Table Of Content

USAFETAC/PR-91 /002 Lfl MINOT AFB3 WIND STUDY N by Major Walter F. Miller JANUARY1991' APPROVED FOR PUBLIC RELEASE; DISTRIBUJTION IS UNLIMITED USAF ENVIRONMENTAL TECHNICAL APPLICATIONS CENTER Scott Air Force Base, Illinois, 62225-5438 REVIEW AND APPROVAL STATEMENT USAFETAC/PR-91/02, Minot AFB Wind Smdv, January 1991, has been reviewed and is approved for public release. There is no objection to unlimited distribution of this document to the public at large, or by the Defense Technical Infonnation Center (DTIC) to the National Technical Infonnation Service (NTIS). PATR ICI ,REITLING Chief Scientist FOR THE COMMANDER WALTFR S.B URGI(IANN Scientific and rechnical Information Program Manager 17 .Ianuary 1991 E * REPORT DOCUMENTATION PAGE 2. Report Date: January 1901 3. Report Type: Project Report 4. Title: Minot AFB Wind Study 6. Author: Maj Walter F. Miller 7. Performing Organization Name and Address: USAF Environmental Technical Applications Center (USAFETAC/DNO), Scott AFB, IL 62225-5438 8. Performing Organization Report Number: USAFETAC/PR-91/002 12. Distribution/Availability Statement: Approved for public release; distribution is unlimited. 13. Abstract: This report documents the development and evaluation of II algorithms for forecasting gusty northwest winds at Minot AFB, ND. Six of the algorithms used pressure gradients between Dickinson, ND, and Portage la Prairie, Canada, and between Glasgow, MT, and Yorkton. Canada. The other five used the 850 or 700 nib wind over Glasgow, MT. One of the algorithms was found to have skill in forecasting gusty winds and was recommended for operational use. 14. Subject Terms: FORECASTING, WEATHER FORECASTING, WIND, GUSTS, ALGORITHMS, MINOT AFB, NORTH DAKOTA. 15. Number of Pages: 52 17. Security Classification of Report: Unclassified 18. Security Classification of this Page: Unclassified 19. Security Classification of Abstract: Unclassified 20. Limitation of Abstract: UL Standard Form 298 ii PREFACE This study describes the results of USAFETAC Project 9N)219, "Minot Wind Study." The analyst was Major Walter F. Miller. The original request (from Det 21,9 WS, Minot AFB, ND-- amended by 3 WW/DOO), tasked USAFETAC to develop a correlation wind study to help forecast gusty northwest winds at Minot AFB between October and March. Det 21 wanted to relate pressure gradients between Dickinson, ND, and Portage a Prairie, Canada, and between Glasgow, MT, and Yorkton, Canada, to Minot surface winds (including gusts) for a 6-hour forecast period. In this study, "pressure gradient" is used loosely to refer to the difference in Pressure between two staions. This deI ..A i'; possible here because the distance between the two points remains constant. Det 21 also asked USAFETAC to examine a relationship between the 70K)-mb wind speed reported at Glasgow, MT, and the maximum surface wind speed at Minot for a 12-hour forecast period. Eleven candidate methods were developed and evaluated. One was found to have skill in forecasting gusty winds for Minot AFB from October through March; it isi dentified and described in this study. Aooession For NTIS GRA&I DTIC TAB Unannounced 1] Justificatio By Distribution/ Availability Codes _ -kjail and/or Dist special 3 0 iv CONTENTS Page I. INT R O D U C T IO N ............................................................................................................ I 1.1 W hy the Study? ............................................................................................................... I 1.2 Candidate A lgorithm s ...................................................................................................... ! 1.3 The Basic Statistical Relationship ................................................................................... 1 1.4 "Inflated" W ind Speed ................................................................................................. 2 1.5 Study O rganization ..................................................................................................... 2 2. T H EO R Y ................................................................................................................................... 3 2.1 The G eostropic W ind Equation ................................................................................... 3 2.2 G eostrophic W ind Lim itations ................................................................................... 3 2.3 Us ing U pper-A ir D ata ................................................................................................ 3 3. D AT A ........................................................................................................................................ 4 3.1 Surface O bservations ................................................................................................. 4 3.2 U pper-Level W inds ................................................................................................... 4 4. M ET H O D O LO G Y .......................................................................................................... 5 4.1 Q uality Control ...................................................................................................... 5 4.2 D atabase Preparation ................................................................................................. 6 4.3 The Pearson Product-M ornent Correlation Coefficient ............................................. 7 4.4 FCST M A X Correlation .............................................................................................. 8 4.5 M odel Developm ent from Surface D ata ....................................................................... !1 4.6 Candidate M odel Evaluation ..................................................................................... 12 4.6. I The Coefficient of Determ ination ......................................................................... 