|POSTPONED DUE TO WEATHER/CAMPUS CLOSING
March 5, 2015
Challenges in the production of observational datasets: The MODIS cloud product perspective
Earth Science Division, NASA/GSFC
There are number of common challenges in the production of observational datasets designed for physical processes studies and/or time series analyses for climate studies. These include: understanding the instrument and its limitations, the applicability of forward radiative transfer models, retrieval algorithm choices and associated data quality filtering, assumptions and ancillary datasets associated with quantities that are not part of the retrieval space, and incorporation of all the above in assessing pixel-level retrieval uncertainties. Lessons learned in developing cloud products for MODIS (Moderate Resolution Imaging Spectroradiometer) on the NASA EOS Terra and Aqua satellites are discussed, with an emphasis on the sensitivity of cloud properties to algorithm choices and estimates of baseline retrieval uncertainties.
The optical and microphysical structure of clouds is of fundamental importance for understanding a variety of radiation and precipitation processes. The MODIS cloud optical product provides daytime global 1km retrievals of cloud thermodynamic phase, optical thickness, effective particle size, and the derived cloud water path. The Collection 6 version of the product (MOD06/MYD06 for MODIS Terra and Aqua, respectively) provides separate effective radii results using the 1.6, 2.1, and 3.7 µm MODIS spectral channels allowing for some assessment of cloud heterogeneity. In addition, unlike the previous version, Collection 6 attempts retrievals for pixel populations that are flagged as likely partly cloudy (or highly inhomogeneous) by the so-call Clear Sky Restoral (CSR) algorithm, but allows a user to isolate or filter out the populations via Quality Assessment (QA) assignments. Collection 6 also includes a new phase algorithm, new ice cloud radiative models, and improved pixel-level uncertainty calculations. In this presentation, we discuss global and regional statistics of cloud properties and their sensitivity to choices made in the development of the Collection 6 algorithm.