I have 3 time-series data.

  1. NDVI(normalized difference vegetation index) mean
  2. Precipitation
  3. Temperature

    All of these have their own unit.
    Now I want to find similarity/correlation between NDVI and precipitation, NDVI and temperature. Basically, my aim is to find "NDVI is more correlated with precipitation or temperature". Should I normalize precipitation and temperature to NDVI values?

I recommend you to scale all three time series to unitless values, with zero mean and standard deviation of 1. After that, you can search for correlation among normalized time series. Python has StandardScaler function that could help with it.

  • $\begingroup$ NDVI values are between (0,1). So can I scale other values to (0,1)? Because I read somewhere that Standardization(zero mean and standard deviation of 1) is only recommended when data follows Gaussian distribution. $\endgroup$ – sankalp Feb 7 '19 at 10:04
  • $\begingroup$ If you want to apply the Pearson correlation, standardisation is needed. You can use other methods to find similarities, e.g. ranking statistics (like Spearman), that are independent to scale. $\endgroup$ – Viacheslav Komisarenko Feb 7 '19 at 13:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.