The Fraser (Fraser et al., 1992) and LOWTRAN 7 (Kneizys et al., 1988) models were used for computation of coefficients used to correct radiances of scattered radiation measured by aircraft and/or satellite during FIFE. The Fraser algorithm is designed to compute the surface reflectance for a given measured radiance, or alternatively, the upward radiance at an arbitrary height when the surface reflectance is given. LOWTRAN 7 is a low-resolution propagation model and computer code for predicting atmospheric transmittance and background radiance from 0 to 50,000 [cm^-1] at a resolution of 20 [cm^-1].
The FIFE Staff Science effort covered those activities that were FIFE community-level activities or required uniform data collection procedures across sites and time. These activities included acquiring and processing data from instruments on several satellites.
As part of the FIFE staff science data collection effort, the FIFE Information System (FIS) utilized atmospheric correction and related algorithms to generate coefficients for deriving corrected values from the FIFE level-1 image products. These coefficients were used by FIFE staff in calculating site reflectances from pixel values extracted from the level-1 imagery. See the documents entitled Site Reflectances Extracted from Landsat TM Imagery, Site Reflectances Extracted from SPOT HRV Imagery, and Site Average Reflectances Extracted from AVHRR-LAC Imagery for a discussion of the site reflectance calculation.
Irradiance, transmittance, and effective wavelength.
See the Objective/Purpose Section.
Staff Science Satellite Data Analysis Program.
Jeffrey A. Newcomer
NASA Goddard Sp. Fl. Ctr.
NASA Goddard Sp. Fl. Ctr.
The Satellite Image Value Conversion Coefficients were produced by the FIFE Information System staff. The dedicated work of Scott Goetz and Jeff Newcomer is especially appreciated.
The correction of remote sensing data acquired from satellites for effects due to the intervening atmosphere has proven to be a difficult problem. Not only does the atmosphere reduce the transmission of the incoming, reflected, and emitted radiation, but it contributes reflected and emitted radiation of its own. Under some conditions, atmospheric radiation comprises over 90% of the satellite-observed radiance, but even much smaller effects would degrade the quantitative use of these data unless they are taken into account.
Therefore, in estimating geophysical and biophysical parameters from remotely-sensed data, the atmospheric effects on the transmitted and reflected solar radiation should be factored into the analysis, so that appropriate correction schemes can be employed to infer reflectivity of the ground from satellite radiometric data. Some of the correction techniques require derived coefficients, as inputs in the algorithms that perform the atmospheric correction. The Fraser (Fraser et al., 1992) and LOWTRAN 7 (Kneizys et al., 1988) models were used for computation of coefficients used to correct radiances of scattered radiation measured by aircraft and/or satellite during FIFE.
The Fraser algorithm is designed to compute the surface reflectance for a given measured radiance, or alternatively, the upward radiance at an arbitrary height when the surface reflectance is given. The algorithm is applicable to many wavelengths in the visible and near-infrared spectrum (with appropriately specified gaseous absorption), for a wide range of solar and observation zenith angles, azimuth angles between the observer and the solar rays, and any height of the observer (aircraft or satellite). Any practical value of the aerosol optical thickness can be used, but the tabulated radiation parameters are computed for a specific aerosol size distribution and refractive indices. LOWTRAN 7 is a low-resolution propagation model and computer code for predicting atmospheric transmittance and background radiance from 0 to 50,000 [cm^-1] at a resolution of 20 [cm^-1].
The Advanced Very High Resolution Radiometer(AVHRR), Thematic Mapper (TM), and the High Resolution Visible (HRV) sensor systems used to collect the original data from which this data set was produced have been described in detail for the Level-1 Advanced Very High Resolution Radiometer (AVHRR) Images, Landsat Thematic Mapper (TM) Averages, and FIFE SPOT High Resolution Visible (HRV) Averages data. See the individual documents describing those data sets for details.
