Ana Barros (PI)
Duke University
barros@duke.edu

Predictability and Attribution Metrics to Characterize Land Controls in the Dynamical Evolution of Cloud and Precipitation Patterns in Models and Observations

The spatial patterns (e.g. orographic precipitation, hydroecological gradients) and temporal rythms (e.g. diurnal cycle, seasonality, active and break phases of  convective activity) of  water cycle processes studied through exploratory data analysis of remote sensing and model data show the emergence of coherent, complex order from the interplay between large-scale forcing and strong local constraints.  That is, land-atmosphere interactions exhibit regionally the nonlinear dynamical behavior of self-organizing phenomena.   Although there is a large body of evidence from observations and model studies linking precipitation occurrence to land-surface conditions, the space-time dynamics by which land-atmosphere feedback mechanisms control and are reflected in the dynamics of moist processes in the mid and upper troposphere are not well understood.  A basic research question is to establish measurable cause-effect relationships between land-atmosphere interactions within limited areas (as described by differential surface water and energy flux patterns) and the dynamical evolution of moist processes in the mid and lower troposphere regionally (as manifest by cloud and precipitation patterns): in other words, to isolate the contribution of land controls to the predictability of clouds and precipitation.  A critical research need is therefore to track and measure the complex space-time lifecycle of convective instabilities from their origin to the unfolding of clouds and the materialization of rainfall across the landscape.   Furthermore, metrics are necessary that can both support diagnostic studies and also permit systematically exploring the range of models’ behaviors that is consistent with observations. For this purpose, our working hypothesis is that the emergent behavior of land-atmosphere interactions associated with cloud and precipitation processes is consistent with deterministic chaos phenomena.  Our working assumption is that complex behavior in coupled systems can be best described in the framework of nonlinear dynamical systems and phase-space topologies.  We propose to build upon an existing theoretical and applied framework for the analysis of non-linear time-series, and to extend it to the context of spatial complexity with two overarching research objectives:1) to gain a better understanding and interpretation of land-atmosphere interactions in the predictability of hydrometeorological phenomena; and 2)  to establish objective metrics to track uncertainty propagation and evaluate models against observations that are based on physical dynamics rather then specific realizations (i.e. RMS, residuals,etc). Our research plan consists of conducting comparative studies using  high resolution MODIS  products for the last five years , and global and regional model simulations using ensembles of cloud and precipitation simulations  from the FSU superensemble database. Deliverables will include spatial fields of Finite-size Lyapunov exponents, space-time coupling indices including directionality and synchronicity.  This proposal addresses specific themes of the NRA with respect to predictability and model evaluation criteria.

+ Back to Participant Listing