scholarly journals Wavelet Support Vector Machines for Forecasting Precipitation in Tropical Cyclones: Comparisons with GSVM, Regression, and MM5

2012 ◽  
Vol 27 (2) ◽  
pp. 438-450 ◽  
Author(s):  
Chih-Chiang Wei

Abstract This study presents two support vector machine (SVM) based models for forecasting hourly precipitation during tropical cyclone (typhoon) events. The two SVM-based models are the traditional Gaussian kernel SVMs (GSVMs) and the advanced wavelet kernel SVMs (WSVMs). A comparison between the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) and statistical models, including SVM-based models and linear regressions (regression), was made in terms of performance of rainfall prediction at the Shihmen Reservoir watershed in Taiwan. Data from 73 typhoons affecting the Shihmen Reservoir watershed were included in the analysis. This study designed six attribute combinations with different lag times for the forecast target. The modified RMSE, bias, and estimated threat score (ETS) results were employed to assess the predicted outcomes. Results show that better attribute combinations for typhoon climatologic characteristics and typhoon precipitation predictions occurred at 0-h lag time with modified RMSE values of 0.288, 0.257, and 0.296 in GSVM, WSVM, and the regression, respectively. Moreover, WSVM having average bias and ETS values close to 1.0 gave better predictions than did the GSVM and regression models. In addition, Typhoons Zeb (1998) and Nari (2001) were selected for comparison between the MM5 model output and the developed statistical models. Results showed that the MM5 tended to overestimate the peak and cumulative rainfall amounts while the statistical models were inclined to yield underestimations.

2019 ◽  
Vol 58 (6) ◽  
pp. 1219-1232
Author(s):  
Yu-Fen Huang ◽  
Yi-Leng Chen

AbstractThe seasonal variations of rainfall over the island of Hawaii are studied using the archives of the daily model run from the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5) from June 2004 to February 2010. Local effects mainly drive the rainfall on the Kona coast in the early morning and the lower slopes in the afternoon. During the summer, the incoming trade winds are more persistent and moister than in winter. The moisture content in the wake zone is higher than open-ocean values because of the convergent airflow associated with dual counterrotating vortices. As the westerly reversed flow moves toward the Kona coast, it decelerates with increasing moisture and a moisture maximum over the coastal area, especially in the afternoon hours in summer months. The higher afternoon rainfall on the Kona lower slopes in summer than in winter is caused by a moister (>6 mm) westerly reversed flow bringing moisture inland and merging with a stronger upslope flow resulting from solar heating. Higher nocturnal rainfall off the Kona coast in summer than in winter is caused by the low-level convergence between a moister westerly reversed flow and offshore flow. On the windward slopes, the simulated rainfall accumulation in winter is higher because of frequently occurring synoptic disturbances during the winter storm season. Nevertheless, early morning rainfall along the windward coast and afternoon rainfall over the windward slopes of the Kohala Mountains is lower in winter because the incoming trades are drier.


2008 ◽  
Vol 136 (7) ◽  
pp. 2488-2506 ◽  
Author(s):  
Qingqing Li ◽  
Yihong Duan ◽  
Hui Yu ◽  
Gang Fu

Abstract In this study, the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) is used to simulate Typhoon Rananim (2004) at high resolution (2-km grid size). The simulation agrees well with a variety of observations, especially for intensification, maintenance, landfall, and inner-core structures, including the echo-free eye, the asymmetry in eyewall convection, and the slope of the eyewall during landfall. The asymmetric feature of surface winds is also captured reasonably well by the model, as well as changes in surface winds and pressure near the storm center. The shear-induced vortex tilt and storm-relative asymmetric winds are examined to investigate how vertical shear affects the asymmetric convection in the inner-core region. The inner-core vertical shear is found to be nonunidirectional, and to induce a nonunidirectional vortex tilt. The distribution of asymmetric convection is, however, inconsistent with the typical downshear-left pattern for a deep-layer shear. Qualitative agreement is found between the divergence pattern and the storm-relative flow, with convergence (divergence) generally associated with asymmetric inflow (outflow) in the eyewall. The collocation of the inflow-induced lower-level convergence in the boundary layer and the lower troposphere and the midlevel divergence causes shallow updrafts in the western and southern parts of the eyewall, while the deep and strong upward motion in the southeastern portion of the eyewall is due to the collocation of the net convergence associated with the strong asymmetric flow in the midtroposphere and the inflow near 400 hPa and its associated divergence in the outflow layer above 400 hPa.


2006 ◽  
Vol 134 (3) ◽  
pp. 897-918 ◽  
Author(s):  
M. Chiriaco ◽  
R. Vautard ◽  
H. Chepfer ◽  
M. Haeffelin ◽  
J. Dudhia ◽  
...  

