What's Wrong with Errors?
Prof. Zoltan Toth
Global Systems Division, NOAA, USA
Numerical models of the atmosphere and coupled Earth system are designed to simulate selected processes of the natural systems. When a simulation starts from an initial condition that in the model’s space captures a part of reality, it is called a forecast. Errors in the initial (analysis) and subsequent simulated states of natural systems are notoriously difficult to estimate or even define, as reality itself is unaccessible. Amidst lingering ambiguity, a unified concept of errors remain elusive.
Dictionary definitions suggest the word “error” is related to being wrong, ignorant, or making mistakes. To what extent are these concepts applicable in the context of weather forecasting? Are errors anthropogenic or natural phenomena? What errors may be due to ignorance, mistakes, being wrong, or may simply be unavoidable? What errors originate from the observing, data assimilation, and modeling systems? And how errors relate to the predictability of the atmosphere? This presentation will review the sources, types, behavior, and estimation of analysis and forecast errors.