Amplitude spectrum of temperature fluctuations
 

Hello Friends,

I am performing LES of T-junction mixing flows at high temperature differences. The amplitude spectrum of temperature fluctuations in the near wall region is a useful way of looking at my results. To do this, I do the following things

1. Take the temperature fluctuation data, do the fft and divide it by length of data. (In this case, A = fft(T')/L, where L = 32768)

2. Then, taking the magnitude of only half the data, B = 2*abs(A(2:L/2)), excluding DC and Nyquist frequency.

3. My LES recording frequency, Fs = 5000. Therefore, f=(Fs/2*linspace(0,1,L/2))';

4. Then, a loglog plot of B vs f.

Here is what I don't understand
(i) What is the unit of B, i.e., the amplitude of temperature fluctuations? Is it only Kelvin (K) or Kelvin/Hz?
(ii) To do a power spectrum plot, should I just take square of magnitude, say,
B = abs(A(2:L/2))^2 and plot it against frequency? In this case what is the unit of B?
(iii) Finally, I would like to compare my experimental data with LES data for amplitude spectrum. My experimental recording frequency is 100 Hz, whereas my LES recording frequency is 5000 Hz. Is there any mathematical function or method to increase the frequency to experimental data?

I am not able to understand conceptually what amplitude spectrum and power spectrum signifies, in spite of using it in my work repetitively? Any help (or reference to an internet link) in making my understanding better would be appreciated. Thank you in advance.