The satellite data conversion coefficients were produced for the FIFE level-1 satellite data by the FIS staff at Goddard Space Flight Center. A variety of algorithms were employed. The primary visible correction algorithm was developed by R.S. Fraser of GSFC under contract to FIFE. It is described in detail in the technical memorandum available in the Scanned Document subdirectory on FIFE CD-ROM Volume 1. In addition, the LOWTRAN 7 algorithm of Kneizys et al. (1988) was obtained from NTIS and the split window technique of Price (1983) was implemented. Auxiliary data such as visibility, atmospheric optical characteristics, and atmospheric water content were obtained from FIS.
The FIFE study area, with areal extent of 15 km by 15 km, is located south of the Tuttle Reservoir and Kansas River, and about 10 km from Manhattan, Kansas, USA. The northwest corner of the area has UTM coordinates of 4,334,000 Northing and 705,000 Easting in UTM Zone 14.
The calculated coefficients are applicable only to imagery acquired over the FIFE study area.
The optical characteristics of the atmosphere vary on many scales. The satellite data conversion coefficients represent single-point theoretical calculations with no intrinsic spatial resolution.
Correction coefficients were calculated for images acquired over the FIFE study area between February 3, 1987 and October 13, 1987.
The correction coefficients are calculated for a specific date and time at which there was a satellite overpass. They are only applicable for a period of a few hours centered on that time.
The SQL definition for this table is found in the SAT_COEF.TDF file located on FIFE CD-ROM Volume 1.
Parameter/Variable Description Range Units Source
OBS_DATE The date on which the image data min = 10-FEB-87, FIS were recorded. max = 13-OCT-89
OBS_TIME The time at which the data at min = 1339, [GMT] FIS the center of the level-1 image max = 2221 were collected.
IMAGE_ID The image identification code for min = 4215216345-1, FIS the FIS satellite image to which max = SX043-1 conversion coefficients apply.
PLATFORM The satellite platform on which min = LANDSAT-4, FIS the data collecting instrument is max = SPOT1 mounted.
INSTR_ID The instrument which collected min = AVHRR-LAC, FIS the image data. max = TM
NORMLZD_PATH_RADNC The normalized path radiance, min = .0013, [Watts] FIS that is the upward radiance caused max = 4.234 [meter^-2] solely by atmospheric scattering [ster^-1] and emission, assuming zero [micrometer^-1] surface reflectance.
IRRADNC The normalized downward solar min = .4599, [Watts] FIS flux (irradiance) through a max = .9267 [meter^-2] horizontal surface at the ground. [ster^-1] [micrometer^-1]
TRNSMTNC The fraction of the total min = .6805, FIS irradiance (diffuse + direct) max = .9392 transmitted from the surface to the observer.
BACKSCAT_RATIO The atmospheric backscattering min = .0042, FIS ratio. max = .2018
EFCTV_WAVLEN The effective wavelength for a min = .486, [microns] FIS given sensor band response max = 11.45 function, calculated as a function of surface reflectance, aerosol optical thickness, and geometry.
SOLAR_ZEN_ANG The solar zenith angle, i.e., the min = 17.1, [degrees] FIS angle between lines drawn from the max = 78 observed point a) vertically upward and b) to the sun position.
CORRECTION_MTHD The atmospheric correction method FIS used to obtain the coefficients: the Fraser, et al. algorithm, Lowtran 7, or Split Window.
FIFE_DATA_CRTFCN_CODE * The FIFE Certification Code for CPI - checked by FIS the data, in the following format: primary investigator CPI (Certified by PI), CPI-??? (CPI - questionable data).
LAST_REVISION_DATE data, in the format (DD-MMM-YY). max = 25-APR-91
Decode the FIFE_DATA_CRTFCN_CODE field as follows:
The primary certification codes are: EXM Example or Test data (not for release) PRE Preliminary (unchecked, use at your own risk) CPI Checked by Principal Investigator (reviewed for quality) CGR Checked by a group and reconciled (data comparisons and cross-checks)
The modified certification codes currently in use are: PRE-NFP Preliminary - Not for publication, at the request of investigator. CPI-MRG PAMS data which is "merged" from two separate receiving stations to eliminate transmission errors. CPI-??? Investigator thinks data item may be questionable.
