SMAP Maps: Radiometer RFI
The maps on this page show the percentage of SMAP samples that were flagged as radio frequency interference (RFI) and removed during data processing.
The algorithm used to flag RFI is described in (Piepmeier et al., 2014; Piepmeier et al., 2016). This algorithm uses several criteria to identify RFI: time-domain detection; cross-frequency detection; kurtosis detection and polarimetric detection.
These criteria are applied to both the full-band data (each sample corresponds to 24 MHz bandwidth and 300 s of integration time) and the sub-band data (16 bands of 1.5 MHz bandwidth and 1.2 ms of integration time) produced by SMAP. Each footprint is represented by a spectrogram. The footprint, which is the fundamental data block in the L1B_TB products, consists of 32 full-band samples or, equivalently, 128 sub-band samples (16 frequency x 8 time samples). The fraction of this spectrogram flagged by any of the RFI detection criteria determines the "RFI percent" for that footprint. Since the flagged samples are discarded before obtaining geophysical parameters (e.g., brightness temperatures), the RFI percent also corresponds to the quantity of data that is lost due to contamination from RFI signals.
To prepare these maps, the RFI percentages (for both V- and H-polarization) of every footprint are binned into a 0.25x0.25 degree grid and every grid cell is filled with the mean value of the RFI percent of all the footprints whose center is within that cell.
The false alarm level is about 5% (i.e., averaging over a region expected to be RFI-free leads to an RFI percent of about 5%). The algorithms have been tuned to avoid detecting changes associated with antenna temperature gradients at coastlines, but the effects of the water-ice boundary can be seen as a faint line near the ice edge around Antarctica.
Click on the images (below) for a closer view.