On the potential of a neural-network-based approach for estimating XCO<sub>2</sub> from OCO-2 measurements
<p>In David et al. (2021), we introduced a neural network (NN) approach for estimating the column-averaged dry-air mole fraction of CO<span class="inline-formula"><sub>2</sub></span> (<span class="inline-formula">XCO<sub>2</sub><...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2022-09-01
|
| Series: | Atmospheric Measurement Techniques |
| Online Access: | https://amt.copernicus.org/articles/15/5219/2022/amt-15-5219-2022.pdf |
| Summary: | <p>In David et al. (2021), we introduced a neural network (NN) approach for estimating the column-averaged dry-air mole fraction of CO<span class="inline-formula"><sub>2</sub></span> (<span class="inline-formula">XCO<sub>2</sub></span>) and the surface pressure from the reflected solar spectra
acquired by the OCO-2 instrument. The results indicated great potential for
the technique as the comparison against both model estimates and independent TCCON measurements showed an accuracy and precision similar to or better than that of the operational ACOS (NASA's Atmospheric CO<span class="inline-formula"><sub>2</sub></span> Observations from Space retrievals – ACOS) algorithm. Yet, subsequent analysis showed that the neural network estimate often mimics the training dataset and is unable to retrieve small-scale features such as CO<span class="inline-formula"><sub>2</sub></span> plumes from industrial sites. Importantly, we found that, with the same inputs as those used to estimate <span class="inline-formula">XCO<sub>2</sub></span> and surface pressure, the NN technique is able to estimate latitude and date with unexpected skill, i.e., with an error whose standard deviation is only 7<span class="inline-formula"><sup>∘</sup></span> and 61 d, respectively. The information about the date mainly comes from the weak CO<span class="inline-formula"><sub>2</sub></span> band, which is influenced
by the well-mixed and increasing concentrations of CO<span class="inline-formula"><sub>2</sub></span> in the
stratosphere. The availability of such information in the measured spectrum
may therefore allow the NN to exploit it rather than the direct CO<span class="inline-formula"><sub>2</sub></span>
imprint in the spectrum to estimate <span class="inline-formula">XCO<sub>2</sub></span>. Thus, our first version of the NN
performed well mostly because the <span class="inline-formula">XCO<sub>2</sub></span> fields used for the training were
remarkably accurate, but it did not bring any added value.</p>
<p>Further to this analysis, we designed a second version of the NN, excluding
the weak CO<span class="inline-formula"><sub>2</sub></span> band from the input. This new version has a different
behavior as it does retrieve <span class="inline-formula">XCO<sub>2</sub></span> enhancements downwind of emission hotspots, i.e., a feature that is not in the training dataset. The comparison against the reference Total Carbon Column Observing Network (TCCON) and the surface-air-sample-driven inversion of the Copernicus Atmosphere Monitoring Service (CAMS) remains very good, as in the first version of the NN. In addition, the difference with the CAMS model (also called <i>innovation</i> in a data assimilation context) for NASA Atmospheric CO<span class="inline-formula"><sub>2</sub></span> Observations from Space
(ACOS) and the NN estimates is correlated.</p>
<p>These results confirm the potential of the NN approach for an operational
processing of satellite observations aiming at the monitoring of CO<span class="inline-formula"><sub>2</sub></span>
concentrations and fluxes. The true information content of the neural
network product remains to be properly evaluated, in particular regarding
the respective input of the measured spectrum and the training dataset.</p> |
|---|---|
| ISSN: | 1867-1381 1867-8548 |