OBS_DATE OBS_TIME IMAGE_ID PLATFORM INSTR_ID --------- -------- ---------------- ---------- ------------ 08-MAY-87 2146 LAC0912378N-1 NOAA-9 AVHRR-LAC 08-MAY-87 2146 LAC0912378N-1 NOAA-9 AVHRR-LAC 09-MAY-87 2136 LAC0912392N-1 NOAA-9 AVHRR-LAC 09-MAY-87 2136 LAC0912392N-1 NOAA-9 AVHRR-LAC NORMLZD_PATH_RADNC IRRADNC TRNSMTNC BACKSCAT_RATIO EFCTV_WAVLEN ------------------ --------- --------- -------------- ------------ .06710 .85150 .85230 .08180 .6390 .02990 .77520 .77630 .03250 .8450 .05260 .85730 .87910 .08180 .6390 .02320 .78290 .81220 .03250 .8450 SOLAR_ZEN_ANG CORRECTION_MTHD FIFE_DATA_CRTFCN_CODE LAST_REVISION_DATE ------------- --------------- --------------------- ------------------ 49.400 FRASER RTC CPI 05-FEB-91 49.400 FRASER RTC CPI 05-FEB-91 47.200 FRASER RTC CPI 05-FEB-91 47.200 FRASER RTC CPI 05-FEB-91
Correction coefficients were calculated for images acquired over the FIFE study area between February 3, 1987 and October 13, 1987. The satellite data conversion coefficients represent single-point theoretical calculations with no intrinsic spatial resolution.
A general description of data granularity as it applies to the IMS appears in the EOSDIS Glossary.
The CD-ROM file format consists of numerical and character fields of varying length separated by commas. The character fields are enclosed with a single apostrophe. There are no spaces between the fields. Each file begins with five header records. Header records contain the following information: Record 1 Name of this file, its table name, number of records in this file, path and name of the document that describes the data in this file, and name of principal investigator for these data. Record 2 Path and filename of the previous data set, and path and filename of the next data set. (Path and filenames for files that contain another set of data taken at the same site on the same day.) Record 3 Path and filename of the previous site, and path and filename of the next site. (Path and filenames for files of the same data set taken on the same day for the previous and next sites (sequentially numbered by SITEGRID_ID)). Record 4 Path and filename of the previous date, and path and filename of the next date. (Path and filenames for files of the same data set taken at the same site for the previous and next date.) Record 5 Column names for the data within the file, delimited by commas. Record 6 Data records begin.
Each field represents one of the attributes listed in the chart in the Data Characteristics Section and described in detail in the TDF file. These fields are in the same order as in the chart.
Reflectance channel derivation using Fraser algorithm:
The relation between the measured radiance and the surface reflectance is expressed as a function of the path radiance, downward flux at the ground, atmospheric transmission, and the atmospheric backscattering ratio (Fraser et al., 1992). Using this relation, a lookup table is constructed, which relates the measured upward radiance to surface reflectance for several aerosol optical thicknesses, solar zenith angles, measurement wavelengths, and a range of observation directions. This lookup table is based on the tabulation of the results of radiative transfer computations that are made using a Dave (1972 a,b,c,d) code. It is assumed that the atmosphere and surface are horizontally homogeneous, and the surface reflects light according to Lambert's law. The light scattered by the atmosphere and the surface is assumed to be unpolarized. The atmosphere is also assumed to be cloud-free. For the most part, aerosol extinction is the dominant parameter in the aerosol component of the atmospheric effect (Fraser and Kaufman 1985). Thus, in this algorithm, the aerosol optical thickness is the only variable aerosol parameter. The algorithm uses a constant aerosol single scattering phase function and scattering albedo chosen to represent a rural environment. In the following, the mathematical basis of the algorithm is given. Details of applying it are discussed. Finally, an error analysis is made.
