Signal Extraction Part II: FGS1 Data Reduction
Building on the success of AIRS-CH0 signal extraction, let’s apply the same intelligent data reduction approach to the FGS1 guidance camera data. The goal is to identify and extract just the signal-bearing pixels from the 2D frames, reducing data volume while preserving the exoplanet transit signatures.
Checkout the FGS1 signal extraction notebook on GitHub.
1. FGS1 signal structure
Unlike AIRS-CH0’s spectral strips, FGS1 frames contain a more compact signal region that extends across both rows and columns. The signal analysis reveals a clear pattern:
The signal is concentrated in a roughly square region around the center of the detector, with the brightest pixels clustered between rows and columns 10-20. This makes sense for a guidance camera - the star’s point spread function creates a compact brightness distribution on the detector.
2. 2D signal extraction approach
The extraction strategy extends the 1D row-based approach used for AIRS-CH0 to work in two dimensions:
- Identify bright rows: Find the top N rows with highest total signal
- Identify bright columns: Find the top N columns with highest total signal
- Extract signal block: Select the intersection of bright rows and columns
- Sum to single value: Collapse the extracted block to one brightness measurement per frame
This creates a “signal block” rather than a “signal strip”, capturing the 2D nature of the FGS1 point source.
3. Extraction results
The results demonstrate excellent performance with dramatic data reduction:
Key findings:
- Transit preservation: The extracted signal shows virtually identical transit signatures to the full frame data
- Data reduction: Using a 6×6 pixel block achieves ~98% data reduction while preserving signal quality
- Signal consistency: The extracted signal maintains the same relative brightness and noise characteristics
- Processing efficiency: Dramatically smaller datasets enable much faster downstream analysis
4. Optimal extraction parameters
Testing different block sizes reveals that a 4×4 to 6×6 pixel extraction region hits the sweet spot:
- Smaller blocks risk losing signal from the point spread function wings
- Larger blocks start including more background noise than useful signal
- The 6×6 approach provides robust signal capture with excellent noise rejection
5. Implementation ready
The FGS1 signal extraction approach mirrors the adaptive threshold method developed for AIRS-CH0, making it straightforward to integrate into the preprocessing pipeline. The two-dimensional extraction naturally handles variations in star brightness and detector response across different planets.
This completes the signal extraction development - we now have intelligent data reduction methods for both instruments that preserve transit signals while achieving ~98% data volume reduction. The next step is integrating these methods into the full preprocessing pipeline for production use.
Just like with AIRS-CH0, sometimes the best signal processing isn’t about more sophisticated algorithms - it’s about intelligently identifying and keeping only the data that matters.