search

FIDUCEO / FCDR_HIRS

Other
1 stars 2 forks source link

Derive cutoffs for outlier detection more formally #194

Open gerritholl opened 6 years ago

gerritholl commented 6 years ago

Currently, my Earth count MEDMAD outlier filter uses a cutoff of 10. That lets some "mild outliers" through. If I set it too low, I might flag away some real masurements. I need histograms per channel, probably per channel per month, of

|C_E-MED|/MAD

perhaps along with some statistics of brightness temperatures.

gerritholl commented 6 years ago

Need to describe how I do outlier detection.