Thermal channel derivation using LOWTRAN 7 code:
The LOWTRAN 7 code calculates atmospheric transmittance, atmospheric background radiance, single scattered solar and lunar radiance, direct solar irradiance, and multiple scattered solar and thermal radiance. The spectral resolution of the model is 20 [cm^-1] (full width at half maximum) in steps of 5 [cm^-1] from 0 to 50,000 [cm^-1] (0.2 um to infinity). The code uses a single-parameter band mode for molecular absorption, and includes the effects of continuum absorption, molecular scattering, and aerosol extinction. Refraction and Earth curvature are considered in the calculation of an atmospheric slant path and attenuation amounts along the path. The code contains representative atmospheric, aerosol, cloud, and rain models, which are provided with the option to replace them with user-provided theoretical or measured values.
The angular scattering of light by the atmosphere is specified by the phase function that gives the differential probability of the scattered radiation going in a given direction. The scattering by the aerosols and air molecules are treated separately using the appropriate phase function for each. The angular distribution from the two types of scattering are combined, weighted by the corresponding scattering coefficients.
The single scattering approximation is valid over a broad range of conditions found in the atmosphere. However, there are also conditions of interest in the atmosphere where multiple scattering and/or internal sources must be included to accurately calculate the atmospheric radiance. There is no simple indicator that predicts the conditions for which the single solar scattering approximation is acceptable; rather the range of applicability depends upon a large set or parameters including the atmospheric profile, the optical path, the solar geometry, the aerosol phase function, and the wave-number region.
FIFE staff creates the satellite extract coefficients data by:
Default Atmospheric Optical Thicknesses (all quantities are dimensionless) AVHRR-LAC Aerosols Water Other gases --------- -------- ------ ----------- Channel-1 0.15 0.0208 0.0247 Channel-2 0.10 0.1156 0.0152 Landsat-TM Water Other ---------- ------- ------- Band1 0.00000 0.00671 Band2 0.00020 0.03194 Band3 0.00673 0.01829 Band4 0.05011 0.00339 Band5 0.12750 0.00849 Band7 0.09722 0.00792 HRV1 HRV2 PAN Water Other Water Other Water Other ------- ------- ------- ------- ------- ------- Band1 0.00740 0.02526 0.00451 0.02732 0.01610 0.02738 Band2 0.01319 0.02881 0.01720 0.02456 - - Band3 0.04908 0.00849 0.05362 0.00127 - -
The accuracy of the satellite data conversion coefficients depends on the accuracy of the input data and the applicability of the assumptions of the atmospheric model being employed. In cases where default inputs are used, error may be greater than when measured values are used. The calibration of sunphotometers to derive optical thickness can be in error. The models vary in how multiple scattering is handled. Some of the models are not well tested under the clear (very low optical thickness) conditions sometimes observed at the FIFE study area.
The FIS staff has used the satellite data conversion coefficients to derive reflectances used in a number of analyses. Obvious errors have been corrected. Comparisons have been made where independent input data were available. However, there is no method of determining these coefficients, which is not dependent on an atmospheric model.
The spatial and temporal variability of atmospheric optical conditions affecting satellite data corrections are not well known. These coefficients are generally useful for the area of the FIFE site near the time for which they are calculated. However, correction of large-scale data (e.g., AVHRR) using default input parameters has produced suspect results.
The precision of high-resolution satellite remote sensing estimates of surface reflectance (Hall et al., 1992), calibrated and corrected for atmospheric effects, was no worse than about 1 percent absolute. The errors may actually be smaller, but an upper bound of 1 percent results from sampling variance caused by differences among the satellite and ground sensors in spatial resolution, atmospheric effects, and calibration.
No quantitative assessment was made, see the Confidence Level/Accuracy Judgment Section.
Other errors mentioned in the Sources of Error Section were not assessed.
FIS staff applied a general Quality Assessment (QA) procedure to the data to identify inconsistencies and problems for potential users. As a general procedure, the FIS QA consisted of examining the maximum, minimum, average, and standard deviation for each numerical field in the data table. An attempt was made to find an explanation for unexpected high or low values, values outside of the normal physical range for a variable, or standard deviations that appeared inconsistent with the mean. In some cases, histograms were examined to determine whether outliers were consistent with the shape of the data distribution.
