CDS-ESPRI site AERIS Site
ESPRI Data Centre (former Ether)

ESPRI is a data centre of the French Atmosphere Infrastructure AERIS

......................................
Access to data
.............................................
Contact & publications

......................................

Carbon monoxide (CO)


How to read the data

Submit your email to receive updates :

Download this Readme LATMOS-ULB FORLI-CO data retrieved from IASI here.

How to read the data since December 2, 2010 :

Each file contains data for one day of observation. In each file, each line corresponds to one observation.
File names include date of observation. Their structure is :
iasi_CO_LATMOS_ULB_YYYYMMDD_vXXXXXXXX.txt
where : YYYY = year, MM = month, DD = day, XXXXXXXX = version number of the retrieval code.
The data are organized in 60 columns :
columns 1 and 2 :latitude and longitude
columns 3 and 4 : date (yyyymmdd) and time (hhmmss)
columns 5 : solar zenith angle
columns 6 : IASI field of view (0, 1, 2 or 3)
columns 7 : information flag about Eumetsat IASI L2 temperature profiles retrieval method (see above)
columns 8 to 15 : quality flags (see hereafter)
columns 16 : “super” quality flag (see hereafter)
columns 17 : Eumetsat cloud coverage in the pixel (in %, between 0 and 25 %)
columns 18 : Degrees Of Freedom of the Signal (DOFS, ~number of independant information pieces concerning CO vertical distribution in the signal)
columns 19 : fit residual root mean square (in W/m2/cm-1)
columns 20 : fit residual bias (in W/m2/cm-1)
columns 21 : CO total column (in molec.cm-2)
columns 22 : CO total column relative error (observation_error/observation ratio)
columns 23 to 41 : Xa, a priori CO profile in partial columns (in molec.cm-2) and on the same vertical layers as the A matrix, with -999 for missing levels
columns 42 to 60 : A, averaging kernel vector for the total column in 19 vertical layers (in partial column), with -999 for missing levels (see hereafter for the description of the layers)


How to read the data before December 2, 2010 :

Each file contains data for one day of observation. In each file, each line corresponds to one observation.
File names include date of observation. Their structure is :
iasi_CO_LATMOS_ULB_YYYYMMDD_vXXXXXXXX.txt
where : YYYY = year, MM = month, DD = day, XXXXXXXX = version number of the retrieval code.
The data are organized in 59 columns :
columns 1 and 2 :latitude and longitude
columns 3 and 4 : date (yyyymmdd) and time (hhmmss)
columns 5 : solar zenith angle
columns 6 : IASI field of view (0, 1, 2 or 3)
columns 7 to 14 : quality flags (see hereafter)
columns 15 : « super » quality flag (see hereafter)
columns 16 : Eumetsat cloud coverage in the pixel (in %, between 0 and 25 %)
columns 17 : Degrees Of Freedom of the Signal (DOFS, ~number of independant information pieces concerning CO vertical distribution in the signal)
columns 18 : fit residual root mean square (in W/m2/cm-1)
columns 19 : fit residual bias (in W/m2/cm-1)
columns 20 : CO total column (in molec.cm-2)
columns 21 : CO total column relative error (observation_error/observation ratio)
columns 22 to 40 : Xa, a priori CO profile in partial columns (in molec.cm-2) and on the same vertical layers as the A matrix, with -999 for missing levels
columns 41 to 59 : A, averaging kernel vector for the total column in 19 vertical layers (in partial column), with -999 for missing levels (see hereafter for the description of the layers)


Significations of quality flags :

1. descriptive flags :
flag 1 : is 1 if surface altitude in our model is negative
flag 2 : is 1 if Tskin is missing in IASI level 2 data and is replaced by the brightness temperature at 2143.25 cm-1 in our model
flag 3 : is 1 if the difference in absolute value between Tskin of IASI level 2 data and the brightness temperature at 2143.25 cm-1 is higher than 5K
flag 4 : is 1 if a desert is detected at the Earth's surface
2. fit flags :
flag 5 : is 1 if the maximum number of iterations is reached without convergence
flag 6: is 1 if the fit residual bias is sloped
flag 7 : is 1 if the contrast of CO lines is weak
flag 8: is 1 if the averaging kernel vector includes strange values (generally too high)

3. “Super” quality flag: :
The “super” quality flag is a summary of the detailed quality flags. To describe it, we use two additional fit flags: one to indicate that the fit residual root mean square is too high and one to indicate that the fit residual bias is too high (see above).
The super quality flag can have 3 values :
0 when
(i) all the previous flags are null or
(ii) the cloud cover is less than 12 % (but higher than 0) and only flag 2 is equal to 1 ;
1 when
(i) all the fit flags are null with at least one descriptive flag equal to 1 or
(ii) all the fit flags are null with only flag 2 equal to 1 and the cloud cover equal to 0 or
(iii) all the fit flags are null with only flag 2 equal to 1 and the cloud cover higher or equal to 12 % ;
2 when at least one fit flag is equal to 1.

For data validation purpose, we recommend to use the data for which the “super” quality flag is equal to 0.


The averaging kernel matrix is dimensionless but calculated from profiles in partial columns units, on 19 layers (level 1 corresponds to the first layer of the atmosphere [0-1km], level 2 [1-2km],.. level 19 [18-60km, top of the atmosphere]).
Because only CO total columns are provided here, we provide the corresponding averaging kernel vector.
When the first levels are not available (because of the orography), data=-999.
To smooth a carbon monoxide profile, one should use:
X_smoothed= A*Xr + (I-A)*Xa with
Xr the profile (in partial columns) to be smoothed
A the IASI averaging kernel vector
Xa the IASI a priori profile (in partial columns).
(Cf Rodgers, 2000)

To properly use the averaging kernel vectors two conditions must be fulfilled :
- Profiles must be in partial columns;
- The profile to be smoothed (Xr) must go at least as low as the IASI profile in altitude (it can go lower than IASI).
If these conditions are not met, smoothing with the vector does not give the same result as smoothing with the complete averaging kernel matrix.

FORLI a priori CO profile in VMR between 0 and 60 km is the following :

01.0161773E-07164.1054126E-08314.6196803E-08461.8742890E-07
19.4855840E-08173.6674835E-08324.9959250E-08472.1709527E-07
29.1867937E-08183.2485751E-08335.4181389E-08482.4970039E-07
38.9939589E-08192.9398235E-08345.8425609E-08492.8425973E-07
48.8739003E-08202.6610051E-08356.3029508E-08503.2205485E-07
58.8418514E-08212.4046862E-08366.7743093E-08513.6187539E-07
68.7545818E-08222.3019179E-08377.2737994E-08524.0372110E-07
78.6420669E-08232.2254821E-08387.7949241E-08534.4595348E-07
88.3986495E-08242.1616224E-08398.3286388E-08544.9068308E-07
98.0001663E-08252.1599277E-08408.9656617E-08555.3657991E-07
107.5277656E-08262.5668884E-08419.7574193E-08565.8427115E-07
116.9161481E-08273.0270625E-08421.0816715E-07576.3513023E-07
126.3326457E-08283.4893978E-08431.2211041E-07586.9045460E-07
135.7343264E-08293.9139751E-08441.4033574E-07597.4988295E-07
145.1686031E-08304.2789457E-08451.6217000E-07608.1519043E-07
154.6027417E-08
Last update : 2017/12/20
            Project call    Legals    Contacts