[Introduction to BRDF]

[ULGS SYSTEM PAGE]

Measurement of BRDF

 

This page will give a brief overview of the measurement of a surfaces bidirectional reflectance.  The data presented here were acquired with the University of Lethbridge Goniometer System (ULGS).   

 

What is a Goniometer?

 

A goniometer is basically a positioning instrument that allows us to maneuver a sensor (a spectroradiometer in this case) over a surface and view it from a hemispherical perspective.  The most basic instrument design involves a fixed base ring and a moveable arch which positions the sensor above the target to gather data from a hemispherical perspective.   The overall design of the University of Lethbridge Goniometer System (ULGS) was focused on portability and flexibility for both laboratory and field use.  The positioning system has a base azimuth ring and a removable zenith arch.  The arch rides in a track on the azimuth ring and is positioned over pre-assigned azimuth angles by way of marked, pre-bored indicators on both sides of the arch.  Similarly, the sensor sled is located at pre-assigned zenith angles using marked, pre-bored positions and is securely held in place using an integrated clamp during measurements (Figure 1). 

 

Figure 1. Set-up of ULGS goniometer in the University of Lethbridge Spectroradiometer Laboratory. Fluxnet moss target, reference panel, illumination source (turned off) and ASD-FR spectroradiometer are shown.

 

The ULGS system was designed to be compact; the radius of the azimuth ring and zenith arch is 60 cm.  This allows for the direct measurement of samples in the laboratory or measurements of relatively short plant canopies in the field.  To increase the utility of the instrument, extendable, variable length legs can be attached to the base ring to enable field measurements on sloped surfaces or for plant canopies up to 130 cm in height in the current design (Figure 2), however, this is easily expanded by simply using longer legs. The zenith arch was designed so that little or no shadow is cast by the arch over the target area. 

While the system is capable of acquiring measurements at any angular precision, it is currently designed to acquire measurements at 10° increments in both zenith and azimuth dimensions. The sensor sled and zenith arch design enables zenith angle measurements from -60° to 60°.   The zenith angle is limited due to the design of the sensor sled.  The sensor sled was designed to accept the pistol grip attachment for the ASD-FR, while an alternate sled could be build to accept only the foreoptic, the incorporation of the pistol grip allows for rapid conversion between hand-held spectra measurement and BRDF measurements.

A given zenith set therefore has 13 angles (including nadir), with 18 azimuths possible (effective azimuth range = 0-170°). However, since the nadir position is the same for all 18 azimuths, this position is only included once (although measured for each zenith set) once resulting in 217 (13 x 18 – 17 OR 12 off-nadir zeniths x 18 azimuths + 1) unique measurements taken for a complete hemisphere. 

At full angular resolution, the acquisition time has averaged 25 minutes, which is adequate for laboratory measurements, and also fast enough to ensure no significant change in solar illumination geometry during field acquisition.

Figure 2.  Field use of ULGS goniometer on a sloped surface at grassland meteorological station located in the coulees of the Oldman River valley in the southern portion of the University of Lethbridge campus.

 

EXAMPLE APPLICATIONS: FIELD AND LABORATORY BRDF ESTIMATES

BRDF estimates using the ULGS goniometer are demonstrated here for a Lambertian reference panels as well as two very different targets and settings: a) field measurement of a grass canopy (crested wheatgrass, Agropyron cristatum) (Figure 2) and b) laboratory measurement of a moss canopy (Pleurozium schreberi) (Figure 1). In all cases, hyperspectral target data were acquired at full angular resolution (10° in zenith and azimuth). For the experiments reported here, a 5° foreoptic barrel was used on the ASD-FR to yield a field of view diameter of 5.24 cm based on a 60cm distance from the sensor to the target located at the centre of the plane formed by the azimuth base ring.

            The protocol used for gathering spectral reflectance data with this instrument followed the procedures outlined in Peddle et al. (2001).   For each set of zenith measurements (i.e. each azimuth), a separate measurement of a calibrated 12”x12” pressed polytetrafluoroethylene (PTFE) reflectance panel (available commercially as Spectralon: Labsphere, 1998), was first acquired at the nadir position prior to for use in computing target reflectance.  Each separate series of azimuth sets therefore has a unique reference panel calibration spectra.  The PTFE panel used offers nearly ideal Lambertian properties.

            Illumination properties are important when estimating BRDF, particularly for field estimates due to changing solar position, weather and atmospheric conditions. Therefore, time is of the essence in the field.  In most field situations a BRDF can be collected at reduced angular resolution to reduce the acquisition time from approximately 25 minutes to 15 minutes.  This is not a concern in a controlled laboratory experiment.  In this study, a 500-watt halogen lamp was used as the source illumination.   As the position of the lamp and optical properties of the illumination environment were constant, measurement time was less critical. 

