next up previous
Next: FASCOD3P Use Up: Abstract and Contents Previous: AERI Microwindow Regions

Return to the Publications. Return to the Index.

Brightness Temperature Linear Fit


The calculation of a brightness temperature within a spectral bandpass is inconsistent with the true temperature; where the error increases with larger bandpass. This is a result of solving


based on a single wavenumber representation of the given bandpass.

A correction can be applied by assuming a linear representation of the temperature in the Planck function, such that


where a and b are the y-intercept and slope, respectively, determined with a least squares fit over an expected temperature range using the central bandpass wavenumber, tex2html_wrap_inline3425.

For ground-based measurements of downwelling radiance in the atmospheric window, a typical brightness temperature range is between 150 and 270 K; dependent upon cloud base altitude and optical depth. The theoretical radiance, integrated over the spectral bandpass, is calculated in 1 K increments in the given temperature domain. Brightness temperature, tex2html_wrap_inline3427, is determined using the theoretical radiance and tex2html_wrap_inline3429. A least squares fit to these values is performed to yield the necessary adjustment; where the corrected form is represented as


Figure 37 illustrates the relative error (upper figure) and percent error (lower figure) in deriving the brightness temperature from radiance data in a spectral bandpass. Corrected and uncorrected values are represented by the solid and dashed curves, respectively, using a bandpass of 909 to 1000 cm tex2html_wrap_inline3431 (10 to 11 tex2html_wrap_inline3433m) over the temperature domain 150 to 270 K.

Figure 37: A comparison of broadband brightness temperature using parameterized Planck function linear in temperature (solid line) and standard Planck function (dashed line); each using the mean wavenumber, relative to the theoretical value. The upper figure represents temperature difference from theoretical, while the lower gives percent error.

Daniel DeSlover
Sun Aug 11 10:02:40 CDT 1996