Quantifying the Variability of Hyperspectral Earth-reflected Radiation for Climate Studies

• Main Author: Roberts, Yolanda
• Format: Text
• Language: unknown
• Published: CU Scholar 2012
• Subjects: Hyperspectral Imaging; Multivariate Analysis; Remote Sensing; Atmospheric Sciences; Climate; Sea ice
• Online Access: https://scholar.colorado.edu/atoc_gradetds/27; https://scholar.colorado.edu/cgi/viewcontent.cgi?article=1028&context=atoc_gradetds

An accurate assessment of the Earth's energy budget is essential to understanding how the Earth's climate is changing and what processes and feedbacks are causing those changes. This is difficult to achieve, in part, because reflected solar irradiance, and therefore albedo, is a challenging quantity to measure from space with sufficient accuracy to monitor climate changes. An alternative to irradiance or albedo is directly measured spectral radiance, which provides information about the Earth's atmospheric composition and surface properties that impact albedo variability. We have applied multivariate spectral decomposition techniques, such as principal component analysis (PCA), to quantify the variability of Earth-reflected hyperspectral solar radiance measured by the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY) aboard ENVISAT. Using multivariate analysis we explored the potential for directly measured hyperspectral Earth-reflected solar radiance to provide sufficient information to study changes in Earth's climate based on the quantified variability of the data. The spectral signatures of the principal components (PCs) reveal that clouds, water vapor, vegetation, and sea ice are among the physical variables that explain the largest fraction of the SCIAMACHY data variance. The extraction of the spectral, spatial, and temporal variability in reflected shortwave hyperspectral radiance using multivariate analysis provides an alternate and complementary approach to applying inverse methods to space-based observations for climate studies. Observation System Simulation Experiments (OSSEs) have been used to simulate solar radiation measurements during the twenty-first century for the NASA Climate and Absolute Radiance and Refractivity Observatory (CLARREO) hyperspectral shortwave instrument being designed. Comparing the spectral shapes of the OSSE and SCIAMACHY PCs shows that the OSSE has a similar variance distribution to that observed by SCIAMACHY. We developed a quantitative comparison technique to quantify the degree to which the OSSE reproduces the variability within Earth's climate system relative to observations. These comparisons showed that the OSSE spectral variability is close to that observed by SCIAMACHY. In addition, for the first time, the near-decadal temporal variability of observed reflectance measured between 2002 and 2010 was quantified; the variance drivers in the nearly decadal variability of SCIAMACHY measurements exhibited temporal signals of physical variables such as the location of the Intertropical Convergence Zone and the annual cycle of the cryosphere. The intersection also allowed for the direct comparison between the temporal variability of SCIAMACHY and OSSE reflectance at the beginning of the twenty-first century. Finally, we quantified the centennial variability of OSSE output during the twenty-first century, demonstrating that the reflectance spectra simulated from the A2 emission scenario model output exhibited secular trends over the simulation period. Applying the multivariate techniques presented in this thesis to evaluate the OSSE centennial variability enables the development of trend detection methods to further study the temporal variability of reflected solar radiation.