navigation image mapnext pagetable of contentsprevious page

Field Instruments and Measurements; Data Collection Platforms

Another major phase of ground truthing involves getting detailed and more complete spectral signatures using spectrometers, spectrophotometers, and radiometers in the laboratory, the field, or from aircraft or manned spacecraft. Although subject to the caveats mentioned earlier about signature extension, these controlled measurements are invaluable in defining reference signatures and designing sensors to better discriminate a greater variety of classes. Signatures obtained in the laboratory are normally obtained for single materials (pure substances).

A typical lab spectrometer generates illumination that irradiates a sample mount with diffused light scattered from a metal-plated hemisphere that integrates over an area measured in steradians. Light reflected at an angle from the sample surface collimates as a beam through a hole in the hemisphere. A chopper (rotating signal divider) interrupts the beam by alternately passing and blocking the light that goes to a grating and/or prism, which disperses the radiation into its spectrum. A second beam from a reference source of known and constant intensity reaches the dispersing element during the blocking phase. The dual beam signals then go to a detector that scans them (i.e., moves through the dispersion angles) to measure reflectances (as intensities) as a function of wavelengths. The signals are amplified and then plotted on an X-Y recorder. We display these signals as the ratio of sample reflectance to the reference output, to derive total reflectance (specular and diffuse components). Some instruments can vary the angles of incidence and observation to derive bidirectional reflectance, in which intensities vary with the angles selected (displayed in a series of curves).

Operating the spectrometer in a near vacuum eliminates the effects of absorption bands in the atmosphere. Using a lab spectrometer, a series of fixed reflectance measurements on some typical rocks from Wyoming, first introduced in Section 2, appear again below. Note the absorption bands at >2.0 mm associated with mineral constituents and pore water.

Diagram showing a series of fixed reflectance measurements on some typical rocks from Wyoming.

In recent years, the term hyperspectral has come to apply to spectral curves that are either continuous over a broad range, such as the Wyoming group, or consist of a large number of individual narrow-wavelength (high spectral resolution) channels that are so closely spaced that they constitute an almost quasi-continuous spectrum. Until the last decade (see AVIRIS described below), it was technically very difficult to operate a spectrometer from fast-moving air and space platforms, because the instrument was unable to dwell on a small target (IFOV) long enough to scan the full spectrum. This limitation is the main reason Landsat, SPOT, and other sensor systems have had to use broad wavelength bands that integrate the variations in spectral intervals into single values for the reflectance ranges they represent.

Using spectrometers and related instruments in the field has the advantage of looking at surfaces that contain the mix of components that make up the classes of usual interest (remember the usual constituents in a field crop). We can often obtain spectra for each component and then back away to get the full mix. This collection of component spectra assists in interpreting the mix response. Also, illumination conditions from solar irradiation, plus diffuse skylight and multiple reflections from ground surroundings that can contribute 10% to 25% of the total, during sunny or even overcast conditions, are best in the outdoors.

13-11: Suggest three other advantages to acquiring spectral data in the field. ANSWER

One of the simplest devices is a hand-held two- or three-band radiometer, such as the one held by the writer in this illustration:

Color photograph of the author using a hand-held 2 or 3 band radiometer.

The instrument has its own portable power source and recording system. The spectral bandwidth is typically 0.05 to 0.10 mm. Common channels are in the green, red, and near-IR. These are especially pertinent to calculating Vegetation Indices (Section 4).

13-12: Specify narrow band wavelengths for vegetation detection using a two or three band radiometer (if you have forgotten what vegetation signatures look like, check Section 3). ANSWER

We can take readings at various look and solar angles, and on different dates and times of year to provide records of spectral variations in the same test areas. These variables can have a pronounced effect on the character of the spectral response, and hence the interpretation of changes, as indicated in this plot of bidirectional reflectances, showing the variation of IR/red-band radiances as a function of view angle and azimuth (compass) direction:

Plot of bidirectional reflectances relating the IR/Red radiance ratio and the view angle (in degrees).

13-13: Formulate a generalization about the implications of these curves. ANSWER

In general, spectra of rocks show much less variability, because of bidirectional reflectance effects, than does vegetation. For forests, irregularities in canopy shape, leaf or needle shapes, and species mix can have notable influence on response as a function of viewing and illumination geometry. Portable field spectrometers are now in common use. The next image shows a typical setup for a reflectance spectrometer developed by the Jet Propulsion Lab (JPL) in the mid-70s:

Color photograph of a portable reflectance spectrometer.

