Search Results

You are looking at 51 - 60 of 705 items for :

  • Regression analysis x
  • Weather and Forecasting x
  • Refine by Access: All Content x
Clear All
William H. Klein and Hal J. Bloom

, Guay-Hong Chen and Bruce Whistler at theUniversity of Maryland; and Don Gilman, Chet Ropelewski and Bob Taubensee at CAC.REFERENCESCayan, D. R., and J. O. Roads, 1984: Local relationships between United States west coast precipitation and monthly mean cir culation parameters. Mon. Wea. Rev., 112, 1276-1282.Draper, H. R., and H. Smith, 1981: Applied Regression Analysis. 2nd ed., Wiley, 294-352.Englehart, P. J., and A. V. Douglas, 1985: A statistical analysis of precipitation frequency in the

Full access
Thomas A. Jones, Daniel Cecil, and Mark DeMaria

significant skill. If the predictor does not, it is removed from the final regression, keeping with the goal of having a model with as few predictors as possible. The high threshold for predictor retention and a degree of subjectivity are necessary to keep the total number of predictors to a minimum, reducing artificial skill ( Neumann et al. 1977 ). The use of more objective means such as principle component analysis failed to produce superior results. Once the final set of predictors is chosen

Full access
Thomas M. Hamill, Robert P. D'Entremont, and James T. Buntin

surface characteristics: water, land, andsnow/ice. Each regression equation is of the form:Tc-clr = Tshd*R~ + Tskin*R2 + (Tshel- T850)*R3 q- Pwat*R4 + BBGS*R5 + C. (2) Here, rc~clr is the conventionally derived estimate ofthe IR clear-column temperature; Tshct and Tskin arethe SFCTMP-supplied estimate of shelter and skintemperature, respectively; ( Tshe~ - T85o) is a measureof low-level stability, with T850 being the 850-mb temperature derived from a global temperature analysis.This term

Full access
John A. Knaff, Charles R. Sampson, and Mark DeMaria

The development of the STIPS model closely follows the development of the SHIPS model in the Atlantic and east Pacific tropical cyclone basins as described in DeMaria and Kaplan (1999) . As a result, STIPS is a multiple linear regression model. The dependent variables (predictands) are the intensity change from the initial forecast time (DELV) at 12-h intervals of all storms not making landfall. Potential predictors (independent variables), or more precisely parameters that have been documented

Full access
Li-Chuan Gwen Chen and Huug van den Dool

consolidation forecasts. Another major challenge often encountered in optimization procedures is overfitting. Overfitting occurs when the length of the training dataset is too short compared to the number of input models ( Peña and Van den Dool 2008 ). To combat this issue, Van den Dool and Rukhovets (1994) and Peng et al. (2002) used information from all grid points in the region of analysis to estimate regression parameters, thereby effectively increasing the training dataset. Wanders and Wood (2016

Full access
Amanda M. Walker, David W. Titley, Michael E. Mann, Raymond G. Najjar, and Sonya K. Miller

determine the most appropriate statistical measure for the surge in a given county. With four storms, two storm surge variables, four fiscal loss metrics, and nine representations, the total number of cases explored was 4 × 2 × 4 × 9 = 288. The goal of this step was to determine which representation would be the most appropriate for the storm surge variables at subsequent steps in the analysis. Step 2 used Minitab statistical software to explore linear and multiple linear regression models using the

Open access
Mark DeMaria and John Kaplan

and date of the storm. The vertical shearvalues along the track of the storm are estimated using the synoptic analysis at the beginning of the forecastperiod. All other predictors are evaluated at the beginning of the forecast period. The model is tested using a jackknife procedure where the regression coefficients for a particular tropicalcyclone are determined with all of the forecasts for that storm removed from the sample. Operational estimatesof the storm track and initial storm intensity

Full access
David John Gagne II, Amy McGovern, and Ming Xue

precipitation forecasts tended to be underdispersive. Linear regression calibration methods have shown some skill improvements in Hamill and Colucci (1998) , Eckel and Walters (1998) , Krishnamurti et al. (1999) , and Ebert (2001) . Hall et al. (1999) , Koizumi (1999) , and Yuan et al. (2007) applied neural networks to precipitation forecasts and found increases in performance over linear regression. Logistic regression, a transform of a linear regression to fit an S-shaped curve ranging from 0 to 1

Full access
Adrienne Tivy, Bea Alt, Stephen Howell, Katherine Wilson, and John Yackel

Chen (1983) , are used in the analysis ( Table 3 ). 4. Methods An automated regression procedure is developed to generate multiple linear regression (MLR) models for the OWRC and four RND time series using as input the pool of over 1500 potential predictors ( Tables 1  – 3 ) defined in sections 2 and 3 . The LOD time series is derived from the maximum RND value in each year. The strength of the relationship between each predictand and predictor, and the relative redundancy of the predictors, are

Full access
Petra Friederichs and Andreas Hense

(GFS) 12-h forecasts. A canonical correlation analysis (CCA) is used to correct for systematic model forecast errors relative to the reanalysis in the large-scale circulation patterns. As this approach uses reanalysis for the training of the probabilistic model, we denote this method the RAN approach, as in Marzban et al. (2005) , although they use multiple linear regression instead of CCA. Forecast skill is assessed using the censored quantile verification (CQV) score ( FH ) and the Brier score

Full access