John Miller SE590 Documentation OTTER Data Description Optical Reflectance Signature Measurements INVESTIGATOR: Dr. J. R. Miller, Co-director, Earth Observations Laboratory, Institute for Space and Terrestrial Science, 4850 Keele Street, North York, Ontario, Canada, M3J 3K1 CONTACT: J. R. Freemantle, Project Scientist, Earth Observations Laboratory, Institute for Space and Terrestrial Science, 4850 Keele Street, North York, Ontario, Canada, M3J 3K1 Tel: (416)-665-5405 Fax: (416)-665-2032 E-mail: freemant@eol.ists.ca Field operatives: J. R. Freemantle (ISTS), M. Belanger (York University), Y. Awaya (Forestry and Forest Products Research Institute, Japan) Data Processing: J. R. Freemantle (ISTS), M. Belanger (YORK), P. Shepherd (ISTS). REQUESTED ACKNOWLEDGMENT: Citation would be sufficient: Suggest "Otter Spectron Spectral Reflectance Measurements" J. Freemantle, M. Belanger, Y.Awaya and J. Miller. ISTS internal report 1992. INTRODUCTION One method of correcting remote sensing imagery is to use prominent man-made or natural features that have a relatively flat, seasonally invariant spectral reflectance (Hall et al. 1991, Miller et al. 1990, Schott et. al. 1988). These targets, which have a size on the order of many imagery pixels have been called flat fields, pseudo-invarent features (pif) or radiometric control sets. The purpose of our field reflectance measurement campaign was to measure the reflectance of as many of these targets as possible. If these targets can be selected from the airborne imagery using an image processing system, a simple calibration constant can be calculated. This constant will convert sensor digital number to percent reflectance. The calibration constant is good only for the image it was calculated for, but the pif target can be used to calibrate other images. This method corrects for atmospheric and sensor calibration effects and has proved useful in other research projects. The field targets where chosen with the airborne sensor in mind. Gravel and asphalt roads, gravel pits and parking lots were measured where available at each site. The intention of the experiment was to provide a spectral reflectance measurement representative of the pseudo-invariant targets. EQUIPMENT In order to collect reflectance measurements of "flat fields" or pseudo- invarent targets, a "Spectron Engineering SE590" was used. This is a portable Spectroradiometer with interchangeable detector heads. Three measurement heads, with the following spectral ranges: 350-1100 nm, 400-800 nm and a shortwave infrared (SWIR) head 1100-2500 nm, were used during the May 1991 OTTER field campaign. Measurements were made with two heads to provide a spectral signature covering the visible, near-infrared and shortwave infrared. The detector head uses a diffraction grating to disperse the incoming light onto a linear photodiode array. The signal is conducted to the controller electronics via a cable. The microprocessor based controller processes the signal from the detector head. Integration times are from 1/60 sec to 1 sec. The integration time can be automatically selected by the instrument or manually chosen by input from a key pad. The SWIR head always collects data with a 1 sec integration time. The SE590 uses a 12 bit A/D converter and stores data in 16 bit words. The vis-nir heads record data in 256 channels and the SWIR head 66 channels. Scans can be internally averaged. Data is output to a laptop computer or to a builtin tape recorder. Data processing by the laptop computer results in an output file normalized to counts/sec. CALIBRATION Spectron reflectance calibration has been confirmed in-house against "Labsphere" standard reflectance disks. PROCEDURE In the field, natural solar illumination was used to illuminate the target and a reference halon ( Labsphere) panel (99% ). Care was taken to ensure that the time between samples of the panel and the target were minimal and that the sky conditions were the same. The automatic selection of integration time mode was employed resulting in different integration times for the panel and the target. The panel was generally much brighter than the target. The resulting ratio gives the target reflectance. In order to facilitate consistent measurements in the field a measurement stand was constructed which held the Spectron measurement heads a fixed distance ( 1.0 m) from the target and allowed movement of the heads (+, - 45 degrees) to obtain measurements at different look angles. This may be important for relating the ground based spectral signature measurements with airborne measurements taken at oblique angles. (This off-nadir data set has not been processed at this time.) Generally two measurements were made of each target in the nadir direction and then other measurements were made at off nadir angles as time permitted. Since only two measurements were taken, a standard deviation is not possible. LAB: A number of samples were returned to the laboratory at York University for measurement. These samples where measured in the following manner: The Spectron heads where placed 20 cm from the target at an angle of 26 degrees. The lamp source was at 0 degrees. Dark current measurements were recorded for each integration time used in the series of experiments. An integration time was manually chosen and 8 scans were internally averaged. A gray reflectance reference disk was used (Labsphere, reflectance 50%). The resultant scans were dark current corrected and then put into ratios to produce a reflectance. This procedure generally reduced or removed the problems experienced in the field (see errors). These measurements are labeled "lab". Soil samples described as "dried" were dried with a heat lamp for 1.5 hours and then left to dry without a heat lamp overnight. The soil was measured by the Spectron and then flattened and re-measured. ERRORS: It was discovered during processing of the data set that the reflectance ratio calculated between 800 and 1100 nm with the broad band head (350-1100nm) suffered from noise and "sag". It was determined that this is due to very low signal levels (due to low solar energy) in this spectral region and incorrect adjustment for dark current. For that reason data from 800nm to 1100nm is not considered valid. In order to correct this problem a number of samples of rock, gravel and asphalt were returned for measurement in the laboratory. In the lab, a current-controlled lamp with high NIR light output was used as a source. The Spectron was used to measure the reflectance of the samples in the lab as described in Procedures, Lab. QUALITY ASSESSMENT Validation: All submitted spectra have been plotted for obvious discrepancies Confidence Level: Subject to problems discussed in ERRORS, I (JF) have confidence in the quality of the data set. Data from 800 nm to 1100 nm is not considered valid. Data taken at Site 3 (Scio) on May 20, 1991 shows poor signal to noise in the SWIR region due to lateness in the day and cloudy conditions. NOTES See important note on problems with field data in errors section. How representative of the targets the measurements are is subjective. Data from 800 nm to 1100 nm is not considered valid. REFERENCES Hall, F. , D. Strebel, J. Nickeson and S Goetz " Radiometric Rectification: Towards a Common Raiometric Response Among Multidata, Multi sensor images. Remote Sensing of Environment 35:11-27 (1991). Miller J, C. Elvidge, B. Rock and J. Freemantle "An Airborne Perspective on Vegetation Phenology from the Analysis of AVIRIS Data Sets over the Jasper Ridge Biological Preserve". Proceedings IGARSS'90 Washington, D. C., pp 565-568, May 20-24, 1990. Reflectance Calibration Standards, Labsphere, Inc. P.O. Box 70, North Sutton, NH, USA 03260. Schott, J. , C. Salvaggio and W. Volchok. " Radiometric Scene Normalization Using Pseudoinvariant Features." Remote Sensing of Environment 26:1-16 (1988). SE590 Field-Portable Data-logging Spectroradiometer Operating Manual, Spectron Engineering, Inc. 225 Yuma Court, Denver CO 80223 USA