Infrared Image-forming System

Instrument Components and Performance

The basic components one considers in designing a radiometer (once the application has been determined) are:

Performance of an instrument is characterized by its:

Instrument Design Considerations

Diffraction

When an image is formed, the energy is not focused on a point but is spread over an area, forming a diffraction pattern. Diffraction occurs when physical limits are imposed on the radiation, as the energy passes through a hole it spreads out. Diffraction defines the ability of a radiometer to resolve to point sources. Even though a cloud may not be in the dircet FOV of the sensor, its presence can affect the measured radiance because of diffraction. is also important when seeking clear FOVs in the vicinty of clouds.

Impulse Response Function

The impulse, or step, response function characterizes the sampling time over which the detector accumulates a voltage response. The impulse response determines how sharp edges appear in the image.

Signal to Noise

We want a large signal to noise ratio so that we have a "clean" image. Noise Equivalent Radiance (NEDR) of an infrared detector is a function of the detector area, the mirror apperature, the dwell time, the FOV size, the spectral bandwidth, and other optical and electrical instrument characteristics. Signal-to-noise is also expressed as the Noise Equivalent Temperature (NEDT) at a specific temperature.

Calibration

Calibration converts the dector output to a known radiance, allowing quantitative use of the measurements. In a linear system the output voltage is given by

V = R + V'

where V is the target output voltage, R is the target input radiance, is the radiometer responsivity and V' is the system offset voltage. Calibration consists of determine and V'. In the IR, this is accomplished by viewing two blackbodies (or space and one blackbody) of known radiance.

Bit Depth

The observed energy in a given spectral bandpass must be converted to a digital count of a fixed bit depth. For n bit data, the radiance must be convereted to 2n even increments. This introduces truncation error, limiting the precisioin of the data.