Nighttime polar cloud detection with MODIS

Abstract

Cloud detection is the first step in studying the role of polar clouds in the global climate system with satellite data. In this paper, the cloud detection algorithm for the Moderate Resolution Imaging Spectrometer (MODIS) is evaluated with model simulations and satellite data collocated with radar/lidar observations at three Arctic and Antarctic stations. Results show that the current MODIS cloud mask algorithm performs well in polar regions during the day but does not detect more than 40% of the cloud cover over the validation sights at night. Two new cloud tests utilizing the 7.2 μm water vapor and 14.2 μm carbon dioxide bands, one new clear-sky test using the 7.2 μm band, and changes to the thresholds of several other tests are described. With the new cloud detection procedure, the misidentification of cloud as clear decreases from 44.2% to 16.3% at the two Arctic stations, and from 19.8% to 2.7% at the Antarctic station.

Introduction

The variation of cloud amount over the polar regions strongly influences planetary albedo gradients and surface energy exchanges (Key & Barry, 1989), which, in turn, affect regional and global climate (Curry et al., 1996). While cloud radiative properties are important in the study of clouds in polar climate systems, the first step is to determine when and where clouds exist. Limited surface observations of cloud cover in the Arctic and Antarctic makes the use of satellite data necessary. However, the detection of polar clouds is inherently difficult due to poor thermal and visible contrast between clouds and the underlying snow/ice surface, small radiances from the cold polar atmosphere, and ubiquitous temperature and humidity inversions in the lower troposphere (Lubin & Morrow, 1998).

Polar cloud detection from remote sensing data has been an area of active research during the past decade (Gao et al., 1998). The International Satellite Cloud Climatology Project (ISCCP) employs a combination of spectral, temporal, and spatial tests to estimate clear-sky radiances and values of cloud forcing Key & Barry, 1989, Rossow & Garder, 1993, Rossow & Schiffer, 1991, Rossow et al., 1993 and increases the sensitivity of low-level cloud detection over snow and ice in polar regions by use of a new threshold test on 3.7 μm radiances (Rossow & Schiffer, 1999). The TOVS Polar Pathfinder cloud detection scheme uses a series of spectral tests to determine if a pixel is clear or cloudy (Schweiger et al., 1999). Statistical classification procedures, including maximum likelihood and Euclidean distance methods, have been applied in cloud detection algorithms Ebert, 1989, Key, 1990, Key et al., 1989, Welch et al., 1988, Welch et al., 1990, Welch et al., 1992. Single- and bispectral threshold methods have been developed and applied to polar data Ackerman, 1996, Gao et al., 1998, Inoue, 1987a, Inoue, 1987b, Minnis et al., 2001, Spangenberg et al., 2001, Spangenberg et al., 2002, Yamanouchi et al., 1987.

The Moderate Resolution Imaging Spectrometer (MODIS) on the NASA Terra and Aqua satellites provides an unprecedented opportunity for earth remote sensing. Its broad spectral range (36 bands between 0.415–14.235 μm), high spatial resolution (250 m for 5 bands, 500 m for 5 bands, and 1000 m for 29 bands), frequent observations of polar regions (28 times a day), and low thermal band instrument noise (roughly 0.1 K for a 300 K scene) provide a number of possibilities for improving cloud detection.

The goal of this study is to present improvements to the MODIS cloud mask algorithm for the detection of polar clouds at night. Changes to some of the current spectral tests are recommended, and new tests are proposed. The physical basis for these tests are described and supported by radiative transfer simulations. Validation of the satellite-derived cloud detection results is accomplished with surface-based cloud radar and lidar data from Alaska and the South Pole. It will be shown that nighttime cloud detection in polar regions with MODIS can achieve a high level of accuracy and is far more robust than what may be obtained with the Advanced Very High Resolution Radiometer (AVHRR). These enhancements will be incorporated into the next version of the NASA MODIS cloud mask.