GOES-12 (Channel Radiometer)
• Channels are typically independent of each other• Need to know each channel’s
– Spectral response function– Noise characteristics
How an Interferometer Works
1. Move one mirror slowly back-and-forth to create an interference pattern (interferogram) at the detector
2. Record the inteferogram as a function of time (or mirror position)
3. Apply a FFT to the interferogram to yield the spectrum
AERI Interferometer AssemblyAERI Interferometer Assembly
BomemInterferometer
ABB
HBB
OpticsOpticsBenchBench
ShockShockMounts (4)Mounts (4)
Interferometer / AERIInterferometer / AERIElectronics Interface BoxElectronics Interface Box
IR DetectorIR DetectorDewar withDewar withCooler Cold FingerCooler Cold Finger
Stirling CoolerStirling CoolerCompressorCompressor
Front End AssemblyFront End AssemblyBlackbodiesBlackbodiesScene Mirror AssemblyScene Mirror AssemblyForced Air InletForced Air InletRain SensorRain SensorSun SensorSun Sensor
Front-endFront-endCloseoutCloseout(thermal)(thermal)
Knuteson et al., JTECH, 2004
Passive IR Satellites In Space
• Wave of future is high-spectral resolution IR remote sensing
• Fourier transform spectrometers (FTS)• Examples:
– Infrared Atmospheric Sounding Interferometer (IASI) on METOP
– Cross-track Infrared Sounder (CrIS) on NPOESS– Geostationary Imaging Fourier Transform Spectrometer
(GIFTS)
Weighting Function
This would be ideal This would
be nice This usually what we get
Retrieving Temperature Profiles
• Signal in different channels is highly correlated due to vertical spread in the weighting functions– Typically have only a few “independent pieces of
information” in the observations
• Multiple temperature profiles (solutions) yield the same observed radiance– Underdetermined, or ill-defined, problem
• Instrument noise further complicates matters
Retrieval Algorithms• Algorithm must be able to handle ill-conditioned problem with
noise• Two general approaches:
– Statistical: use a priori data to generate regressions to relate radiance to T(z) profile
• Easy to develop• Handles noise well• Computationally fast
– Physical: iterative approach whereby a forward RT model is used to derive T(z) profile
• Need a priori data to help constrain solution• Computationally slow• Provides error bars as part of the retrieval
Ground-based IR Profiling Capability
Warm & humid mid-lat clear sky case
Cold & dry mid-lat clear sky case
Example Results from Mid-latitude Site
winter: cold and dry summer: hot and humid