a C program will want to define the dimensions of the arrays as shown above. FORTRAN users will have to reverse the order of their dimensions and define their arrays as follows:

array(NCOL,NROW, NMONTH)

for the three-dimensional case, and

array(NCOL,NROW)

for two-dimensional arrays.
Since FORTRAN uses 1-based indexing, FORTRAN users will translate (r,s) coordinates to array indices in i and j by adding 1.

All three-dimensional scientific data sets in the AORadFlux data set have dimensions of NROW=67, NCOL=67 and NMONTH=90. respectively. The two-dimensional scientific data sets LATITUDE_GRID and LONGITUDE_GRID have dimensions of NROW= 67, NCOL = 67. All data sets are in 32-bit floating point format (DFNT_FLOAT32). . The data dictionary containing the definition, unit, data type, dimension and format for each metadata attribute and scientific data set, can be found in Appendix A or in the provided ASCII file AORadFlux.data.dictionary.txt. Additional information of the file structure can be found in Appendix C of this Document.

Data contained in the AORadFlux Radiation data set.

AORadFlux Radiation Parameters 
-------------------------------------------------------------------
Parameter Name  Parameter Definition                           Unit  
-------------------------------------------------------------------
LATITUDE_GRID   Latitude  at each grid point                   DEG   
LONGITUDE_GRID  Longitude at each grid point                   DEG   
DWNVSSRF        Downwelling solar radiation at the surface     Wm-2  
DWNIRSRF        Downwelling longwave radiation at the surface  Wm-2  
UPVSSRF         Upwelling solar radiation at the surface       Wm-2  
UPIRSRF         Upwelling longwave radiation at the surface    Wm-2  
DIRCTOP         Direct solar radiation at the TOA              Wm-2  
UPVSTOP         Upwelling solar radiation at the TOA           Wm-2  
UPIRTOP         Upwelling longwave radiation at the TOA        Wm-2  
-------------------------------------------------------------------

(for further details see Appendix A and file AORadFlux.data.dictionary.txt)

Validation and Accuracy Assessment

The radiative fluxes in this data set have been compared with climatological data based on Russian drifting stations and other satellite based studies (ERBE). A detailed account of these comparisons is presented in Schweiger and Key, 1994. Due to the lack of sufficient correlative data, an absolute validation of this data set is currently not possible. While top-of-the atmosphere radiative fluxes are in good agreement with direct satellite measurements from ERBE, large discrepancies between calculated fluxes and climatological values at the surface can be ascribed to deficiencies of the ISCCP algorithm in the polar regions. The user is advised of these potentially large errors and of the experimental nature of this data set. Although surface fields may contain errors, mostly due to underestimation of cloudiness by the ISCCP algorithm, we believe that this data set provides the unique opportunity to include temporal and spatial variability in radiative the forcing fields used in sea ice modeling experiments.

Data availability

The AORadFlux Radiation data set is available via anonymous ftp from:

ftp://psc.apl.washington.edu/pub/poles/AORadFlux

Utility Programs

A FORTRAN function to access the hdf-formatted data and a sample program are provided in the anonymous ftp directory (ftp://psc.apl.washington.edu/pub/poles/AORadFlux).

fortran_sample.f:

Sample FORTRAN program that shows how to access the hdf-formatted AORadFlux radiation data sets via function read_sub. Each scientific data set is written to a separate output file. Output files from this program are provided, as binary files (*.ftst.gz). Users without access to the hdf library may choose to use these files instead. Note, that these files are written as binary files on the SUN Sparc architecture. Users on different machines may have to accomodated different byte orders or floating point representations.

read_sub.f:

This file contains a number of functions that allow the access to the HDF formatted data files. It contains function read_sub which provides a single function interface to all the variables in the hdf-formatted file.

plotgrid_all.pro

IDL routine to contour-plot output from the fortran_sample program. Requires ch.pro.

Makefile

Use this makefile to build sample programs or as a template for developing your own applications. Edit macro HDFROOT to point to your local HDF installation and type make.

Notes

The data set and access programs have only been tested on SUN Sparc Architectures under SUN-OS 4.1.X. The HDF library is supported on a wide range of architectures and access should be identical.

Access of to scientific data sets via FORTRAN programs requires a "work around" due to a known bug in version 3.3r3 of the HDF library. The STRIDE variable in function ppgetsds (call to sfrdata) is set to 1,1,1 instead of the 0,0,0 as would be expected from the documentation. Future versions for HDF will fix this bug, so that the STRIDE variable in the sfrdata call will have to be changed to 0,0,0. (If you have installed patch level 2 of release 3.3r3 to your hdf library a STRIDE value of 0,0,0 is appropriate)

Requirements

HDF libraries version 3.3 release 3 or higher. If your site doesn not have the HDF libaries, you can obtain a copy via anonymous ftp from

ftp.ncsa.uiuc.edu

Fortran 77 compiler if you are planning to use the supplied Fortran programs.

Contacts

Axel Schweiger, Polar Science Center, 1013 NE 40 th Street Seattle,WA. 98105, 206-543-1312 axel@apl.washington.edu
Jeff Key, Cooperative Institute for Research in Environmental Sciences (CIRES). University of Colorado. P.O. Box 216. Boulder, Co. 80309. 303-492-3450. jeff@stratus.colorado.edu.

References

Schweiger, A. J. and Key. J. 1994. Arctic Ocean Radiative Fluxes and Cloud Forcings Estimated from the ISCCP C2 dataset, 1983-1990. Journal of Applied Meteorology, Vol. 33., No. 8, 948-963
Schweiger, A. J., and Key. J. 1992. Arctic Cloudiness: Comparison of ISCCP-C2 and Nimbus-7 satellite-derived cloud porducts with a surface-based cloud climateology. Journal of Climate, 5, 1514-1527.
Rossow, B.W. , and R. A. Schiffer. 1991. ISCCP Cloud Data Products. Bull Amer. Meteor. Soc., 72(1), 2-20.
Rossow, B. W., L.C. Gardener, P. Lu and A. Walker, 1991: International Satellite Cloud Climatology Pojerct (ISCCP). Documentation of Cloud Data. World Climate Research Programme. WMO/TD No. 266
GSFC. EOSDIS Version 0 Data Product Implementation Guidelines. Draft version 1.0. GSFC 50-003-04, GSFC
NCSA. HDF users and reference manual, University of Illinois, Champagne/Urbana, 1994

Acronyms and Glossary

ISCCP: International Satellite Cloud Climatology Project
C2 data set: Monthly averages of ISCCP data set
AVHRR: Advanced Very High Resolution Radiometer
HDF: Hierachical Data Format
EASE: Equal Area SSM/I Earthgrid
SSM/I: Special Sensor Microwave/Imager
NSIDC: National Snow and Ice Data Center
Level 4: Gridded geophysical Parameters derived from lower level data. (eg. model results)
Level 3: Gridded geophysical data
Level 2: Geophysical data in orbital swaths
Level 1b: Georeferenced sensor data in orbital swaths
Granule: Unit of a data set, for example a file.

Table of Contents

Next Chapter