Mapping methane emissions from a marine geological seep source using imaging spectrometry

Methane (CH4) is an extremely important greenhouse gas that has increased significantly in pre and post-industrial times. Due to CH4ʹs strong absorptions in the shortwave infrared (SWIR), the potential exists to use imaging spectrometers, such as the Airborne Visible Infrared Imaging Spectrometer (AVIRIS), to map CH4 emissions. Here, we present research evaluating the ability of AVIRIS to map CH4 emitted by one of the largest marine geologic CH4 sources in the world, the Coal Oil Point seep field in the Santa Barbara Channel, California. To develop algorithms for detecting CH4 and to establish detection limits, initial analysis focused on simulated radiance spectra, calculated using Modtran 5.2 radiative transfer code that was parameterized to match scene conditions for a 6 August 2007 AVIRIS flight over the area. Model simulations included a range of surface albedos, variable column water vapor, and CH4 concentrations ranging from 1.7 ppm (background) up to the equivalent of 2500 ppm in the lower 20 m of the boundary layer. A multistep CH4 detection algorithm was developed using Modtran simulations. First, surface albedo was estimated from the radiance at 2139 nm. Next, albedo-specific radiance for background CH4 was used to calculate spectral residuals for CH4 above background. A CH4 index, C, calculated as the average residual between 2248 and 2298 nm, showed high sensitivity to CH4, with minimal sensitivity to water vapor or surface albedo. Detection limits in simulations depended on CH4 concentrations and albedo, ranging from 990 ppm for a 0.5% albedo surface, to as low as 18 ppm for albedos higher than 22%. Application of this approach to the AVIRIS data demonstrated considerable potential for mapping CH4. Due to specular reflectance off of the ocean surface, albedo in the scene varied significantly, from less than 0.5% to over 30%. Strong CH4 anomalies were observed in the data acquired over the seep field, which produced large C values with spectral residuals consistent with CH4 and estimated radiance spectra that matched measurements. All strong anomalies were located in close vicinity to and downwind from known CH4 sources. However, contrary to simulated data, C was overly sensitive to albedo, restricting high confidence anomalies only to the brightest surfaces, and showing high frequency spatial variation throughout the AVIRIS image. CH4 concentration was overestimated by C, potentially due to a spectral trend in sea surface reflectance and/or the impact of diffuse light on dark surfaces (< 1%) leading to the over-expression of CH4 absorptions.