All Things TAF A Briefing to Central Region

All Things TAF A Briefing to Central Region September 24, 2009 Chuck Kluepfel National Weather Service Headquarters Silver Spring, Maryland 301 -713 -0090 x 132 Charles. Kluepfel@noaa. gov Prepared October 2008 1

TAF System Objectives • • • Verify all terminals with a TAF More continuous – Every 5 minutes Scheduled TAFs/ Amendments / Both Evaluate TEMPO / PROB groups Combine CIG and VIS into one element – Called: Flight Category – Example: < 1000 feet or < 3 miles • Compare TAF to MOS or persistence 2

Part 1 TEMPO Forecasts Are we serving the people who use our forecasts? 3

Is a TEMPO Group Justified? Variability Test Look at every 5 minutes of the TEMPO. Look for two or more changes +/- 90 minutes. Definition of “Change” • • Sig Wx Types: Beginning or end of an event Ceiling / Visibility: Defined categorically 4

Ceiling and Visibility Categories All Ceiling or visibility changes are defined categorically • • • Ceiling > 3000 feet 2000 -3000 feet 1000 -1900 feet 500 -900 feet 200 -400 feet < 200 feet 5 • • • Visibility > 5 statute miles 3 to 5 miles 2 to < 3 miles 1 to < 2 miles ½ to < 1 mile < ½ mile

KABC 301728 z 301818 24010 KT P 6 SM SCT 030 TEMPO 1820 1 SM +TSRA BKN 008 CB FM 2100 VRB 05 KT P 6 SM SKC= Question: If it only thunders for 15 minutes out of 2 hours (the length of my TEMPO group), will my False Alarm Ratio be 1 hour 45 minutes 2 hours 6 = 0. 88 ?

KABC 301728 z 301818 24010 KT P 6 SM SCT 030 TEMPO 1820 1 SM +TSRA BKN 008 CB FM 2100 VRB 05 KT P 6 SM SKC= Answer: Absolutely not! We do not compute False Alarm Ratios (FAR) from the TEMPO forecast, alone. TEMPOs are not intended to be applied to the entire TEMPO period. We only compute FARs for the § Prevailing Forecast § “Operational Impact” Forecast 7

TAF Example 1 Obs’n 1635. . 1805. . 1905 1910 1915 1920 1925 1930. . 1955 2000 Prevail TEMPO Pass VAR Test? OIF No TS TS Yes No TS TS No TS No TS TS TS TS Yes Yes Yes No TS TS No TS No TS TS TS Yes No TS 8

Thunderstorm TEMPOs Central Region All Scheduled TAFs 0 -6 hour forecasts TEMPO FORECASTS (evaluated every 5 min. ) FY 09* (FY 06) a. TEMPO Forecast (Hours) 14, 756 (23, 593) b. Justified TEMPO (Hours) 2047 (6477) c. Justified TEMPO (% of line a) 14% (27%) d. Justified TEMPO – Hit (% of line b) 73% (38%) 12, 709 (17, 115) e. Unjustified TEMPO (Hours) f. Unjustified TEMPO – Should be FM (% of line e) g. Unjustified TEMPO – Benign (% of line e) h. Unjustified TEMPO – Hurt (% of line e) * FY 09 through September 21, 9 2009 10% (5%) 0% (0%) 87% (92%)

10

Fog (BR and FG) TEMPOs Central Region All Scheduled TAFs 0 -6 hour forecasts TEMPO FORECASTS (evaluated every 5 min. ) FY 09* (FY 06) a. TEMPO Forecast (Hours) 63, 685 (87, 230) b. Justified TEMPO (Hours) 7851 (9747) c. Justified TEMPO (% of line a) 12% (11%) d. Justified TEMPO – Hit (% of line b) 61% (60%) e. Unjustified TEMPO (Hours) 56, 023 f. Unjustified TEMPO – Should be FM (% of line e) 48% (45%) 5% (4%) 32% (38%) g. Unjustified TEMPO – Benign (% of line e) h. Unjustified TEMPO – Hurt (% of line e) * FY 09 through September 21, 2009 11 (77, 572)

Ceiling TEMPOs Central Region All Scheduled TAFs 0 -6 hour forecasts TEMPO FORECASTS (evaluated every 5 min. ) FY 09 * (FY 06) a. TEMPO Forecast (Hours) 129, 799 (181 K) b. Justified TEMPO (Hours) 54, 947 (65 K) c. Justified TEMPO (% of line a) 42% (36%) d. Justified TEMPO – Hit (% of line b) 29% (29%) e. Justified TEMPO – Improved TAF 12% (13%) 74, 852 (117 K) 44% (48%) 5% (4%) f. Unjustified TEMPO (Hours) g. Unjustified TEMPO – Should be FM (% of line f) h. Unjustified TEMPO – Benign (% of line f) i. Unjustified TEMPO – Hurt (% of line f) * FY 09 through September 21, 2009 12 43% (41%)