Abstract The ability of the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5) to simulate midlatitude ice clouds is evaluated. Model outputs are compared to long-term meteorological measurements by active (radar and lidar) and passive (infrared and visible fluxes) remote sensing collected at an atmospheric observatory near Paris, France. The goal is to understand which of four microphysical schemes is best suited to simulate midlatitude ice clouds. The methodology consists of simulating instrument observables from the model outputs without any profile inversion, which allows the authors to use fewer assumptions on microphysical and optical properties of ice particles. Among the four schemes compared in the current study, the best observation-to-simulations scores are obtained with Reisner et al. provided that the particles’ sedimentation velocity from Heymsfield and Donner is used instead of that originally proposed. For this last scheme, the model gives results close to the measurements for clouds with medium optical depth of typically 1 to 3, whatever the season. In this configuration, MM5 simulates the presence of midlatitude ice clouds in more than 65% of the authors’ selection of observed cloud cases. In 35% of the cases, the simulated clouds are too persistent whatever the microphysical scheme and tend to produce too much solid water (ice and snow) and not enough liquid water.


2006 ◽  
Vol 63 (1) ◽  
pp. 19-42 ◽  
Author(s):  
Scott A. Braun ◽  
Michael T. Montgomery ◽  
Zhaoxia Pu

Abstract The fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) is used to simulate Hurricane Bonnie at high resolution (2-km spacing) in order to examine how vertical wind shear impacts the distribution of vertical motion in the eyewall on both the storm and cloud scale. As in many previous studies, it is found here that the shear produces a wavenumber-1 asymmetry in the time-averaged vertical motion and rainfall. Several mechanisms for this asymmetry are evaluated. The vertical motion asymmetry is qualitatively consistent with an assumed balance between horizontal vorticity advection by the relative flow and stretching of vorticity, with relative asymmetric inflow (convergence) at low levels and outflow (divergence) at upper levels on the downshear side of the eyewall. The simulation results also show that the upward motion portion of the eyewall asymmetry is located in the direction of vortex tilt, consistent with the vertical motion that required to maintain dynamic balance. Variations in the direction and magnitude of the tilt are consistent with the presence of a vortex Rossby wave quasi mode, which is characterized by a damped precession of the upper vortex relative to the lower vortex. While the time-averaged vertical motion is characterized by ascent in a shear-induced wavenumber-1 asymmetry, the instantaneous vertical motion is typically associated with deep updraft towers that generally form on the downtilt-right side of the eyewall and dissipate on the downtilt-left side. The updrafts towers are typically associated with eyewall mesovortices rotating cyclonically around the eyewall and result from an interaction between the shear-induced relative asymmetric flow and the cyclonic circulations of the mesovortices. The eyewall mesovortices may persist for more than one orbit around the eyewall and, in these cases, can initiate multiple episodes of upward motion.


2011 ◽  
Vol 26 (1) ◽  
pp. 26-43 ◽  
Author(s):  
P. Goswami ◽  
S. Mallick

Abstract One factor that limits skill of the numerical models is the bias in the model forecasts with respect to observations. Similarly, while the mesoscale models today can support horizontal grid spacing down to a few kilometers or fewer, downscaling of model forecasts to arrive at station-scale values will remain a necessary step for many applications. While generic improvement in model skill requires parallel and comprehensive development in model and other forecast methodology, one way of achieving skill in station-scale forecasts without (intensive effort) calibration of the model is to implement an objective bias correction (referred to as debiasing). This study shows that a nonlinear objective debiasing can transform zero-skill forecasts from a mesoscale model [fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5)] to forecasts with significant skill. Twelve locations over India, representing urban sites in different geographical conditions, during May–August 2009 were considered. The model MM5 was integrated for 24 h with initial conditions from the National Centers for Environmental Prediction Global Forecast System (final) global gridded analysis (FNL) for each of the days of May–August 2009 in a completely operational setting (without assuming any observed information on dynamics beyond the time of the initial condition). It is shown that for all the locations and the four months, the skill of the debiased forecast is significant against essentially zero skill of raw forecasts. The procedure provides an applicable forecast strategy to attain realizable significant skill in station-scale forecasts. Potential skill, derived using in-sample data for calibrating the debiasing parameters, shows promise of further improvement with large samples.


2010 ◽  
Vol 49 (11) ◽  
pp. 2230-2245 ◽  
Author(s):  
Sara A. Michelson ◽  
Irina V. Djalalova ◽  
Jian-Wen Bao

Abstract A season-long set of 5-day simulations between 1200 UTC 1 June and 1200 UTC 30 September 2000 are evaluated using the observations taken during the Central California Ozone Study (CCOS) 2000 experiment. The simulations are carried out using the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5), which is widely used for air-quality simulations and control planning. The evaluation results strongly indicate that the model-simulated low-level winds in California’s Central Valley are biased in speed and direction: the simulated winds tend to have a stronger northwesterly component than observed. This bias is related to the difference in the observed and simulated large-scale, upper-level flows. The model simulations also show a bias in the height of the daytime atmospheric boundary layer (ABL), particularly in the northern and southern Central Valley. There is evidence to suggest that this bias in the daytime ABL height is not only associated with the large-scale, upper-level bias but also linked to apparent differences in the surface forcing.


