Although typhoon forecasting has been improved notably over the years, it remains a difficult challenge to make timely and accurate rainfall forecasts, which is crucial to saving lives and reducing damage. Under the assumption that a proportional relationship exists between the accumulated rainfall of a typhoon hitting Taiwan and climatological accumulated rainfall, the Rainfall Potential Analog Forecasting (RaPAF) technique is proposed for predicting the accumulated rainfall distribution when a typhoon makes landfall in Taiwan by an analog forecasting technique. The RaPAF is a combined method involving an operational SSM/I rainfall retrieval algorithm, the tropical rainfall potential technique, and an empirical relationship of precipitation associated with typhoon. In addition, The RaPAF needs information on forecast of typhoon track as an input. In the case study of Typhoon Morakot (2009), it is proven that the method not only provided a more accurate accumulated rainfall distribution prediction and that the probable mapping of the maximum precipitation areas was also close to the actual observations. The calculations involved were easy and quick to implement, enabling it to serve as a viable tool for producing accumulated rainfall forecasts and potential damage assessments during different stages after typhoon genesis. The RaPAF technique may be useful for accumulated rainfall prediction of typhoon.
We investigate the feasibility of dynamical seasonal predictions of the interannual variability of the mean location of typhoon formation in the western North Pacific (WNP) and its physical mechanisms during the active typhoon season from June to October. We performed seven-month integrations for 28 years starting from late April using the El Niño Southern Oscillation (ENSO) prediction system of the Japan Meteorological Agency. Typhoons detected with an objective method using model outputs are verified with best track data from the Regional Specialized Meteorological Center, Tokyo. The good overall deterministic skill in predicting the interannual variability of the mean location of typhoon formation fundamentally stems from the ability to predict the interannual variability of the atmospheric circulation in the WNP influenced by ENSO. The interannual variability of indices representing a latitudinal shift of the atmospheric circulation in the WNP is better predicted than that of indices representing a longitudinal shift in this experiment. This difference in skill among these indices provides a physical basis for the difference in prediction skill between the mean latitude and longitude regarding the interannual variability of typhoon formation. Probabilistic predictions also demonstrate the skillful predictions of the mean location of typhoon formation for tercile-based categories. Therefore, both deterministic and probabilistic predictions using our ENSO prediction system provide useful information about the mean location of typhoon formation.
In the present study, we perform a set of numerical simulations in a moderately stable boundary layer with four types of subgrid-scale parameterization schemes and attempt to evaluate the error of the vertical flux for these schemes in terms of self-consistency on the basis of the Germano identity. If the effects of grid-scale components in higher wavenumbers are excluded from the analysis, the error estimated by the Germano identity is insensitive to the reference data utilized. The subgrid-scale flux, evaluated by the Smagorinsky model, tends to excessively weaken the positive temperature gradient at the top of the boundary layer with decreasing model resolution. The Deardorff and two-part models overestimate the subgrid-scale temperature flux at a coarser resolution, and the dynamic Smagorinsky model tends to underestimate both the subgrid-scale momentum and the temperature fluxes throughout the entire boundary layer. The underestimation of the subgrid-scale flux found in the dynamic Smagorinsky model could be attributed to a low correlation between the resolved and the parameterized components.
A new scheme, termed Vortex Initialization with the Assimilation of Retrieved Variables (VIRV), is presented to improve the initialization of regional numerical model for Tropical Cyclone (TC) prediction. In this scheme, the horizontal winds in Planetary Boundary Layer (PBL) and the sea level pressure (SLP), retrieved from Quick Scatterometer (QuikSCAT) data obtained using a modified University of Washington Planetary Boundary Layer (UWPBL) model, are assimilated with a cycled three-dimensional variational (3DVAR) technique to produce the initialized analysis. The procedures of retrieval are implemented under the joint dynamical constraints of the gradient wind, secondary circulation, and thermal stratification. Moreover, in order to improve the analysis of TC intensity, the roughness parameterization in the UWPBL model was modified for the case of strong surface wind. The sensitivities of the structure, intensity, and track of TC to the VIRV are then examined by two numerical experiments for TC Bilis (2006) and TC Fung-wong (2008). The maximum Wind Speed (MWS) and minimum Sea Level Pressure (MSLP) retrieved from the QuikSCAT data obtained using the modified UWPBL model show more agreement with the observations relative to those derived from the analysis of the National Center for Environmental Prediction (NCEP global model). The analysis of TC intensity cfm enhanced using VIRV by modifying the low-level (upper-level) convergence (divergence), vertical shear of horizontal wind, transportation of moisture. Significant improvement on 48-h TC simulation is identified in the MWS, with 22.8% error reduction. In particular, the Modification of Roughness Parameterization (MRP) enhanced the simulation of MWS by 6.9%. Finally, the VIRV also reduces the simulation error in the track of TC by affecting the steering flow throughout the troposphere.
The Skyrad software package (SKYRAD.pack), which is used to analyze sky radiometer data, utilizes aerosol retrieval algorithms that are based on the Mie theory. In modeling the optical properties of dust-like aerosols, however, it is necessary to account for particle nonsphericity. We applied aerosol retrieval algorithms based on the spheroid model to SKYRAD.pack. We tested the model using simulated data from aerosol models and verified that dust-like aerosol optical properties can be accurately retrieved using the spheroid model. Moreover, we analyzed sky radiometer data collected in Tsukuba, Japan, in April 2006. The results demonstrate that when the diffuse sky flux normalized by the direct flux is used, i.e., the aerosol optical thickness is unknown, the spheroid model can retrieve aerosol optical properties more accurately than the sphere model in the analysis of observations in the case of Asian dust. Furthermore, if solar calibration constants can be determined, aerosol optical properties can be accurately retrieved in both the case based on the Mie theory and the case based on the spheroid model.
To solve the problems of the Mellor-Yamada-Nakanishi-Niino (MYNN) model, we attempted to modify it by following the method proposed by Canuto et al. (2008) for the stable stratification case. In contrast to the original MYNN model, the modified model has no critical Richardson number, and the effect of turbulent motions remains for all Richardson number values in the level 2 model. Furthermore, the modified model possesses an advantage even for the level 2.5 and level 3 models. The flux Richardson number is unbounded, and a lower limit of the turbulent kinetic energy must be imposed in the original MYNN model; these features do not appear in the modified model.