The discrepancies, which were identified, are reported as problems in the Known Problems with the Data Section.
The data verification performed by the ORNL DAAC deals with the quality of the data format, media, and readability. The ORNL DAAC does not make an assessment of the quality of the data itself except during the course of performing other QA procedures as described below.
The FIFE data were transferred to the ORNL DAAC via CD-ROM. These CD-ROMs are distributed by the ORNL DAAC unmodified as a set or in individual volumes, as requested. In addition, the DAAC has incorporated each of the 98 FIFE tabular datasets from the CD-ROMs into its online data holdings. Incorporation of these data involved the following steps:
Each distinct type of data (i.e. "data set" on the CD-ROM), is accompanied by a documentation file (i.e., .doc file) and a data format/structure definition file (i.e., .tdf file). The data format files on the CD-ROM are Oracle SQL commands (e.g., "create table") that can be used to set up a relational database table structure. This file provides column/variable names, character/numeric type, length, and format, and labels/comments. These SQL commands were converted to SAS code and were used to create SAS data sets and subsequently to input data files directly from the CD-ROM into a SAS dataset. During this process, file names and directory paths were captured and metadata was extracted to the extent possible electronically. No files were found to be corrupted or unreadable during the conversion process.
Additional Quality Assurance procedures were performed as follows:
As errors are discovered in the online tabular data by investigators, users, or DAAC staff, corrections are made in cooperation with the principal investigators. These corrections are then distributed to users. CD-ROM data are corrected when re-mastering occurs for replenishment of CD-ROM stock.
The satellite extract coefficients may be used to correct other portions of the imagery acquired for FIFE.
The Satellite Image Value Conversion Coefficients Data Set contains derived coefficients that are used as inputs in algorithms that infer reflectivity of the ground from satellite radiometric data. The satellite extract coefficients may be used to correct other portions of the imagery acquired for FIFE.
The FIFE field campaigns were held in 1987 and 1989 and there are no plans for new data collection. Field work continues near the FIFE site at the Long-Term Ecological Research (LTER) Network Konza research site (i.e., LTER continues to monitor the site). The FIFE investigators are continuing to analyze and model the data from the field campaigns to produce new data products.
Software to access the data set is available on the all volumes of the FIFE CD-ROM set. For a detailed description of the available software see the Software Description Document.
ORNL DAAC User Services
Oak Ridge National Laboratory
Telephone: (865) 241-3952
FAX: (865) 574-4665
ORNL Distributed Active Archive Center
Oak Ridge National Laboratory
Telephone: (865) 241-3952
FAX: (865) 574-4665
Users may place requests by telephone, electronic mail, or FAX. Data is also available via the World Wide Web at http://daac.ornl.gov.
FIFE data are available from the ORNL DAAC. Please contact the ORNL DAAC User Services Office for the most current information about these data.
The Satellite Image Value Conversion Coefficients data are available on FIFE CD-ROM Volume 1. The CD-ROM file name is as follows:
Note: capital letters indicate fixed values that appear on the CD-ROM exactly as shown here, lower case indicates characters (values) that change for each path and file.
The format used for the filenames is: yyyyMULT.sfx, where yyyy are the four digits of the century and year (e.g., 1987, 1989). The filename extension (.sfx), identifies the data set content for the file (see the Data Characteristics Section) and is equal to .SEC for this data set.
Dave, J.V. 1972a. Development of programs for computing characteristics of ultraviolet radiation. Technical Report. Scalar Case. Program I. FSC-72-0009. IBM Federal Systems Division. Gaithersburg, Maryland.
Dave, J.V. 1972b. Development of programs for computing characteristics of ultraviolet radiation. Technical Report. Scalar Case. Program II. FSC-72-0011. IBM Federal Systems Division. Gaithersburg, Maryland.
