Surface reflectance biases in XCH4 retrievals from the 2.3 µm band are enhanced in the presence of aerosols

Abstract

In this work, we present the results of an observing system simulation experiment (OSSE) in which we investigate the emergence of a surface-reflectance-dependent bias in retrieved column-averaged dry-air mole fractions of methane (XCH4). Our focus is on single-band retrievals in the shortwave infrared (SWIR) at 2.3 µm. This particular bias manifests as artificial gradients in XCH4 fields that relate to surface features on the ground and can, for example, cause erroneous estimates of methane source emission rates. We find that even for near-ideal conditions (that being a perfectly calibrated instrument, perfect knowledge of meteorology and trace gas vertical distributions, and an absence of clouds and aerosols) a surface-reflectance-related bias appears in the retrieved XCH4. While the magnitude of the bias is much lower than is observed in, for example, real data from the TROPOspheric Monitoring Instrument (TROPOMI), the overall qualitative shape is strikingly similar. When we study a more realistic scenario by considering synthetic measurements that are affected by aerosols, the surface bias increases in magnitude roughly by a factor of 10. We hold all other properties of the synthetic measurements fixed and thus can make the following statements about these surface biases from the 2.3 µm absorption band. First, the bias already appears in the near-perfect scenario, meaning that its origin is likely fundamental to XCH4 retrievals from this particular absorption band and using an optimal-estimation-type retrieval approach. Second, the magnitude of the bias increases significantly when aerosols are encountered. As aerosols give rise to a magnification of the bias, we have implemented a retrieval configuration in which the retrieval algorithm knows the true aerosol abundance profiles along with their optical properties. With this configuration, the surface bias returns mostly to the level first seen when synthetic measurements were not affected by aerosols. The results we present in this work should be considered for new missions where XCH4 is a target quantity and the design relies on the 2.3 µm absorption band. Since the surface bias will likely emerge, it is crucial that a validation approach is planned which sufficiently samples the needed range of surface reflectance in areas of near-uniform methane concentrations in order to capture the bias and thus correct for it.