Sensor Calibration

Spaceborne/Airborne Sensor Calibration

Vicarious calibration is an important tool used to monitor temporal responsivity changes that may occur in spaceborne and airborne Earth-observing sensors. The term vicarious generally refers to calibration methodologies that are not onboard the sensor platform. It has been used successfully to monitor such satellite sensors as Landsat-5 TM, Landsat-7 ETM+, Landsat-8 OLI, ASTER, MODIS, MISR, Ikonos, Quickbird, Hyperion, ALI, and RapidEye, and airborne systems such as AVIRIS, MASTER, and MAS. Vicarious calibration is typically performed using ground-based methods, lunar irradiance measurements, and cross calibration with other well-understood sensors.

The reflectance-based approach is a ground-based method that has been used successfully since the 1980s by the Remote Sensing Group (RSG) at the University of Arizona. It involves measuring the surface reflectance and atmospheric properties at a ground test site during an overpass of the sensor under test. The measured surface and atmospheric properties are used in a radiative transfer code to predict the band-averaged top-of-atmosphere radiance for each spectral band. The results produced by the radiative transfer code are compared to the DNs (or spectral radiance values) extracted from the image produced by the sensor under test using image-processing software.

The surface BRF is measured using a portable spectroradiometer that measures reflected radiance from 350–2500 nm. The instrument is carried over the test site in a well-defined pattern. It views a reference panel at regular, well-defined intervals during the surface reflectance collection process. The ground BRF is determined by taking the ratio of DNs produced when viewing a panel versus viewing the ground. The spectral BRF of the panel is measured in the laboratory on a regular basis to monitor any changes that may have occurred. This also allows long-term changes in the panel to be monitored for future reference. Atmospheric transmission is the second input to the radiative transfer code, and must be measured before, during, and after sensor overpass.

Solar transmission is measured using an automated solar radiometer that has ten discreet spectral bands between 380 and 1030 nm. The output voltage is linearly related to the in-band solar irradiance, and is used to determine such atmospheric parameters as aerosol size distribution, water vapor, and columnar ozone. The top-of-atmosphere radiance measured by the sensor under test is modeled by RSG using a radiative transfer package based on a modified version of MODTRAN. It uses input parameters such as the illumination and viewing geometry, surface BRF, meteorological conditions, and atmospheric properties.

Laboratory Sensor Calibration

Laboratory calibrations are performed by the RSG both for validation of in-house instrumentation and also as a service for customer sensors, especially those built here at RSG. Generally speaking, these calibrations are radiance calibrations and they can be performed a few different ways. One method makes use of a NIST calibrated FEL lamp. This lamp is aligned at a well-known distance and the light is then reflected off of a known reflectance standard which is viewed by the radiometer at an angle of 45°. Using the calibration of the lamp, the radiance viewed by the sensor can be directly computed. An irradiance calibration is achieved similarly with the sensor directly viewing the lamp. Another method makes use of an integrating sphere as the radiance standard. These standards, however, must be characterized by a calibrated radiometer in order to predict the radiance output from the sphere. RSG has calibrated radiometers which can perform this prediction out to 2.5 μm.

There are many additional relevant measurements associated with these calibrations. Hyperspectral filter transmission, linearity testing, field of view measurement, and repeatability testing are a few of the necessary additional tests necessary to characterize an instrument and calculate an accurate calibration coefficient.