12 4.6.2 The Heidke Skill Score .......................................................................................... 12 4.6.3 The Probability of Detection ................................................................................ 12 4.6.4 The False A larm Rate ............................................................................................ 12 4.6.5 The C ritical Success Index ......................................................................................... 2 4.7 Frequency D istribution ............................................................................................ 12 4.8 The N W S "Inflation Technique". ............................................................................. 12 4.9 "Inflation" M ethdology A pplied to U pper-A ir D ata ................................................. 13 4. 10 Correlation Coefficients for Up per-A ir Variables ................................................. 13 4.11 Candidate M odels U sing U pper-A ir D ata .............................................................. 14 5. A N A LYSIS, R ESU LT S, A N D V ER IFIC AT IO N ............................................................ 18 5. 1 Initial Evaluation ........................................................................................................... 18 5.2 W ind Speed Frequency D istributions ...................................................................... 20 5.3 Reducing the Pressure G radient .............................................................................. 2 I 5.4 The Advantages of a New Threshold ........................................................................ 23 5.5 Inflation A pplied ....................................................................................................... 24 5.6 Independent Verification .......................................... 26 V 5.7 Sum m ary of Results ......................................3. 1............. 5.8 Leid Times .................................................................................. 31 5.9 Refinement: Adding 7(X)-nib Winds ....................................................... 31 5. 10 Independent Verification--Upper-Air ..................................................... 33 6. DISCUSSION ...................................................................................... 34 6. 1 Limitations .................. I....I............................................................. 34 6.2 Persistence................................................................................... 34 6.3 MOS Forecasting ............................................................................. 34 6.4 Geostrophic Wind Forecasting ............................................................... 34 6.5 The Recommended Model............................................................... .. 95 7. CONCLUSIONS................................................................................... 36 7. 1 The Study Summarized ...................................................................... 36 7.2 Inflation Recommended...................................................................... 36 7.3 Upper-Air Data Not Recommended ......................................... .............. 36 7.4 Model "E" Recoimended..................................................... .............. 36 &.REFERENCES.................................................................................... 37 9. SPECIALIZED TERMS AND ACRINABs..................................................... 38 vi0 FIGURES Page Figure 1. The Four Stations Used for Pressure Gradient Comparisons .......................................... Figure 2. Scatter Diagram of FCSTMAX as a Function of MAXWND--Both Pressure G radients G reater than 4 m b ........................................................................ 8 Figure 3. Scatter Diagram of FCSTMAX as a Function of DIKMYPG--Both Pressure G radients G reater than 4 m b ........................................................................ 9 Figure 4. Scatter Diagram of FCSTMAX as a Function of GGWMYQV--Both Pressure G radients G reater than 4 m b ........................................................................ 9 Figure 5. Scatter Diagram of FCSTMAX as a Function of SLP--Both Pressure Gradients G reater than 4 m h ..................................................................................................... 10 Figure 6. Scatter Diagram of FCSTMAX as a Function of DELP24--Both Pressure G radients G reater than 4 nib ..................................................................................... 10 Figure 7. Scatter Diagram of FCSTMAX (12 Hoursg) as a Function of WND850-- WND850 Greater than 30 Knots, Wind Direction Between 240 and 3600. ................. 15 Figure 8. Scatter Diagram of FCSTMAXD (1 2 Hours) as a Function of WND7(X)-- WND850 Greater than 30 Knots, Wind Direction Between 240 and 3600 .................. 16 Figure 9. Scatter Diagram of FCSTMAX 12 Hours) as a Function of STABLE--WND850 Greater Than 30 Knots, Wind Direction Between 240 and 3600. ............................ 16 Figure 10. Scatter Diagram of FCSTMAX (12 1t ours) as a Function of STRENGTH-- WND850 Greater Than 30 Knots, Wind Direction Between 240 and 3600 ................ 17 vii TABLES Page Table I. Upper-Air Variables Used in the Regression Study ................................................... 6 Table 2. Pearson Product-Moment Correlation Between the Dependent and Independent Variables When Both Pressure Gradients Are Greater than 4 mb ...... 7 Table 3. Candidate M odels Using Surface Observations ............................................................. I I Table 4. Pearson Product-Moment Correlation Coefficient for Upper-Air Variables-- 850 nib Wind Speed Greater than 30 kts, Wind Direction Between 240 and 360. ...... 13 Table 5. Candidate Models Using Upper-Air Data--Wind Speed Greater than 30 kts, W ind D irection Between 240 and 3600 .................................................................... 