            These data were then processed to reflectance using a modified version of the protocol outlined in Peddle et al. (2001).  The primary difference in the approach was the inclusion of positional information in the goniometer data.  The zenith and azimuth measurements were then transformed from spherical into rectangular coordinates to facilitate improved visualisation, graphing and manipulation, using the following equations:

 

                       

 

where ρ is the diameter of the sphere, Φ is the zenith angle, and θ is the azimuth angle. Once the data were transformed, surfaces were created using a thin-plate spline technique (ER Mapper, 2002).  This technique maintains the original data points and provides a smooth interpolation (Hutchinson, 1993).

 

4.1 Field Results

 

            The field data gathered for the estimation of the BRDF of crested wheatgrass was collected on August 24, 2004.  At this time, the grass plants were starting to undergo senescence and, therefore, did not display the same spectral signature as a healthy plant.  Several BRDF measurement sets were taken between 12:00 and 13:30 hours, with most of the measurements requiring 20 – 25 minutes to complete. 

The results of this experiment (Figure 3) show a distinct hot spot located in the direction of solar illumination.  This figure shows both planimetric and orthographic projections of the same data and gives a clear indication of the amount of variability possible with respect to view angle properties.  Figure 4 shows the continuous spectra for a variety of different azimuth positions with a constant view zenith angle of 40°.  This illustrates the spectral variability encountered with measurements taken over different azimuth positions.

Figure 3.  Planimetric (top) and orthographic (bottom) projections of BRDF data for Crested Wheatgrass measured in the field.  Note the pronounced hot-spot located near 120°.

Figure 4.  A full-spectrum representation of the Crested Wheatgrass data at a constant 40° zenith for different azimuths.

 

Laboratory Results

            The laboratory studies were conducted with a pleurozium moss canopy collected from the Fluxnet-Canada Western Peatland site in northern Alberta, and transported to the UL SR Laboratory using standard refrigerated shipment methods.  The results of this experiment indicate that there was not a pronounced hot-spot; it was more of a region of low-angle, higher reflectivity around the southern edge (Figure 5).  When these data were plotted as a full spectrum for a variety of angles (Figure 6), little difference is apparent in the visible wavelengths.  The infrared wavelengths, especially between 800 to 1200 nm, show a greater degree of variability with change in azimuth angle and in some parts the changes were not as continuous as in most other portions of the spectrum (Figure 6).  There were also small differences in the magnitude of variation between azimuths.

Figure 5.  The planimetric and orthographic representations of BRDF data for the Pleurozium surface measured in the laboratory.  The hot-spot is more of a band of higher reflectance in the southern portion of the plots.

Figure 6.  A full-spectrum representation of the Pleurozium data at a constant 40° zenith for different azimuths.

 

CONCLUSIONS

The development of this low-cost field and laboratory ULGS goniometer system allows for the acquisition of BRDF data under a variety of conditions in support of remote sensing studies.  The actual quality of an individual spectral measurement is a function of the spectroradiometer sensor used, and largely independent of the goniometer platform.  Accordingly, despite the low-cost of the ULGS, it still yields very high quality spectra.  The application of a low-cost approach has yielded usable data with this system as demonstrated by the field and laboratory examples presented in this paper.  The primary difference between this system and others currently in use is the relative simplicity and greatly reduced cost (ULGS - $1000 compared to $400 000 typically for SFG) of the device when compared to the computer operated style of goniometer. 

            The angular sampling resolution of this instrument can be adjusted.  At 10° the BRDF data showed a lot of detail not always present in lower resolution studies, but made measurements in field situations difficult.  For field use, the reduction of the angular resolution to 20° should provide a good balance between detail and time of acquisition.  The frequent calibration of the instrument with the reference panel, while increasing field acquisition time, allowed for the changing illumination conditions (diffuse irradiance, aerosol optical thickness etc.) to be accounted for during acquisition. 

            The increasing availability of multiangular orbital satellite data and the increasing use of spectroradiometers for field validation of remotely sensed data require the collection of BRDF data for a wide variety of ground conditions.  The development of instruments capable of estimating the BRDF in a low-cost fashion will improve the availability of these data and increase our knowledge of the reflectance characteristics of the earth’s surface.

 

Reference:

 

Peddle, D.R., H.P. White,  R.J. Soffer,  J.R. Miller  and E.F. LeDrew, 2001.

Reflectance processing of remote sensing spectroradiometer data. Computers  & Geosciences.  Vol. 27, pp. 203-213.<p>