📌 TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

Inverse modeling of CO₂ emissions

using satellite observations

from OCO-2 and OCO-3

Sojung Sim, Sujong Jeong

(simsj0304@snu.ac.kr)

28 May 2024 @ NSF NCAR Mesa Lab in Boulder, Colorado

01 Introduction

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

• Anthropogenic emissions of carbon dioxide (CO₂), have dominated since the Industrial Era, resulting in rising atmospheric CO₂ levels and driving climate change.

​

• The IPCC 1.5℃ Special Report declared that achieving global net-zero CO₂ emissions by 2050 is imperative to limit the increase in global temperature to 1.5 ℃ above pre-industrial levels.

1. Net-zero CO₂ emissions

01 Introduction

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

2. Cities

• Cities bear responsibility for emission reduction and are expected to play a major role in meeting net-zero goals.

​

• CO₂ Emission estimates at city scale are needed to provide detailed guidance to establish a baseline for prioritizing climate action and assessing policy progress over time.

• By 2050, it is projected that 68 percent of the world's population will reside in urban areas.

​

• At least 70% of global anthropogenic CO₂ emissions originate from cities.

01 Introduction

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

3. Top-down approach

• A complementary and independent approach to verify CO₂ emission estimates is deemed necessary.

​

• The accuracy of bottom-up CO₂ emission estimates can be improved via real-time CO₂ measurements and atmospheric transport models (Top-down approach).

01 Introduction

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

4. Seoul

• Given its dense population, concentrated emissions, as well as extensive measurement networks, Seoul can be an optimal testbed city for studies aimed at verifying CO₂ emission estimates and assessing the effectiveness of mitigation policies.

01 Introduction

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

5. Objective

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

1. Bayesian Inverse Modeling Framework

• The Bayesian inverse modeling framework was developed to derive optimized posterior CO₂ emission estimates over Seoul.

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

2. Anthropogenic CO₂ Emissions

• We used anthropogenic CO₂ emissions data from ODIAC (Open-source Data Inventory for Anthropogenic CO₂) as the state vector of prior CO₂ emissions (s_p).

​

• The ODIAC is based on the downscaling of bottom-up CO₂ emission estimates using spatial proxies.

​

• The ODIAC produces global fossil fuel CO₂ emission estimates at a 1 × 1 km² resolution using power plant profiles and space-based nighttime light data.

< Spatial distribution of CO₂ emissions >

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

3. Atmospheric CO₂ measurement

• For the ground measurements, we used observed CO₂ concentrations from five sites in Seoul.

​

• We used only daytime (10-16 KST) for the inverse modeling in order to minimize the influence of model biases in PBL height.

​

• The OCO-2 and OCO-3 satellite passed over Seoul on Dec 4 and 5, 2021, yielding 60 and 167 soundings, respectively.

< OCO-2 overpass over Seoul >

< OCO-3 overpass over Seoul >

< Ground measurements >

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

4. Biogenic CO₂ Fluxes

• Given that Seoul has forests covering 25.3% of its total area, the impact of biogenic CO₂ activities cannot be ignored.

​

• To account for the influence of biogenic CO₂ on the observed enhancements, we incorporated biogenic CO₂ fluxes estimated using a data-based model known as CASS (Carbon Simulator from Space).

​

• We utilized hourly net ecosystem exchange (NEE) data, which were resampled from a 250-meter resolution to 0.01°, as the biogenic CO₂ flux data within Seoul.

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

5. Lagrangian Transport Model

• For WRF-(X)STILT, one thousand air particles were released from each observation site (each column level) and tracked backward in time for 24 h.

• In Bayesian inverse modeling, we require the footprint, which functions as an operator (H) connecting CO₂ measurements and emissions, and WRF-(X)STILT was used to derive the footprints.

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

6. Prior Error Covariance

• The prior error covariance matrix (Q) can be derived from both the variance in prior emissions uncertainty (σ) and the temporal and spatial covariances (D and E).

02 Method & Data

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

7. Observational Error Covariance

• The residual error method was used to estimate observation error covariance (R).

03 Results

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

1. Comparison between prior and posterior emissions

• The average of prior and posterior emissions were 35.26 and 32.6 μmol/(m² s), respectively.

​

• While the spatial patterns of the posterior still resemble those of the prior emissions, CO₂ emissions decreased after the inversion run.

03 Results

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

1. Comparison between prior and posterior emissions

• The averaged emission correction between prior and posterior emissions was about -7.53%. �🡪 Prior emissions were overestimated.

​

• Most parts of the city have been adjusted through Bayesian inverse modeling.

​

• Posterior corrections were mostly negative in center area, except in the far south and north of Seoul.

03 Results

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

2. Uncertainty reduction

• The most significant reductions in uncertainty were observed in areas where observation sites are concentrated 🡪 We need to establish observation sites in the northern part of Seoul.

​

• Uncertainty reduction also extended within the footprint influence range �🡪 Notable decrease in uncertainty was observed in the area northwest of the sites.

03 Results

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

3. Sensitivity test

< Emission corrections >

• Five ground observations significantly contribute to the corrective magnitude of the inversion.

​

• Although each satellite observation has passed once a month, they had the effect of spatially correcting many areas of Seoul.

​

• In particular, the OCO-3 satellite covered the northeastern area of Seoul, which was not covered by ground observations.

03 Results

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

3. Sensitivity test

< Uncertainty reduction >

​

• Less uncertainty reduction for satellite data was achieved due to its limited coverage for only one hour during the month.

​

• To maximize the performance of the inversion, combining observational information from both ground- and space-based measurements is essential.

* Different range!

All (11.43%)

Only ground (10.03%)

Only OCO-2 (0.36%)

Only OCO-3 (1.83%)

04 Conclusion

TransCom-2024: Quantifying errors in inversions of satellite trace gas retrievals

• We will expand the target area from Seoul to major cities and Korea, using ground and satellite observations from OCO-2 and OCO-3.

​

• By quantifying current emissions and uncertainties, we will ultimately contribute scientifically to mitigating climate change by reducing Korea's emissions.

E-mail

simsj0304@snu.ac.kr

Thank you for your attention