15 Table 6. Verification Matrix for Candidate Model A--Both Pressure Gradients Greater than 4 m b (D ependent D ata Set) .......................................................................................... N Table 7. Verification Matrix for Candidate Model E--Both Pressure Gradients Greater than 4 nib (D ependent D ata Set) ..................................................................................... 19 Table 8. Verification Statistics for Candidate Models Using Surface Observations-- Both Pressure Gradients Greater than 4 nib (Dependent Data Set) .......................... 19 Table 9. Wind Speed Frequency Distribution-- Both Pressure Gradients Greater than 4 nib (D ependent D ata Set) .............................................................................................. 20 Table 10. Wind Speed Frequency Distribution--At Least One Pressure Gradient Less than 4 trb, Direction Between 270 and 360', Speed Greater than 10 kts (Dependent D a ta Se t) ........................................................................................................................ 20 Table I I. Wind Direction Frequency Distribution--Both Pressure Gradienis Greater than 4 nib (D ependent D ata Set) ..................................................................................... 2 I Table 12. Verification Matrix for Candidate Model A--Both Pressure Gradients Greater tha I nib (D ependent D ata Set) .............................................................................. 21 Table 13. Verification Matrix for Candidate Model E--Both Pressure Gradients Greater than I nib (D ependent D ata Set) .............................................................................. 22 Table 14. Verification Statistics for Candidate Models Using Surface Observations-- Both Pressure Gradients Greater than i mb (Dependent Data Set) ........................ 22 viii Table 15. Wind Speed Frequency Distribution--Both Pressure Gradients Greater than 1 m b (D ependent D 'tta Set) ........................... ........................................................ 23 Table 16. Wind Speed Frequency Distribution--At Least One Pressure Gradient Less than I tb,D irection Between 270 and 3600, Speed Greater than 10 kts (Dependent D at a Se t . ........................................................................................................................ 23 Table 17. Wind Direction Frequency Distribution--Both Pressure Gradients Greater than I m b (D ependent D ata Set) ..................................................................................... 24 Table 18. Verification Matrix for Candidate Model A--Both Pressure Gradients Greater than I mb, Inflation Used (Dependent Data Set) ................................................... 25 Table 19. Verification Matrix for Candidate Model E--Both Pressure Gradients Greater than I tob, Inflation Used (Dependent Data Set) ................................................... 25 Table 20. Verification Statistics for Candidate Models Using Surface Observations-- Both Pressure Gradients Greater than I tub, Inflation Used (Dependent Data Set) ..2.6. Table 2 1.V erification Matrix for Candidate Model A--Both Pressure Gradients Greater than I tub (Independent D ata Set) ............................................................................ 27 Table 22. Verification Matrix for Candidate Model P--Both Pressure Gradients Greater !ha- n'h (Indepenm-_e t D ata Sct) ....................................................................... 27 Table 23. Verification Statistics for Candidate Models Using Surface Observations-- Both Pressure Gradients Greater than I mb (Independent Data Set) ....................... 28 Table 24. Wind Speed Frequency Diso ibution- Both Pressure Gr: dicna (i'-, thMI I m b (Independent D ata Set) ................................................................................... 28 1 able 25. Wind Speed Frequency Distribution--At Least One Pressure Gradient Less than I mb, Direction Between 270 and 360', Speed Greater than 10 kts. (Independent D at a Se t ). ..................................................2.8 .................................................................... Table 26. Wind Direction Frequency Distribution--Both Pressure Gradients Greater than I m b (Independent D ata Set) ................................................................................... 29 Table 27. The Verification Matrix for Candidate Model A--Both Pressure Gradients Greater than I tub, Inflation Used (Independent Data Set) ................................................... 29 Table 28. Verification Matrix for Candidate Model E--Both Pressure Gradients Greater than I nib. Inflatiotn U sed (lndccift .it Data Sct) ........................3.0............................... ix Table 29. Verification Statistics for Candidate Models Using Surface Observations-- Both Pressure Gradients Greater than I nib, Inflation Used (independent Data Se t ). ........... .................................................................................................................... 30 Fable 30. Verification Statistics for Candidate Models Using Upper-Air Observations-- W inds Greater than 30 kis (Dependent Data Set) ..................................................... 32 Table 31. Verification Statistics for Candidate Models Using Upper-Air Observations-- Winds Greater than 30 kts, Inflation Used (Dependent Data Set) .......................... 32 Table 32. Verification Statistics for Candidate Models Using Upper-Air Observations-- W inds Greater than 30 kts (Independent Data Set) ................................................. 33 Table 33. Verification Statistics for Candidate Models Using Upper-Air Observations-- Winds Greater than 30 kts, Inflation Used (lndependent Data Set) ........................ 33 x0 0