Accurate characterization of cloud horizontal and vertical distributions is needed to assess the clouds produced by climate models. Large differences among satellite observations and between models and observations necessitate better estimates of uncertainties in satellite-derived cloud amounts, heights, and other properties and development of ways to account for the effects of multilayered clouds in the satellite retrievals. Determining errors in those cloud properties and providing guidance for future error reduction are necessary for improved global climate modeling. All previous estimates of the errors are limited to a few locations and cloud types. The proposed research will use CALIPSO-CloudSat (CSCAL) data products to
- Determine errors in cloud phase, top and base heights, optical depth, particle size, and liquid/ice water paths for standard CERES products, day and night as a function of surface and cloud type over the globe. From the results, we will develop refined parameterizations of cloud thickness and test new approaches to improve the algorithms using different spectral channels.
- Determine errors in multilayer cloud detection and retrieval from both Aqua & GEO data using 4 different approaches.
- Develop refined multilayer detection methods for night & daytime by using theoretical calculations at specific wavelengths to assess potential multilayered signals, using combinations of AMSRE, MODIS, GOES & CSCAL to evaluate and refine the techniques, and determine limits of each method & provide uncertainty estimates.
We will use CSCAL cloud layering, phase, optical depth, effective particle size, and liquid/ice water path data to address one of area of study, "Atmospheric processes that are influenced by cloud and aerosol vertical distribution" that supports the NASA objective to "conduct a program of research and technology development to advance Earth observation from space, improve scientific understanding, and demonstrate new technologies with the potential to improve future operational systems." Development of improved multi- (ML) and single-layered (SL) cloud detection and retrieval methods have a direct avenue to operational systems via our GOES icing and cloud assimilation programs, the continuing CERES program through the NOAA NPOESS system, and the future reprocessing of CERES cloud and flux data. Error assessments are a key component for using the retrieved products in weather, climate, and energy applications.
We will match MODIS, AMSR-E, GEO, and CSCAL data to develop a comprehensive set of error bars for cloud products being derived for CERES and other projects using a common set of retrieval algorithms on different satellites. Comparisons between the cloud properties derived from MODIS and GEO satellites with CSCAL cloud-truth data will be categorized according surface type, cloud layering, and time of day. We will use differences in the results to quantify uncertainties in each CERES product and improve the algorithms, reducing errors in SL cloud retrievals. We will develop a more globally robust method for inferring cloud thickness by correlating 3 CERES cloud parameters to CSCAL thicknesses. We will implement currently operative ML cloud detection methods & our own techniques and compare the results to matched CSCAL data to learn which ML clouds are detectable. We will learn how efficiently we detect them and their range of characteristics over a given surface type and layering condition. We will establish a radiance database associated with cloud layering to help refine one or more of the methods in terms of detection and retrieval. To assess and improve off-nadir performance of the SL and ML retrievals, we will use the GEO data, as if they were MODIS data, when the A-train passes in their fields of view, and repeat the analyses used for the MODIS data. The improvements will be implemented in the CERES reprocessing of Terra and Aqua data.
