-
[show abstract]
[hide abstract]
ABSTRACT: Radiative transfer calculations in atmospheric models are computationally expensive, even if based on simplifications such as the δ-two-stream approximation. In most weather prediction models these parameterisation schemes are therefore called infrequently, accepting additional model error due to the persistence assumption between calls. This paper presents two so-called adaptive parameterisation schemes for radiative transfer in a limited area model: A perturbation scheme that exploits temporal correlations and a local-search scheme that mainly takes advantage of spatial correlations. Utilising these correlations and with similar computational resources, the schemes are able to predict the surface net radiative fluxes more accurately than a scheme based on the persistence assumption. An important property of these adaptive schemes is that their accuracy does not decrease much in case of strong reductions in the number of calls to the δ-two-stream scheme. It is hypothesised that the core idea can also be employed in parameterisation schemes for other processes and in other dynamical models.
Atmospheric Chemistry and Physics. 01/2007;
-
[show abstract]
[hide abstract]
ABSTRACT: In this study, the statistical properties of a range of measurements are compared with those of their surrogate time series. Seven different records are studied, amongst others, historical time series of mean daily temperature, daily rain sums and runoff from two rivers, and cloud measurements. Seven different algorithms are used to generate the surrogate time series. The best-known method is the iterative amplitude adjusted Fourier transform (IAAFT) algorithm, which is able to reproduce the measured distribution as well as the power spectrum. Using this setup, the measurements and their surrogates are compared with respect to their power spectrum, increment distribution, structure functions, annual percentiles and return values. It is found that the surrogates that reproduce the power spectrum and the distribution of the measurements are able to closely match the increment distributions and the structure functions of the measurements, but this often does not hold for surrogates that only mimic the power spectrum of the measurement. However, even the best performing surrogates do not have asymmetric increment distributions, i.e., they cannot reproduce nonlinear dynamical processes that are asymmetric in time. Furthermore, we have found deviations of the structure functions on small scales.
Nonlinear Processes in Geophysics. 01/2006;
-
[show abstract]
[hide abstract]
ABSTRACT: A stochastic version of the Iterative Amplitude Adjusted Fourier Transform (IAAFT) algorithm is presented. This algorithm is able to generate so-called surrogate time series, which have the amplitude distribution and the power spectrum of measured time series or fields. The key difference between the new algorithm and the original IAAFT method is the treatment of the amplitude adjustment: it is not performed for all values in each iterative step, but only for a fraction of the values. This new algorithm achieves a better accuracy, i.e. the power spectra of the measurement and its surrogate are more similar. We demonstrate the improvement by applying the IAAFT algorithm and the new one to 13 different test signals ranging from rain time series and 3-dimensional clouds to fractal time series and theoretical input. The improved accuracy can be important for generating high-quality geophysical time series and fields. The traditional application of the IAAFT algorithm is statistical nonlinearity testing. Reassuringly, we found that in most cases the accuracy of the original IAAFT algorithm is sufficient for this application.
Nonlinear Processes in Geophysics. 01/2006;
