Principle Component Analysis as a Tool for Detection of Land-Atmosphere Freeback Regimes Using Flux Site Observations

The land surface energy balance is an important factor in understanding the Earth’s climate system related to the coupling between land and atmosphere through the water cycle. Soil moisture is also a key determinant of interactions with the surface energy terms, such as net radiation and sensible/latent heat fluxes at the surface, affecting the lower atmosphere, daytime boundary layer development, and cloud formation. In recent years, a number of land-atmosphere coupling metrics have been devised to understand these interactions. Some are purely statistical in nature, while others are informed by theoretical considerations or understanding of the processes that link land variability back to the atmosphere.

We propose a new application of a well-known statistical approach to investigate the major relationships between soil moisture and the surface energy budget: principal component analysis (PCA). The methodology is demonstrated using flux tower site observations. Modes of variability are determined across the relatively small dimension of the various surface energy terms. The method is able to illuminate the dominant processes in each season by withholding, in turn, each of the energy terms from the analysis and quantifying the resulting impact. We find that across seasons, latent (sensible) heat flux is a more important term in summer (winter), but radiation is the most important factor in the whole season. Compared with the common approach using the Pearson correlation coefficient to determine water- and energy-limited regimes, PCA finds that latent (sensible) heat flux can be a more important component in energy- (water-) limited regimes, depending on season and land type. As PCA allows all surface energy terms to contribute to each mode of variability, this study gives new insights to the relationships between soil moisture and the surface energy balance.