Is it possible to use the tracer-to-tracer method for satellite data?
Obvious point sources of CO and CH4 and discernable seasonal variation show the potential of TROPOMI
Meaning and Purpose
For environmental policy-making, there is a need to quantify the megacity CH4 emission in this “point source”, but the bottom-up emission inventory has large uncertainties, which can be >30%. In contrast, the CO bottom-up evaluation is more precise. Therefore, the tracer-to-tracer was proposed to restrain the ambiguous emission in a certain area and has been widely adopted by previous studies based on ground observation data.
However, for those cities without sophisticated ground observation networks, it is still unrealistic to assess CH4 emission from them. Here we initially test the tracer-to-tracer method on CO and CH4 data in Los Angeles, which is retrieved by scientific algorithm WFMD from TROPOMI-S5P satellite observation. In comparison with previous studies, we hope to test the validity of this method for satellite data, whereby precise CH4 emission in those cities without ground observation will be available based on satellite data.
The general idea of the tracer-to-tracer method is to calculate one ambiguous emission of a tracer gas by multiplying another tracer gas's emission, which is more precise, by their ratio. This ratio is fitted from the observed enhancement value due to this "point source", the Los Angeles megacity here.
We designated the city area (33.75~34.4°N, 117.7~118.6°W) and background area (35~36°N, 117~118°W), and regard the difference between city and background as enhancement value (excess). We took TCCON sites in Caltech (34.14°N, 118.13°W) and in Edwards (34.96°N, 117.88°W) as standards for city and background data respectively.
Corrections for data
To ensure the accuracy of the TROPOMI data, we corrected the impacts of apparent albedo, elevation, and average kernel on the fitting results. We fitted quadratic functions to describe and eliminate the CH4 and CO's dependence on apparent albedo. A model was built to make up the difference between the background vertical profile and that of the city due to different elevations. The mean value of all the average kernels within the boundary layer (>910hPa) divides gas concentrations because the enhancement is concentrated within the boundary layer.
Test results: negative
The final fitted correlation of CO and CH4 is
CH4 = 2.50 × CO + 17.27
The negative constant is due to excessive CH4 in the background led by the nearby oil industry. The larger coverage of TROPOMI data than the TCCON site enhances this impact. The higher slope of TROPOMI is likely attributed to the biased XCOex value, since if 2.50 times the biased ratio of CO(TROPOMI)/CO(TCCON) of 0.5~0.6, it can be quite close to TCCON fitted slope of 1.44. And the discrepancy between CO(TROPOMI) and CO(TCCON) is due to the lack of calibration, like the shallow learning used for CH4.