Determination of characteristic features of cloud liquid water from satellite microwave measurements

In this paper we present an algorithm for determining liquid water path (LWP) in clouds using satellite microwave measurements from the special sensor microwave/imager (SSM/I). This physically based algorithm is derived using the guidance of a 32‐stream plane‐parallel model of microwave radiative transfer in the cloudy atmosphere. Our investigation using both radiative transfer model and observed SSM/I data indicates that the retrieved LWP is sensitive to the emission temperature of the liquid water, to the presence of large hydrometeors, to the nonuniformity of the rainfall in the satellite field of view, and to the radiation from atmosphere and ocean. Therefore we have incorporated cloud temperature into our algorithm using ancillary infrared satellite data. To account for the effects of rainfall on the LWP retrievals, we use a threshold for the onset of precipitation to determine whether there is any precipitation in the satellite field of view. Two separate retrieval schemes are derived for non‐precipitating and precipitating clouds. A clear‐sky radiation scheme is developed to separate radiation of atmosphere and ocean from that of liquid water. This scheme could be easily tuned by observed microwave data so that the bias in LWP retrieval is minimized. This algorithm is indirectly validated by comparing retrieved values of LWP for clear versus cloudy situations. The retrievals are examined for regions in the tropical western Pacific Ocean, the North Atlantic Ocean, and the subtropical southeast Pacific Ocean. The retrievals obtained using our algorithm are also compared with simultaneous retrievals using other LWP algorithms described in the literature. The retrievals are compared for clear sky, liquid cloud only, cloud with ice, raining water cloud, and raining cloud with ice to assess and interpret the performance of the algorithms in each of these situations. The present algorithm is shown to perform consistently well in all three regions under all conditions. It is determined that features of our algorithm that results in more consistent global retrievals of LWP are inclusion of the tunable clear‐sky scheme, estimation of cloud temperature using satellite infrared radiances, and consideration of precipitation.