Capturing time series of stochastic forcing

Linear Inverse Modeling (LIM) estimates the deterministic dynamics of a linear system and the statistical structure of stochastic driving noise. Having the time series of that noise would be very useful for diagnostic studies of interactions between weather and climate in both models and in the real system, and for developing stochastic parameterizations for stochastic numerical models. While there already exist several methods of diagnosing plausible sources of stochasticity from data, none of these is guaranteed to reproduce dynamical consistency. For example, in a fluid system, contemporaneous correlations between state variables and stochastic forcing is not zero, thus precluding many regression techniques. In this poster, we present a dynamically consistent method of using LIM output to capture the elusive time series of stochastic forcing, apply it seasonal sea surface temperature data, and relate the stochastic forcing to meteorological variables.