The OCO‐2 oxygen A‐band response to liquid marine cloud properties from CALIPSO and MODIS

Abstract

Spectra of reflected sunlight in the oxygen A‐band contain information about cloud properties such as cloud top pressure, optical depth, and pressure thickness. Here we show, for the first time, that high‐spectral‐resolution A‐band Orbiting Carbon Observatory‐2 (OCO‐2) spectra respond largely as simulated to the optical properties of water clouds over ocean during November 2015 (N = 184,318) using input cloud properties from the Moderate Resolution Imaging Spectroradiometer (MODIS) on Aqua and the Cloud‐Aerosol Lidar with Orthogonal Polarization on Cloud‐Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO). In A‐band continuum channels the standard deviation of simulated minus observed radiance is ±37%. Selecting horizontally homogeneous clouds to mitigate three‐dimensional cloud effects and collocation error with the other satellites, the standard deviation of the residuals is reduced to ±18%. Using a look‐up table developed from simulations, OCO‐2's estimated cloud top pressure for low clouds (Ptop > 680 hPa) has a standard deviation of ±61 hPa relative to CALIPSO retrievals, and bias is dependent on assumed cloud pressure thickness, with our smallest value being −5 hPa. Versus MODIS optical depth, the standard deviation is ±9.0 and the bias is −2.0, although these shrink for clouds of τ < 30. These values include collocation error between the different satellites, meaning that they place an upper bound on the OCO‐2 retrieval uncertainty. The theoretical precision limit from OCO‐2's instrumental uncertainty is shown to be ±2.4 hPa in above‐cloud path and ±0.2% in optical depth for a two‐channel retrieval. Options for retrieving cloud optical depth, cloud top pressure, and pressure thickness are discussed in the context of a formal OCO‐2 cloud property retrieval.

Information content of OCO-2 oxygen A-band channels for retrieving marine liquid cloud properties

“…Prior covariance is assumed to be diagonal, equivalent to error of ±1.5, Ptop of ±60 hPa and Δ of ±7.5 hPa. Our prior error comes from applying the ±15 % error in simulated radiance for homogeneous clouds when provided with MODIS optical depth found in (Richardson et al, 2017). Our Ptop uncertainty is approximately equal to that determined in the same study when comparing prior OCO-2 Ptop with CALIPSO.…”

“…It is a flight spare from the original OCO mission and a number of pixels have failed. 853 of the 1,016 channels are available across all soundings and over 94 % of the damaged pixels occur in the A-band continuum where there is redundancy, meaning little loss of information (Richardson et al, 2017).…”

“…Although the L2RTM was designed for clear-sky XCO2 retrievals, it has been validated in cloudy atmospheres by comparing OCO-2 observations with L2RTM output assuming collocated MODIS and CALIPSO cloud properties (Richardson et al, 2017). For homogeneous single-layer liquid clouds over ocean, the root mean square error (RMSE) in continuum channels was ±18 %, an overestimate of the model-only error as this includes 3d cloud effects, collocation error, parallax effects and 30 uncertainty in the MODIS and CALIPSO retrievals.…”

“…We assume prior covariances based partially on the MODIS and CALIPSO cross-validation from (Richardson et al, 2017), and calculate the observation covariance by perturbing atmospheric profiles. The calculation of the covariances is described in Sect.…”

“…Here we develop an optimal-estimationbased retrieval (Rodgers, 2000) for single-layer water clouds over oceans using nadir-view OCO-2 measurements and subject it to several idealised tests. This study goes beyond (Richardson et al, 2017) by (i) considering information content aspects to select groups of channels rather than combined super-channels, (ii) accounting for local meteorological conditions and (iii) adding cloud pressure thickness to the retrieval state vector. 5…”