Contour surgery: a topological reconnection scheme for extended integrations using contour dynamics
Journal of Computational Physics
Machine learning, neural and statistical classification
Machine learning, neural and statistical classification
Journal of Computational Physics
| Hi-index | 0.01 |
The study of chaotic mixing is important for its potential to improve our understanding of fluid systems. Contour advection simulations provide a good model of the phenomenon by tracking the evolution of one or more contours or isolines of a trace substance to a high level of precision. The most accurate method of validating an advected contour is to divide the tracer concentration into discrete ranges and perform a maximum likelihood classification, a method that we term, ''isoline retrieval.'' Conditional probabilities generated as a result provide excellent error characterization. In this study, a water vapour isoline of 0.001 mass-mixing-ratio is advected over five days in the upper troposphere and compared with high-resolution Advanced Microwave Sounding Unit (AMSU) satellite retrievals. The goal is to find the same fine-scale, chaotic mixing in the isoline retrievals as seen in the advection simulations. Some of the filaments generated by the simulations show up in the conditional probabilities as areas of reduced probability. By rescaling the probabilities, the filaments may be revealed in the isoline retrievals proper with little effect on the overall accuracy. Limitations imposed by the specific context, i.e., water vapour retrieved with AMSU in the upper troposphere, are discussed. Nonetheless, isoline retrieval is shown to be a highly effective technique for atmospheric sounding, showing good agreement with both European Centre for Medium-range Weather Forecasts (ECMWF) assimilation data and radiosonde measurements.