Statistical extrapolation of vertically resolved cloud information from CloudSat / CALIPSO observations to regional swaths

Cloud base (ceiling) and internal structure (water content profiles) are historically difficult to measure from space. These parameters are vital to aviation, and truth on cloud vertical structure is needed to validate and improve weather forecast models. Surface observations of cloud base are extremely sparse and typically are limited to the altitude of the lowest cloud layer. Most environmental satellites carry passive radiometers, which observe the tops of the clouds and integrated information about cloud content. Information about the cloud base height must be inferred from gross assumptions on the vertical structure of the clouds. The most effective way to provide the detailed vertical profile information is via active sensors, such as radar and lidar. Such systems are only nascent to the satellite platform.

The NASA CloudSat (W-band radar) and CALIPSO (NdYAG lidar) active observing system missions, which have flown as part of the A-Train since Spring 2006, provide an unprecedented global dataset of cloud boundary (base, top, multi-layer) and internal water content structure information. Due to the non-scanning nature of these observing systems, the observations are limited to a 2-D cross section along the ground track, and therefore provide no direct information on the surrounding cloud field. At the head of the A-Train constellation, on board the Aqua satellite, is the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument. A passive radiometer with 36 narrow bands in the optical spectrum, MODIS provides the necessary information to determine cloud type, cloud top properties, and cloud water path over a 2330 km swath. What it cannot provide is the detailed vertical structure of an active observing system.

This research appeals to the synergy of A-Train observations to enable new capabilities unattainable from the individual components. Here, we present a new technique to estimate cloud profile information from CloudSat/CALIPSO/MODIS observations. The approach uses cloud class-dependent statistics to relate the limited active sensor data to the regional cloud field observed by MODIS. This work addresses the following science questions: How useful and robust is cloud classification information in the context of understanding cloud vertical structure? What is the spatial variability of cloud base and top as a function of these cloud classifications, and can this variability be used as a basis for relating a reduced set of observations to a broader region? How can we quantify the uncertainty of predictions based on such an approach in a way that makes them useful in the context of model validation and decision support systems?

Early results from this work indicate that there is skill in combining the active and passive measurements to translate 2-D swath information into 3-D cloud fields (for the topmost cloud layer). Applications of these ‘enhanced' cloud fields to aviation and forecast model validation, as well as limitations of the approach, will be discussed.