Generalized multi-band methane enhancement retrievals using PACE-OCI

Abstract

This thesis investigates the capabilities of the Ocean Color Instrument (OCI) aboard the PACE satellite for methane enhancement retrievals, and explores the potential of a new methane reference band centered at 2.13 μm. We assess the performance of this band relative to the commonly used 1.61 μm band, and study how the existing multi-band–multi-pass (MBMP) method can be generalized to use more than two spectral bands.We apply existing matched filter (MF) methods to multispectral OCI data, and extend them by including the usage of a methane-free reference day. In addition, we develop a new retrieval method based on generalized least squares (GLS) which does not require linearization and which is the generalization of the MBMP method to ≥2 spectral bands. Retrieval performance is evaluated using a set of 17 real-world case studies and a roundtrip simulation framework in which synthetic methane plume absorptions are injected into observed backgrounds, allowing for controlled comparisons of retrieval accuracy and precision.Using the new 2.13 μm band instead of the 1.61 μm band with the MBMP method resulted in a factor 4 reduction in background variability averaged over all tested scenes. The GLS method consistently outperformed the standard MBMP approach in terms of both accuracy and precision. When using the H and I bands for both methods, GLS showed a background variability which was 7.2% lower compared to MBMP. When using all three methane-sensitive bands, the relative reduction was 8.8%.The new GLS method can be readily applied to existing multispectral instruments, and is likely to reduce background variability and increase retrieval precision. The method is of special interest for (future) missions which have multiple methane-sensitive bands, such as WorldView-3 and Sentinel-2 Next Generation.