Application Of Conditional Nonlinear Optimal Perturbation To Ensemble Prediction

By using a barotropic quasi-geostrophic model and a T21L3 baroclinic quasi-geostrophic model, we explore the possible application of conditional nonlinear optimal perturbation (CNOP) to medium range ensemble prediction. Based on the above researches, a method of generation of special ensemble perturbation for weather regime transition is studied. Major results of this thesis are as follows:1. CNOP is applied to ensemble prediction study by using a barotropic quasi-geostrophic model under the perfect model assumption.a) SVs and CNOPs have been utilized to generate the initial perturbations for ensemble prediction experiments. The results are compared for forecast lengths of up to 14 days. It is found that the forecast skill of samples, in which the first SV is replaced by CNOP, is comparatively higher than that of samples composed of only SVs in the medium range (day 6 ~ day 14). This conclusion is valid under the condition that analysis error is a kind of fast-growing ones regardless of its magnitude, whose nonlinear growth is faster than that of SV in the later part of the forecast.b) Similarity index and empirical orthogonal function (EOF) analysis are performed to explain the above numerical results.c) Ensemble prediction made by the new method of CNOP has higher reliability than that by SV in the medium range by the analysis of spread/skill relations and Talagrand diagrams.d) Sensitivity analysis shows that local CNOP may play an even more important role than global CNOP in ensemble prediction.2. The differences between CNOP and SV are revealed from the initial patterns and their evolutions with a T21L3 baroclinic quasi-geostrophic model, which shows the impact of nonlinearity on predictability.a) CNOP depends sensitively on the norm chosen, which is similar to SV. The streamfunction squared norm yields small-scale disturbances, the total energy norm is characterized by intermediate-scale disturbances, and the enstrophy norm is typified by large-scale disturbance with large zonal flow contribution.b) Both SV and CNOP are much localized in the streamfunction squared norm or the total energy norm. However, with the increasing of the initial constraint condition or the optimization time interval, CNOP has less localized structures than the corresponding SV, What's more, the wave train structures may even be found in the whole zonal direction in the northern hemisphere.c) When SV in the total energy norm over some optimization time interval has not produced a clear blocking-type circulation, CNOP appears to have made the regime transition, which shows that the nonlinearity plays a fundamental role for studying weather regime transitions in the case.3. CNOP is applied to ensemble prediction study by using a T21L3 quasi-geostrophic model under the perfect model assumption.a) The forecast skill of geopotential height at 500 hPa in north hemisphere is verified by anomaly correlation coefficient (ACC). The results show that if the predictability of the atmosphere is low, for example, there is a weather regime transition in the medium range weather forecasts, the forecast skill is improved by the introduction of CNOP compared with SV.b) In the construction of ensemble perturbations, compared with global CNOP, local CNOP may play an even more important role.4. Conditional nonlinear optimal perturbations that trigger the blocking onset, based on blocking index, is acquired by nonlinear optimization technique, which may provide another possibility to determine perturbations for prediction of weather regime transition.