I added a feature that lets you compare and experiment with RFI removal algorithms. First, drill down to a waterfall plot. A typical one is here. You can see that there's drifting RFI.

At the bottom of the page, there's a form that lets you choose RFI algorithms and their parameters. Click OK at the bottom. You'll see that the RFI has been removed.

Uncheck the "Drifting" checkbox and click OK again. The RFI is back.

You can experiment with the parameters of the drifting algorithm and see what works best (the default parameters seem to work pretty well).

This interface, in addition to being informative, will greatly accelerate the process of evaluating new RFI removal algorithms.

Thanks for the update. Going to try that out...

After some experimentation it looks like the default settings remove several spikes that don't seem to have a direct connection to the RFI. (frequency higher and lower than the obvious RFI in the middle)

I tried tuning down the drifting RFI-detection parameters as low as possible, however this might only fit this specific and obvious RFI example ;-).

Things that surprised me most:

• the min dt parameter doesn't seem to have any effect. I expected that drifting RFI-removal must fail completely, if I set min dt >= max dt
  
• I was able to set the threshold really, really low before some of the RFI started appearing again, much lower than default setting
  

On the other hand, there seems to be a lot of mutibeam RFI (pulses appearing at the same time across all frequencies) that did not get filtered with the default settings, so I tuned up the frequency scale parameter until it looked like most of this got removed.

These are the settings I came up with in the end:

https://setiathome.berkeley.edu/nebula/waterfall.php?action=explorer&t=2455525.319763&dt=0.01&f=1418998711.1092&df=10000&zone=on&multibeam=on&tscale=11&fscale=45&lo_chirp=on&drifting=on&min_dt=300&max_dt=300&nbins=6&thresh=1.0E-16&min_dist_beams=1.5&max_drift_rate=5&spike=on&gaussian=on&pulse=on&triplet=on&autocorr=on

A nice-to-have would be, if the zoom buttons on the waterfall plot would not reset the rfi explorer to its default settings.