Title:	Inverse modeling of greenhouse gases
DNr:	NAISS 2025/5-111
Project Type:	NAISS Medium Compute
Principal Investigator:	Hans Chen <hans.chen@chalmers.se>
Affiliation:	Chalmers tekniska högskola
Duration:	2025-02-25 – 2025-09-01
Classification:	10508 10501
Keywords:

Abstract

This project advances the state-of-the-art in regional greenhouse gas (GHG) flux estimation through innovative inverse modeling techniques. By leveraging high-performance computing, we will develop and implement sophisticated data assimilation methods that integrate prior GHG flux estimates, atmospheric transport modeling, in situ measurements, and cutting-edge satellite observations. At the core of this research is the TRACE Regional Atmosphere–Carbon Ensemble (TRACE) system, a data assimilation framework developed by the PI (Hans Chen). TRACE specifically addresses the challenge of constraining regional carbon dioxide fluxes at high spatial and temporal resolutions. Our primary innovation lies in developing novel methodologies to assimilate high-resolution satellite measurements. This represents a significant advancement over conventional atmospheric inversion techniques, which struggle with satellite data due to fundamental assumptions in their formulations. TRACE consists of two major components: atmospheric transport modeling using the Weather Research and Forecasting Model (WRF), a full-physics mesoscale atmospheric model, and an advanced data assimilation system based on the PSU WRF EnKF framework. The data assimilation system employs ensemble-based algorithms to propagate probability density functions of the system’s state through time. The current WRF setup operates at a 27 km horizontal resolution, which we plan to increase to 9 km to better leverage the detail in high-resolution satellite imagery. To make the ensemble generation more efficient, we have modified WRF to simulate multiple tracers within a single run. Additionally, TRACE supports fully independent WRF simulations with varying atmospheric transport fields. This enables comprehensive characterization of transport uncertainties - a novel capability that will be further developed and explored in this project. Initially, we will focus on shorter (~month-long) simulations to optimize computational and storage resources while producing scientifically valuable outputs. After developing and validating these methods, we plan to extend their application to longer timescales to detect changes and trends in GHG emissions. The project’s outcomes will significantly advance our understanding of regional carbon dynamics and support climate policy decisions.