Visibility TEMPOs Central Region Alll Scheduled TAFs 0 -6 hour forecasts TEMPO FORECASTS (evaluated every 5 min. ) FY 09 * (FY 06) a. TEMPO Forecast (Hours) 104, 383 (136 K) b. Justified TEMPO (Hours) 49, 990 (55 K) c. Justified TEMPO (% of line a) 48% (41%) d. Justified TEMPO – Hit (% of line b) 25% (26%) e. Justified TEMPO – Improved TAF 16% (17%) f. Unjustified TEMPO (Hours) 54, 393 (81 K) g. Unjustified TEMPO – Should be FM (% of line f) 18% (16%) 1% (1%) 77% (78%) h. Unjustified TEMPO – Benign (% of line f) i. Unjustified TEMPO – Hurt (% of line f) * FY 09 through September 13 21, 2009

Part 2 Traditional Statistics 14

Review • POD is an event-based statistic • For TAFs, IFR and below usually define the event. • It is the percentage of total “events” that hit. • It is alternatively called the hit rate. • FAR is a warning-based statistic • Any forecast of IFR or below defines the “warning. ” • FAR is the percentage of time that the forecaster false alarms for IFR or below. 15

Review • You can drive your POD up by over-forecasting IFR and below conditions. • However, this practice will also drive up your FAR. • The CSI provides a mathematical way of correcting the POD for a high FAR. (It adds the number of false alarms to the denominator). 16

Operational Impact Forecast (OIF) • When there is no TEMPO forecast: • OIF is set equal to the prevailing forecast. • When TEMPO passes variability test: • Two forecasts are in effect simultaneously. • OIF is set equal to forecast with the least error. • When TEMPO fails variability test: • Two forecasts are in effect simultaneously. • OIF is set equal to the forecast with the worst conditions. • This rule usually, but not always, lowers the OIF. 17

Central Region FY 09* Flight Category: IFR and Below * FY 09 through September 21, 2009 18

Central Region FY 09* Flight Category: IFR and Below * FY 09 through September 21, 2009 19

Central Region FY 09* Flight Category: IFR and Below * FY 09 through September 21, 2009 20

Central Region FY 09* Flight Category: IFR and Below * FY 09 through September 21, 2009 21

Central Region FY 09* Flight Category: IFR and Below * FY 09 through September 21, 2009 22

Central Region Flight Category: IFR and Below Warm Season FY 09* (April 1 to September 21, 2009) * FY 09 through September 21, 2009 23

Part 3 Lead-Time Software 24

25

Part 4 The Future 26

The Future • Output to CSV files o Starting with Flight Category and Sig Wx Data o Ceiling / Visibility (next) o Winds (last) • • Plots of POD / FAR / CSI Sort Elements by Sig Wx Type 27

Improving the Current System: Geometric Interpretation 1 FOH (Z) (X) FAR FOM (Ma, Mf) FORECAST (F) PON Ma: Average of Observations (x+y)/N POD POFD 0 (W) DFR 0 (Y) FOCN OBSERVED (A) 1 A(F): Regression of Observations upon the forecast F(A): Regression of Forecast upon observations Mf: Average of Forecasts (x+z)/N

Basic Interpretation: Extreme Cases PERFECT FORECAST RESIGN FROM THE NWS Observed Forecast Observed Yes No Total Yes 6 0 6 No 0 6 Total 6 6 Yes No Total Yes 0 6 6 6 No 6 0 6 12 Total 6 6 12 29 Forecast

Basic Interpretation: Random Chance RANDOM CHANCE Observed Forecast Yes No Total Yes 3 3 6 No 3 3 6 Total 6 6 12 30

Basic Interpretation: False Alarms vs. Misses Over-forecast Under-forecast Observed Forecast Yes No Total Yes 50 50 100 150 No 0 100 200 Total 50 150 200 Yes No Total Yes 50 0 50 No 50 100 Total 100 10 Forecast Assesses Bias in one glance!! 31

The Future • Output to CSV files o Starting with Flight Category and Sig Wx Data o Ceiling / Visibility (next) o Winds (last) • • Plots of POD / FAR / CSI Sort Elements by Sig Wx Type 32