What stage of processing requires the linear to/from dB command?

s1tbx
#1

None of the tutorials i looked at seem to use it - where/when is it useful?

Thanks in advance,

Conor.

1 Like
#2

It log-scales intensity data (dB=10*log10()), which can be useful for visualisation and/or data compression (e.g. I scale dB output to byte to reduce output with a factor 4, mostly for visualisation of time series). Calibrated data has a limited range in dB (e.g. -30 to 3 dB for VV land scenes) which is easier to scale than corresponding 0.001 - 2 range in linear values).

2 Likes
#3

ok great - thanks.

So i am not missing anything if I do not apply that, other than a potential increase in performance and also some minor differences due to compression?

#4

Depends what you intend to do afterwards. If you intend to apply further steps (calibration, multi-looking, geocoding, speckle filtering etc.), then do NOT do linear to dB conversion first (it changes the statistics and does not make mathematical sense). It is typically the last step you want to apply (i.e. visualization, compression).

#5

if we use **10log10(sigma0) for db conversion from linear , why sometimes **20log10(sigma0) is used instead?

Thanks

1 Like
#6

maybe some folks start from amplitude instead of intensity data. Amplitude*Amplitude = intensity, hence 20. But then it should not be called sigma0, but that may be just laziness.

4 Likes
#7

I have done GRD S1 Calibration to Beta0 and etc… and at the final step I did Range Doppler Terrain Correction for my time series data sets so the final outcomes are Gamma0.
So I have a question, should I convert the data sets to db as far they are Calibration to Beta0?

#8

This depends on what you use the data for. Please have a look at these considerations:

1 Like
#9

Is it correct to apply linear to db after Range Doppler Terrain Correction?

As far I am dealing with intensity should I apply 10*log10(DN) for db?

#10

It is valid at any stage after calibration I’d say.
As Terrain Correction involves resampling of your data I would personally even perform db conversion before that.

Your equation is correct. Your can also right-click at the band and select convert to db.