Reference Measurements: The What, Why, When and How

Introduction

Accurate field spectral reflectance measurements are important in the calibration and validation of airborne and space-borne hyperspectral and multispectral imagery. The field portable ASD FieldSpec® 4 spectroradiometer was designed with these requirements in mind. However, even the best instrument can deliver inaccurate data if the measurements are not carried out properly.

One of the most important actions a researcher takes in the field is frequent reference measurements. This paper serves to explain what a reference measurement is in relation to reflectance measurements, why reference measurements are important, and how to properly perform them in the field.

What is a reference measurement?

Before discussing reference measurements, it is important to provide a definition of a reflectance measurement. Reflectance is not a direct measurement. A reflectance measurement is actually a ratio derived from two separate radiance measurements. It is calculated by dividing the radiance of a sample by the radiance from a reference panel of known reflectance, usually having nearly 100% reflectance. This reflectance panel measurement is known as a reference measurement. This relative reflectance spectrum is then sometimes corrected to absolute reflectance by multiplying it be the known reflectance of the reference panel.

Why is a reference measurement so important?

The solar spectrum contains many absorption features, the strength of which are a function of solar elevation angle, altitude, and atmospheric conditions (Fig. 1). By ratioing the radiance of the observed sample to the radiance observed when viewing the reference panel, these features are removed from the resulting reflectance spectrum. An Inherent assumption in this method of ratioing radiances is that the solar irradiance and light scattering effects from the local surroundings is the same for both the sample and reference measurements.

To obtain an accurate reflectance spectrum, the irradiance illuminating the reference panel must be the same as the irradiance illuminating the sample. Any change in the irradiance between the measurement of the reference panel and the sample will result in atmospheric absorption features being imprinted on the measured reflectance spectrum. This is what drives the need to frequently re-measure the reference panel.

Figure 1. Solar radiance reflected from a 99% reflective reference panel. The overall shape of the spectrum is a 5780°K black body with absorption features associated with gases in the Earth’s atmosphere.

What can cause the irradiance to be different between sample and reference measurement? 

The most important time varying influences on irradiance are the solar elevation angle and the intervening atmosphere. Changes in these parameters dictate how often a measurement of the standard should be performed.

The geometry of objects in the vicinity of both the target and reference panel are also important. The position of your body relative to the reference measurement and the sample measurement should be the same. Differences in collection orientation, back scatter from clothes and local environmental surroundings, and shifting sun elevation between sample and reference measurements all can lead to irradiance variances.

Think about how the reference panel is illuminated relative to how the sample is illuminated; ideally they should be the same.

When should a reference measurement be taken?

The frequency with which reference measurements should be taken is totally dependent on atmospheric conditions at the time samples are being taken. Even on a clear day when conditions seem stable, the sun angle and atmospheric moisture levels are constantly shifting. In the following example, a FieldSpec spectroradiometer viewed a fixed reference panel and a reference measurement was performed. All subsequent spectra were ratioed to this measurement. Figure 2 shows two such subsequent collections. The black curve was collected immediately following the reference measurement and the red curve was measured five minutes later. With high total atmospheric water vapor and relatively stable atmospheric conditions, irradiance changes are minor at wavelengths outside of the major atmospheric water vapor absorption bands centered near 1400, 1800, and 2500 nm. Reference measurements taken once every 10 or 15 minutes are usually sufficient in these atmospheric conditions.

Figure 2. Effects of irradiance changes over a five-minute period at solar noon minus one hour in relatively stable atmospheric conditions.

The data displayed in Figure 3 indicate the presence of less stable conditions. Again, a FieldSpec spectroradiometer viewed a fixed reference panel and a reference measurement was performed. All subsequent spectra were ratioed to this measurement. The black curve was collected immediately following the reference measurement, the red curve was measured five minutes later, and the green curve was collected after an additional five minutes. Large changes in the depths on the minor water vapor features are visible, particularly at 10 minutes after measuring the reference panel, indicating that atmospheric water vapor was much more variable than the conditions present in the previous example.

Figure 3. Effects of irradiance changes over a five minute period at solar noon minus three hours 

With intermittent cloud formation and the presence of high cirrus clouds, which are often not visible, much more frequent reference panel measurements are required. In worst case scenarios with rapidly changing conditions, a reference measurement should precede each sample measurement. The goal is to keep the time gap between reference and sample measurement as small as possible to ensure the irradiance is the same between reference and sample.

Another consideration is the required accuracy of the reflectance measurement: the more modest the requirement, the longer the interval available between reference measurements.

How is a reference measurement taken?

Also, a reference measurement is best accomplished by measuring a reflectance standard close to the area in which the reflectance measurements are taken. This measurement integrates the irradiance from all the components of the radiation field including that scattered from surrounding objects.

1. Level the panel
2. Point the fiber optic at the panel
3. Optimize the instrument (this sets all the parameter in the detector electronics to match the dynamic range of the ambient light source, be it artificial or solar)
4. Collect the scans corresponding to the reference panel.

Noise reduction technique for reference measurements

Set up the instrument so that the number of scans it is averaging when it looks at the reference panel is around 2x the number of actual sample scans. The reason for that is that the noise you see in the resulting ratio (sample spectrum/reference panel spectrum) is the square root of the sum of the squares of the noise you see in each of your two spectra. If you are collecting on a day when you only have to take a reference measurement once every five or 10 minutes, why not spend a little more time to get the noise in that portion of your measurement down a little bit more?

Summary

In order to minimize changes in irradiance between the measurement of the reference panel and sample, maintain an awareness of the factors influencing that irradiance:

• Solar elevation angle changes
• Changes in water vapor
• Changes in cirrus cloud cover
• Differences in the scattering geometry of surfaces in the vicinity of samples.

Adjust the frequency of reference panel measurements to accommodate current conditions. With more variable conditions, increase the frequency of reference panel measurements.

The instrument itself is far more stable than the atmosphere, and, if it has been running for an hour or so, will nominally be stable to 0.1 percent.