2006 ◽  
Vol 134 (7) ◽  
pp. 1987-2008 ◽  
Author(s):  
Sytske K. Kimball ◽  
F. Carroll Dougherty

Abstract In the course of studying the development of hurricanes using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5), a relationship between storm intensity and the distribution of vertical levels became apparent, even when the same total number of sigma levels was used. A specific case of an idealized hurricane, on an f plane, in a quiescent environment, with constant and uniform SST of 28°C, was used to study the sensitivity of hurricane structure and evolution to the distribution of sigma levels. The distribution of vertical levels in the inflow, outflow, and middle layers of the atmosphere clearly affects the intensity, size, and structure of the storms, causing certain physical processes to be under- or overresolved. A well-resolved outflow layer is found to be necessary for proper storm intensification, while a well-resolved inflow layer does not necessarily correspond to an intense storm. In fact, when a well-resolved inflow layer is coupled with a poorly resolved outflow layer, a particularly weak storm evolves. When too few levels are assigned to the upper layer, the storm’s outflow is restricted, causing the eyewall column to become statically stable until surface fluxes can replenish low-level equivalent potential temperature content. Convection in the eyewall and compensating subsidence in the eye occur at a moderate rate and weak storms evolve. However, too few levels in the planetary boundary layer (PBL) can cause a storm to overintensify because of overestimated surface fluxes. When such a PBL is coupled with a poorly resolved outflow, the excessive surface fluxes can compensate for the stifled secondary circulation. Hence, this storm may develop to an expected intensity, but for the wrong reasons. Better guidelines for vertical-level distribution in numerical models, perhaps developed from observations of real-case hurricanes, are required.


2006 ◽  
Vol 23 (3) ◽  
pp. 424-436 ◽  
Author(s):  
Frank H. Ruggiero ◽  
John Michalakes ◽  
Thomas Nehrkorn ◽  
George D. Modica ◽  
Xiaolei Zou

Abstract Updated versions of the Tangent Linear Model (TLM) and adjoint of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) have been developed and are now available to the meteorological community. The previous version of the MM5 TLM and adjoint were designed for single-processor computer architectures, based on version 1 of MM5, and were hand coded, which made it difficult to maintain up-to-date versions of the TLM and the adjoint as MM5 evolved. The new TLM and adjoint are based on version 3 of MM5 and run efficiently on multiple-processor computers. The TLM and adjoint were developed with the aid of the Tangent Linear and Adjoint Model Compiler (TAMC) automatic code generator. While some manual intervention is still necessary, the use of the automatic code generator can significantly speed code development and lower code maintenance costs. The new TLM and adjoint contain most of the physics packages and observation operators that were available in the MM5 version 1 TLM and adjoint. The new adjoint has been combined with the MM5 version 3 nonlinear model and an updated minimization module in a four-dimensional variational data assimilation analysis configuration. Accuracy of the new TLM and adjoint has been verified by individual unit and system tests as well as comparisons with the adjoint from MM5 version 1. Timing tests showed substantial decreases in time to solution when increasing the number of processors devoted to the problem.


2007 ◽  
Vol 46 (5) ◽  
pp. 605-614 ◽  
Author(s):  
Miriam L. Rorig ◽  
Steven J. McKay ◽  
Sue A. Ferguson ◽  
Paul Werth

Abstract Dry thunderstorms (those that occur without significant rainfall at the ground) are common in the interior western United States. Moisture drawn into the area from the Gulfs of Mexico and California is sufficient to form high-based thunderstorms. Rain often evaporates before reaching the ground, and cloud-to-ground lightning generated by these storms strikes dry fuels. Fire weather forecasters at the National Weather Service and the National Interagency Coordination Center try to anticipate days with widespread dry thunderstorms because they result in multiple fire ignitions, often in remote areas. The probability of the occurrence of dry thunderstorms that produce fire-igniting lightning strikes was found to be greater on days with high instability and a deficit of moisture at low levels of the atmosphere. Based on these upper-air variables, an algorithm was developed to estimate the potential of dry lightning (lightning that strikes the ground with little or no rainfall at the surface) when convective storms are expected. In the current study, this algorithm has been applied throughout the western United States, with modeled meteorological variables rather than the observed soundings that have previously been used, to develop a predictive scheme for estimating the risk of dry thunderstorms. Predictions of the risk of dry thunderstorms were generated from real-time forecasts using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) for the summers of 2004 and 2005. During that period, 240 large lightning-caused fires were ignited in the model domain. Of those fires, 40% occurred where the probability of dry lightning was predicted to be equal to or greater than 90% and 58% occurred where the probability was 75% or greater.


Sign in / Sign up

Export Citation Format

Share Document