For this instrument, an optical head gathers the reflected light and passes it through a filter wheel operating between 0.4 and 2.5 mm onto a cooled detector, made of lead sulphide (PbS). The backpack contains a power source, amplifier, and recording (analog to digital) assembly. After the ground target scan (30 seconds or less) is complete, they quickly repeat the scan on a a flat reference plate made of a high reflectance (near white) material. Sometimes they use a black plate as well, to fix both ends of the reflectance range. Over a brief time span, the scene lighting remains about the same, but over longer periods, changes in clouds, sun angle, etc., cause variations in spectral response. However, they normalize the spectra by dividing the target readings with the reference values. The assumption here is that variations in irradiance over a series of paired readings taken minutes to hours apart cancel, since differences in the illumination conditions affect both target and reference in the same way at each sampling time . For example, if at time 1, at some wavelength, the target reads 30 and the reference 90 and at time 2 they read 20 and 60 (percent), the normalized spectra are both 0.33 (33%) in reflectance units. In the next image, we show field (in situ) reflectance spectra of rocks (1-5) and ponderosa pine (6) acquired by this instrument:

Diagram illustrating the reflectance spectra of rocks and ponderosa pine acquired in the field by a reflectance spectrometer.

13-14: These spectral curves show much less structure (peaks and troughs) and less amplitude (intensity) than the laboratory-produced curves for the Wyoming rocks (above). Explain this difference. ANSWER

JPL engineers, leaders in developing ground truth instruments, developed a portable field emission spectrometer that uses argon-cooled, mercury-cadmium-telluride (HgCdTe) detectors to sense thermal IR responses in the 5 to 15 mm spectral region. In keeping with the trend toward using linear array multispectral systems (e.g., SPOT), instead of scanners with filters (Landsat), some field spectrometers today use a fixed grating that spreads radiation over a range of angles onto array detectors made of indium-gallium-arsenic (InGaAs) alloys capable of more rapid scanning and greater sensitivity. For further insights into field spectrometry, consult the review found on the Home Page prepared by Analytical Spectral Devices, Inc., a company founded by Dr. Alexander Goetz (formerly at JPL), a leader in this field.

Another approach is to operate a spectrometer from a truck in which we mount the sensor head on a movable cherry picker, as illustrated below. This allows us to vary the Instantaneous Field Of View (IFOV) height, so that we can examine larger surface areas.

Color photograph of a reflectance spectrometer mounted on a truck's cherry picker.

Often we obtain the most valuable supporting data from sensors mounted on aircraft that fly over study areas. In remote sensing programs administered by NASA, investigators specify test sites for acquiring ancillary data or system developers do to test prototype (breadboard) instrument and sensor designs, proposed for future missions. Such research missions help to determine the spectral and spatial resolutions, the signal-to-noise (S/N) response, and the time of day and year that optimize detecting and identifying. We show here part of an earlier fleet of airplanes used for these purposes:

Color photograph of several airplanes used to carry various sensors.

The large plane on the left is a Lockheed Electra that operates up to 7,600 m (25,000 ft). On the right is a C-130, capable of higher altitudes. The two center planes are U-2's (designated ER-2s) that can reach to 19,000 m (62,000 ft). Not shown is the RB-57, another jet capable of high flight levels. The small jet inside the group is primarily for support, whereas the jet near the hanger has sensors for lower-altitude missions. Among the compliment of sensors are one or more film camera systems (including multiband arrays), multispectral scanners (including those using charge-coupled detectors [CCDs]), thermal-IR scanners (such as the Thermal IR Multispectral Scanner [TIMS], see Section 9), microwave sensors (including radiometers and scatterometers, and multiband radar), and special request equipment not routinely flown.

At the other extreme, we often need to collect continuous data on the ground at widely separated stations or over extended time periods, often in inaccessible areas. Costs from repeat trips and other factors may preclude sending field parties after initial visits. With such requirements, we set up automated, remote-sensing, sampling sites, at which we measure several defining properties constantly or at fixed intervals.

To accomplish this during the Landsat program, we deployed Data Collection Platforms (DCPs) to measure certain properties on site, coding the results and transmitting these by radio whenever Landsat or some other satellite was in line of sight, and then relayed the data to appropriate ground stations for processing. Typical remote field measurements include: 1) stream heights and velocities; 2) silt loads; 3) snow pack densities; 4) meteorologic parameters; 5) point source pollution; 6) seismic disturbances, and 7) surface tilt on volcano slopes. We now commonly use networks of remote stations in the U.S. and other countries, such as this example of a stream hydrograph and its transmitter:

Color photograph of a Data Collection Platform.

13-15: Cite three other plausible uses for DCPs - these need not be directly pertinent to Landsat-type observations. ANSWER

We have demonstrated so far in this tutorial that remote sensing is an efficient way to gather large amounts of information from vast areas without being on the observed surface. But, interpreters will seldom apply this knowledge effectively unless they have first-hand familiarity with the surface of interest, or at least, with models of the surface. They gather this intelligence by several means: from circumspect field observations, judicious investigations at training sites, sophisticated measurements in the laboratory, on the ground, in the air, and from space, and, ultimately, from a rigorous mathematical analysis of the data to test for validity and correlation.

navigation image mapnext pageprevious page

Primary Author: Nicholas M. Short, Sr. email:

Collaborators: Code 935 NASA GSFC, GST, USAF Academy
Contributor Information
Last Updated: September '99

Webmaster: Bill Dickinson Jr.
Site Curator: Nannette Fekete

Please direct any comments to