Virginia Tech
Browse
1/1
2 files

Software and Data for "Long term temporal trends in synoptic-scale weather conditions favoring significant tornado occurrence over the central United States"

dataset
posted on 2022-12-20, 16:14 authored by Mohamed El Khouly, Stephanie ZickStephanie Zick, Marco FerreiraMarco Ferreira

This is software and data to support the manuscript  "Long term temporal trends in synoptic-scale weather conditions favoring significant tornado occurrence over the central United States," which is under review with PLOS One. The software includes all code that is necessary to follow and evaluate the work. Additional public datasets include tornado data from the Storm Prediction Center (http://www.spc.noaa.gov/wcm/#gis), MERRA-2 reanalysis data (https://doi.org/10.5067/VJAFPLI1CSIV), North American Regional Reanalysis data (https://psl.noaa.gov/data/gridded/data.narr.html), Global Wind Oscillation data (https://psl.noaa.gov/map/clim/gwo.data.txt), and the Nino 3.4 index.(https://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/detrend.nino34.ascii.txt).

History

Publisher

University Libraries, Virginia Tech

Corresponding Author Name

Stephanie Zick

Corresponding Author E-mail Address

sezick@vt.edu

Files/Folders in Dataset and Description

This folder contains the R-codes and the data to reproduce the results for the paper, "Long term temporal trends in synoptic-scale weather conditions favoring significant tornado occurrence over the central United States" by Mohamed Elkhouly, Stephanie E. Zick, and Marco A.R. Ferreira. Each folder contains the needed R-code and data to reproduce parts of the paper’s analyses and results: [EOF Analysis] This folder contains the R-code and data to reproduce the EOF modes for the two groups of variables, including Tables 2 and 3 titled "Table 2. Variability explained by EOF modes of air temperature and relative humidity group" and "Table 3. Variability explained by EOF modes of horizontal wind components (U and V wind components) and vertical velocity group." [SVDS all_corrTempRhum.R] to conduct EOF analysis for the temperature and relative humidity group [SVDS all_corrWind.R] to conduct EOF analysis for the horizontal wind components (U and V wind components) and vertical velocity group. [Tropopause_All_corr.R] to conduct EOF analysis for the tropopause data. [Scores.R] to retain the scores of the different EOF scores for the four regions. [Fig 2 plots]This folder contains the R-code and the data to retain the plots in Fig 2 for the loadings plots for the temperature relative humidity variables and the wind group of variables. [Plot Loadings.R] to plot the loadings of the different EOF modes. [Fig 3 STF] This folder contains Python code to reproduce Figure 3 [negative_merra_pv_03020202.py] to calculate and plot potential vorticity for most negative days/times in the dataset corresponding to the STF mode [positive_merra_pv_03020202.py] to calculate and plot potential vorticity for most positive days/times in the dataset corresponding to the STF mode [Fig 4 ACM] This folder contains Python code to reproduce Figure 4 [most_negative_merra_wind_011302021] to plot 250 mb winds and geopotential heights for most negative days/times in the dataset corresponding to the ACM mode [most_positive_merra_wind_011302021] to plot 250 mb winds and geopotential heights for most positive days/times in the dataset corresponding to the ACM mode [Prepare Data] This folder contains the R-codes to down-sample and get the case-control data sets for the different models. You can reproduce Table 1 that summarizes counts of tornadic days per region and intensity from the R codes in the folders: A. Tornadic Non for the LEOF models. B. Strong Weak for the IEOF models. [Raw data] contains the Raw data to select from. [Final Data] has all the final datasets to be used for modeling. [Tornadic Non] for the LEOF models. [Strong Weak] for the IEOF models. [Positive Negative] for the selected times to make the composites of the different EOF model for explaining them. [Table 4] This folder contains the R-code and the data to retain the results in table 4 titled "Table 4. The estimated Logistic EOF (LEOF) regression models" [SSH Tornadic_Non.R] to search for the best model using the EOF modes [LEOF_Tornadic NON.R] to fit the LEOF model for the four different regions [Table 5] This folder contains the R-code and the data to retain the results in table 5 titled "Table 5. The Intensity classification (IEOF) models" [SSH StrongWeak.R] for searching for the best model using the EOF modes. [StrongWeak.R] to fit the best EOF models for the four regions. [Table 6 and Fig 5] This folder contains the R-code and the data to retain the results in : A- Table 6 titled "Table 6. Summary of temporal changes of the noteworthy EOF modes in the LEOF models" B- Figure 5 [ChangeTempRhumR.R] to fit all possible models in the section titled ``Supporting information’’ [ChangeWindR.R] to fit all possible models in the section titled ``Supporting information’’ [Results.R] to retain the results in Table 6. [EOF temporal plots.R] to retain the figures in Figure 5. [Table 7] This folder contains the R-code and the data to retain the results in table 7 titled "Table 7. Summary of the Monte Carlo leave-20%-out validation". There are four folders for the four regions. Each folder contains the following folders and codes (for the corresponding regions - R1, R2, R3, and R4). [IEOF models] [R1_ALL_StrongWeak.R] to retain the data set with all the predictors for the cross validation. [R1_SW_Predictions.R] to run the 20% leave out cross validation. [R1_StrongWeak_models.R] to find the best models for each group of models (CP, INDICES) [R1_ROC_ST_WK.R] to calculate the AUC and plot the ROC curves. [LEOF models] [R1_Tornadic_NON_allpredictors.R] to retain the data set with all the predictors for the cross validation. [R1_Indices.R] to find the best models for each group of models (CP, INDICES) [R1_TornadicNON_models.R] to run the 20% leave out cross validation. [R1_ROC_TOR_NON.R]to calculate the AUC and plot the ROC curves.