Brief documetation of the RCPM output


Statistical output

File types

There are usually five different file types, co?, st?,cov,tst and kpf. "?" represents the numbers 0 to 5 (see below). For two of the file types the last number indicates the state of the post-processing:

Steps of the post processing
step
 data source
0
 output from the un-changed raw data
1
 output after despiking data
2
 output after flow distortion correction (if selected) + transformation to a right handed coordinates system, if necessary
3
 output after two rotations (average z =0 average y = 0)
4
 output after linear detrending
5
 output after covariance maximisation

The file is named after the date of the first raw data file, the ID from the site info configuration file

Variable descriptions

The records of all these files start with the variables rnam,NoObs, i.e. the name of the raw data file and the number of observations. The name of fthe raw data file is the only information that relates the rcpm output to these files. 
  1. co means covariances, they are named like cov_Sens1_Sens2_?. The file contains additionally covariances that have been corrected for low pass filtering, if applicable. Those are tagged by _LPFM or _LPFT. The correction factors (e.g. LPFT_Sens1_Sens2_?) is also given.
  2. st contains monovariate statistical output.  The following list gives an example for the sensor x_43m at post processing level 5:
Mono variate statistics in files *.st?
Variable
 Explanation
x_43m_Lag_5 The lag for X_43m
x_43m_NoSpikes_5 Number of spikes
x_43m_NoDrops_5 number of drops (these are consecutive values that are constant )
x_43m_NoSteps_5 number of relative steps, i.e. staeps that exceed a given multiple of the local standard deviation (this routine is in a beta testing stage)
x_43m_NoAbsSteps_5 number of absolute steps, ie.e steps that exceed a certain range (this routine is in a beta testing stage)
x_43m_LP_sdev_5 the low passed standard deviation (the window for segmentation of the data is to be given in a configuration file)
x_43m_min_5 the absoulte minimum
x_43m_max_5 the absoulte maximum
x_43m_ave_5 the average
x_43m_adev_5 average deviation or mean absolute deviation (numerical recipes in c)
x_43m_sdev_5 standard deviation (numerical recipes in c)
x_43m_var_5 variance (numerical recipes in c)
x_43m_skew_5 skewness  (numerical recipes in c)
x_43m_curt_5 kurtosis (numerical recipes in c)
x_43m_a_5* slope of the linear regression with time (numerical recipes in c)
x_43m_b_5* offset of the linear regression with time
x_43m_siga_5* standard error of a
x_43m_sigb_5* standard error of b
x_43m_chi2_5* chi squared of the residuals
x_43m_rsqr_5* squared correlation coeffcient
theta_5 vertical rotation angle in degrees
phi_5
 azimuth (mathematical coordinate system)
u_123_5
 wind speed
u_12_5
 horizontal wind speed
windir
 horisontal wind direction (meteorological coordinate system)
* outputs are only given in processiong level 5
  1. cov gives the results of the covariance maximisation

Covariances in the files *.co?
Variable
 Explanation
Lag_CovMax_CO2_ppm The time lag in samples at maximum covariance for the first of the sensors, for which covariance maximisation is requestet
CovMax_CO2_ppm the maximum covariance
LagFlag_CO2_ppm a flag 1 is no maximum was found within the window, otherwise 0
Lag_CovMax_H2O_pm The time lag in samples at maximum covariance for the second sensors, for which covariance maximisation is requestet
CovMax_H2O_pm the maximum covariance
LagFlag_H2O_pm a flag 1 is no maximum was found within the window, otherwise 0
L50_CO2_ppm the first lagged covariance sensor one
...
L150_CO2_ppm the last lagged covariance sensor one
L50_H2O_pm the first lagged covariance sensor two
...
L200_H2O_pm the last lagged covariance sensor two
...

  1. tst contains the results of the stationarity test (Foken and Wichura, 1996), calculated after roration but before linear detrending. For each sensor the following list is output.
stationarty test in files *.tst
Variable
 Explanation
QF_m_x_43m
 quality flag (1-9) according to Foken et al. (2003) for Sens 1
m0_x_43m mean of Sens1 the 1st 5 minute data
m1_x_43m mean of Sens1 of the 2nd 5 minute data
m2_x_43m mean of Sens1 of the 3rd 5 minute data
m3_x_43m mean of Sens1 of the 4th 5 minute data
m4_x_43m mean of Sens1 of the 5th 5 minute data
m5_x_43m mean of Sens1 of the 6th 5 minute data
QF_c_x_43m
 quality flag (1-9) according to Foken et al. (2003)
c0_x_43m_z_43m covariance of Sens1 with the vertical wind velocity of the 1st 5 minute data
c1_x_43m_z_43m covariance of Sens1 with the vertical wind velocity of the 2nd 5 minute data
c2_x_43m_z_43m covariance of Sens1 with the vertical wind velocity of the 3rd 5 minute data
c3_x_43m_z_43m covariance of Sens1 with the vertical wind velocity of the 4th 5 minute data
c4_x_43m_z_43m covariance of Sens1 with the vertical wind velocity of the 5th 5 minute data
c5_x_43m_z_43m covariance of Sens1 with the vertical wind velocity of the 6th 5 minute data
...
 quality flag (1-9) according to Foken et al. (2003) for Sens 2
The test is only made with categorie 5 data (see below) and can be used in comparison with the respective output of co5 and st5.
Foken, T., M. Göckede, M. Mauder, L. Mahrt, B.D. Amiro and J.W. Munger 2003. Post-field data quality control. In Handbook of Micrometeorology: A Guide for Surface Flux Measurements Ed. X. Lee. Kluver, Dordrecht, p. 19.

  1. kpf contains header informations (not useful)

Spectral output

Spectral information is written into ASCII files for each raw data file separately. Spectral output files have the same name as the raw data file but the extension .fat.
RCPM uses the finally conditioned raw data, currently after step 5., for spectral analysis. The kernel function for spectral analysis of discretely sampled data  is realft of  numerical recipes in c.

A typical appearance of the output is:

#20050802_0300_LlValbyRimi.dat, u, 2.769727, z-d, 2.000000,
#f, Blockanfang,  f_x_x, ... , f_x_z, ... , x_x, ..., x_z,
#, , 0.287324, ...,  -0.055693, ...
0.000610, 1,  0.0363602, ... , -0.0820392,..., 0.00346263, ... , -0.0820392, ...
...
followed by the spectral data fro every ferquency bin.

1. Header

The first  header starts with the name and extension of the raw data file, the horizontal wind speed, u, in m s-1 and the measurement height in m as expressed as the difference between measurement height and displacement height (see).

2. Header

names of the variables:
Variable
 Explanation
f
 Frequency, i.e. the average of the natural frequency for each of the currently 50 frequency bins in Hz
Blockanfang
 Number of frequencies, where a new bin starts.
f_x_x
 Power spectrum of the sensor x multiplied with the frequency. I.e. the bin average of the product of spectral density and the number of the frequency ranging  from 1 to the number of observations / 2.
...
f_x_z
 Cross-spectral density of the sensor x with z multiplied with the frequency. I.e. the bin average of the product of cross-spectral density and the number of the frequency ranging  from 1 to the number of observations / 2.
...
x_x
 Power spectral density  of the sensor x
...
x_z
 Cross-spectral density  of the sensor x

3. Data section

The respective data for each bin that contains at least one frequency.




Back to the Main Page

  © Andreas Ibrom
Author Andreas Ibrom Date: 11.06.2008

Risø National Laboratory
BIO-ECO Byg. 309
Frederiksborgvej 399
DK-4000 Roskilde