It is a recognized advantage of satellite-derived measurements that they can be regarded as a reliable benchmarks for model simulation evaluation. However, it seems we should remove an initial guess (or a priori) considered in the procedure of trace gases retrieval used in the satellites. I found two ways to wipe out this influence when using OMI tropospheric NO2:
1) From Duncan et al., 2014 (supplementary): The end-user must sum over all model layers the product of the scattering weight and model partial column (molecules/cm2) in each model layer. This sum divided by VCDM is called the air mass factor (AMF) of the model (AMFM). Second, the end-user must divide the product of VCDD and AMFD from the data file by AMFM to obtain a modified form of VCDD (VCD'D): VCD'D = (VCDD*AMFD)/AMFM.
VCDD is vertical column density of satellite and VCDM is for model.
This procedure is looking for modification of OMI observations.
2) From Kumar et al., 2012: The procedure for transforming the WRF-Chem simulated tropospheric column NO2 abundances for comparison to OMI and GOME-2 retrievals is different from that used for TES and MOPITT. This procedure requires the user to calculate the tropospheric averaging kernels (Atrop) through the following equation:
Atrop=A*(AMF/AMFtrop)
where A is the total column averaging kernel and AMF and AMFtrop are the air mass factors for the total columns and tropospheric columns, respectively. The tropospheric averaging kernels are then applied to the tropospheric vertical profiles of NO2 simulated by WRF-Chem.
This procedure is looking for modification of WRF-CHEM outputs.
Now my questions:
1) Are above ways exactly the same? If not, which one is more accurate?
2) Is AMF (total air mass) equivalent to sum of AMFtrop and AMFstr; because I could not find an AMF variable in HDF of OMI.
3) It seems only daily granule OMI observations do have needed variables to perform the comparison: http://aurapar2u.ecs.nasa.gov/opendap/Aura_OMI_Level2/OMNO2.003//2013/
The main problem of daily OMI observations is that they have very poor spatial resolution in pixels far from nadir (panoramic effect). See the attachment. I guess, there is an approach used in global product (instead of daily granules) that overlays two adjacent granule and makes these marginal observations less coarse; Am I right?
Is there any suggestion for doing so?
Thank you in advance,
To 1) The two methods you are describing are for the (1) NASA OMI NO2 standard product and (2) Dutch OMI NO2 product. Since you are planning to use the NASA product, I suggest that you are using the method described by Duncan et al. (2014).
To 2) No, you cannot just sum them up because the summed scattering weights are divided by the partial column.
To 3) Yes, only the Level 2 data provide AMFs because you cannot apply AMF corrections to the gridded (mappped) data. To create the maps, you can use a gridding algorithm we recently developed (Kuhlmann et al., 2014, see attachment).
The source code is available on my GitHub page: https://github.com/gkuhl/omi
Article A novel gridding algorithm to create regional trace gas maps...
Article A novel gridding algorithm to create regional trace gas maps...
To 1) The two methods you are describing are for the (1) NASA OMI NO2 standard product and (2) Dutch OMI NO2 product. Since you are planning to use the NASA product, I suggest that you are using the method described by Duncan et al. (2014).
To 2) No, you cannot just sum them up because the summed scattering weights are divided by the partial column.
To 3) Yes, only the Level 2 data provide AMFs because you cannot apply AMF corrections to the gridded (mappped) data. To create the maps, you can use a gridding algorithm we recently developed (Kuhlmann et al., 2014, see attachment).
The source code is available on my GitHub page: https://github.com/gkuhl/omi
Article A novel gridding algorithm to create regional trace gas maps...
Article A novel gridding algorithm to create regional trace gas maps...
Dear Dr. Kuhlmann
I should express my sincere appreciation for your prompt and comprehensive reply. Your modification in the paper is indeed remarkable compared to previous product; however some of my ambiguities remain:
1) How can I retrieve total air mass for using Duncan method? I can only see AMFtrop and stro. Should I use the ratio of slant and vertical columns?
2) It sounds I cannot directly use your code, since I should involve WRF-CHEM output (AMFM) in the calculation (with different spatial subset and resolution) (I enclosed my MATLAB code). What is your suggestion?
BTW, do you think neglecting this removal (i.e. initial guess) could result in large discrepancies between observed and modeled NO2 values?
To 1) You can correct the stratospheric and tropospheric AMF separately and then sum up the resultung VCDs. The OMI Level 2 file contains the height of the troposphere to do this. The SCDs can be calculated from the provided VCDs and AMFs. If WRF-Chem does not provide suitable NO2 profiles in the stratosphere, I wouldn't update the stratospheric VCD but only work on the tropospheric part.
To 2) My code is only for gridding the Level 2 data. You can write a programme for updating the VCDs and afterwards use my programme with your improved data.
To BTW) If you are interested in regional NO2, the discrepancies can be quite large (see Russell et al 2011, doi:10.5194/acp-11-8543-2011).
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing