Cloud retrievals from satellite data using optimal estimation: evaluation and application to ATSR

Abstract

Abstract. Clouds play an important role in balancing the Earth's radiation budget. Hence, it is vital that cloud climatologies are produced that quantify cloud macro and micro physical parameters and the associated uncertainty. In this paper, we present an algorithm ORAC (Oxford-RAL retrieval of Aerosol and Cloud) which is based on fitting a physically consistent cloud model to satellite observations simultaneously from the visible to the mid-infrared, thereby ensuring that the resulting cloud properties provide both a good representation of the short-wave and long-wave radiative effects of the observed cloud. The advantages of the optimal estimation method are that it enables rigorous error propagation and the inclusion of all measurements and any a priori information and associated errors in a rigorous mathematical framework. The algorithm provides a measure of the consistency between retrieval representation of cloud and satellite radiances. The cloud parameters retrieved are the cloud top pressure, cloud optical depth, cloud effective radius, cloud fraction and cloud phase.The algorithm can be applied to most visible/infrared satellite instruments. In this paper, we demonstrate the applicability to the Along-Track Scanning Radiometers ATSR-2 and AATSR. Examples of applying the algorithm to ATSR-2 flight data are presented and the sensitivity of the retrievals assessed, in particular the algorithm is evaluated for a number of simulated single-layer and multi-layer conditions. The algorithm was found to perform well for single-layer cloud except when the cloud was very thin; i.e., less than 1 optical depths. For the multi-layer cloud, the algorithm was robust except when the upper ice cloud layer is less than five optical depths. In these cases the retrieved cloud top pressure and cloud effective radius become a weighted average of the 2 layers. The sum of optical depth of multi-layer cloud is retrieved well until the cloud becomes thick, greater than 50 optical depths, where the cloud begins to saturate. The cost proved a good indicator of multi-layer scenarios. Both the retrieval cost and the error need to be considered together in order to evaluate the quality of the retrieval. This algorithm in the configuration described here has been applied to both ATSR-2 and AATSR visible and infrared measurements in the context of the GRAPE (Global Retrieval and cloud Product Evaluation) project to produce a 14 yr consistent record for climate research.

Known and unknown unknowns: uncertainty estimation in satellite remote sensing

“…Due to the 4708 A. C. Povey and R. G. Grainger: Uncertainty estimation in satellite remote sensing lack of information about the vertical extent of the cloud, it is common to assume the cloud is infinitely thin (e.g. Poulsen et al, 2012), and the measurand would be more accurately described as the "effective cloud radiating height".…”

Synergy of stereo cloud top height and ORAC optimal estimation cloud retrieval: evaluation and application to AATSR

“…The ORAC algorithm (Poulsen et al, 2012, Watts et al, 1998) is an optimal estimation retrieval that can be used to determine both aerosol and cloud properties from visible/infrared satellite radiometers. In the case of cloud retrievals the algorithm fits radiances computed from LUTs (look-up tables) created from DIScrete Ordinates Radiative Transfer (DISORT) (Stamnes et al, 1988) to the TOA (top of atmosphere) signal measured by the satellite by varying the cloud optical depth, effective radius cloud top pressure, phase and surface temperature simultaneously.…”

“…Near the point of intersection, due to stereo algorithm smoothing effects (Zabih and Woodfill, 1994), the census stereo retrieved CTH relates to the overlying cloud feature, not the underlying water cloud, resulting in inaccurate a priori values. The ORAC+stereo retrieval is constrained against these CTH a priori values, and the degree of the constraint is dependent on the provided a priori error (Poulsen et al, 2012). In our analysis the census stereo a priori was found to be wildly inaccurate at intersections between water clouds and overlying clouds (∼ 8 km mean CTH vs. ∼ 3 km mean CTH retrieved by ORAC), in such instances the assumed error on the a priori is unsuitable, and does not provide a sensible constraint for the retrieval.…”

“…In order to reproduce the observed top of atmosphere brightness temperatures the algorithm must modify other, unconstrained parameters. The inadequacy of the CTH a priori values and the provided error constraints are therefore compensated for in the COD and RE parameters, which are unconstrained in the ORAC retrieval (Poulsen et al, 2012). This results in the observed differences between the ORAC and the ORAC+stereo COD and RE retrievals for water clouds located in proximity to clouds at increased elevations.…”

“…The ORAC (Optimal Retrieval of Aerosol and Cloud) algorithm (Poulsen et al, 2012;Watts et al, 1998) employs the optimal estimation approach (Rogers, 2000) based on radiometric retrieval principles and has been extensively applied to the Along Track Scanning Radiometer Instruments (ATSR), specifically ATSR-2 (1995, Mutlow et al, 1999 and the Advanced-ATSR (2002, LlewellynJones et al, 2001. The radiometric configuration of ATSR-2 and AATSR comprises seven channels at 0.…”