Dave, J.V. 1972c. Development of programs for computing characteristics of ultraviolet radiation. Technical Report. Scalar Case. Program III. FSC-72-0012. IBM Federal Systems Division. Gaithersburg, Maryland.
Dave, J.V. 1972d. Development of programs for computing characteristics of ultraviolet radiation. Technical Report. Scalar Case. Program IV. FSC-72-0013. IBM Federal Systems Division. Gaithersburg, Maryland.
Fraser, R.S., R.A. Ferrare, Y.J. Kaufman, B.L. Markham, and S. Mattoo. 1992. Algorithm for atmospheric corrections of aircraft and satellite imagery. Int. J. Remote Sens. 13:541-557.
Kneizys, F.X., E.P. Shettle, W.O. Gallery, J.H. Chetwynd, L.W. Abreu, J.E.A. Selby, S.A. Clough, and R.W. Fenn. 1988. Atmospheric transmittance/radiance: computer code LOWTRAN-7.AFGL-TR-88-0177. Air Force Geophysics Lab. Hanscomb AFB, Massachusetts.
Newcomer, J.A., S.J. Goetz, D.E. Strebel, and F.G. Hall. 1989. Image processing software for providing radiometric inputs to land surface climatology models. IGARSS '89. 12th Can. Symp. on Remote Sensing. pp. 1779-1782.
Fraser, R.S., and Y.J. Kaufman. 1985. The relative importance of aerosol scattering and absorption in remote sensing. IEEE Transactions on Geoscience and Remote Sensing. 23, 625-633.
Hall, F.G., D.E. Strebel, J.E. Nickeson, and S.J. Goetz. 1991. Radiometric Rectification: Toward a Common Radiometric Response Among Multidate, Multisensor Images. Remote Sens. Environ. 35:11-27.
Hall, F.G., K.F. Huemmrich, S.J. Goetz, P.J. Sellers, and J.E. Nickeson. 1992. Satellite remote sensing of surface energy balance: success, failures, and unresolved issues in FIFE. J. Geophys. Res. 97:19061-19089.
Markham, B.L., and J.L. Baker. 1986. Landsat MSS and TM post-calibration dynamic ranges, exoatmospheric reflectances and at-satellite temperatures. EOSAT Landsat Tech. Notes 1:3-7. Lanham, Maryland.
Price, J. C. 1983. Estimating surface temperatures from satellite thermal infrared data - a simple formulation for the atmospheric effect. Remote Sensing Environment. 13:353-361.
Contact the EOS Distributed Active Archive Center (DAAC) at Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee (see the Data Center Identification Section). Documentation about using the archive and/or online access to the data at the ORNL DAAC is not available at this revision.
A general glossary for the DAAC is located at Glossary.
A general list of acronyms for the DAAC is available at Acronyms.
May 6, 1994 (citation revised on October 14, 2002).
Warning: This document has not been checked for technical or editorial accuracy by the FIFE Information Scientist. There may be inconsistencies with other documents, technical or editorial errors that were inadvertently introduced when the document was compiled or references to preliminary data that were not included on the final CD-ROM.
Previous versions of this document have been reviewed by the Principal Investigator, the person who transmitted the data to FIS, a FIS staff member, or a FIFE scientist generally familiar with the data.
Newcomer, J. A., and S. Goetz, 1994. Satellite Atmos[pheric]. Correction Coef[ficients]. (FIFE). Image Value Conversion Coefficients Data Set. Data set. Available on-line [http://www.daac.ornl.gov] from Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge, Tennessee, U.S.A. doi:10.3334/ORNLDAAC/76. Also published in D. E. Strebel, D. R. Landis, K. F. Huemmrich, and B. W. Meeson (eds.), Collected Data of the First ISLSCP Field Experiment, Vol. 1: Surface Observations and Non-Image Data Sets. CD-ROM. National Aeronautics and Space Administration, Goddard Space Flight Center, Greenbelt, Maryland, U.S.A. (available from http://www.daac.ornl.gov).