Application of the Surrogate Test to Detect Dynamic Non-Linearity in Ground-Level Ozone Time-Series from Berlin
Recent applications of non-parametric methods to forecast ground level ozone concentrations are based on dynamic non-linearity of the data series. We explain the surrogate method to test this assumption, illustrate the method with non-linear data generated by the Lorenz system, and discuss our test results for Berlin ozone time-series. We find that the null-hypothesis of linearity is clearly rejected for 12- and 24-step-ahead predictions of hourly ozone concentrations.
KeywordsPrediction Accuracy Ozone Concentration Atmospheric Environment Query Point Lorenz System
Unable to display preview. Download preview PDF.
- Schlink, U., Herbarth, O., Richter, M.: Ozone data analysis to develop an immediate warning system. Proceedings of the Air Pollution’ 96 conference 1996.Google Scholar
- Takens, F.: Detecting Strange Attractors in Turbulence. In Rand, D.A., Young, L.S. (Eds.): Dynamical Systems and Turbulence. Lecture Notes in Mathematics 898, 366–381, Berlin: Springer 1981.Google Scholar
- Kantz, H., Schreiber, T.: Nonlinear Time Series Analysis. Cambridge: Cambridge University Press 2000.Google Scholar
- Merkwirth, C.: TSTOOL Documentation, http://www.dpi.physik.uni-goettingen.de/tstool. (2000).Google Scholar