Heat and Health: MethodologicalConsiderations for Warning SystemDevelopment

Dr. Laurence S. KalksteinSynoptic Climatology Laboratory

Center for Climatic ResearchUniversity of Delaware

Washington, D.C.May 2005

How Are These Systems Unique?

A custom-made system is developed for each urban area, based on specific meteorology for each locale, as well as urban structure and demographics

These systems are based on actual weather-health relationships, as determined by daily variations in human mortality

These systems are based on much more than just temperature and humidity

Systems We are Presently OperatingUSA Chicago, IL Dallas/Ft. Worth, TX* Dayton/Cincinnati, OH Jackson/Meridian, MS Lake Charles/Alexandria, LA Little Rock/Pine Bluff, AR Memphis, TN/Tupelo, MS New Orleans/Baton Rouge, LA Philadelphia, PA* Phoenix, AZ Portland, OR Seattle, WA St. Louis, MO Shreveport/Monroe, LA Washington, D.C. Yuma, AZ

International Bologna, Italy Florence, Italy Milan, Italy Naples, Italy Palermo, Italy Rome, Italy Shanghai, China Turin, Italy Toronto, Canada

* New five-day forecast systems based on point forecast matrices

Maximum Temperature vs. Daily Mortality: New York and Jacksonville, Florida

New York Jacksonville

Steps in System Development Step 1: Determine air masses daily over a city using

newly-developed Spatial Synoptic Classification (SSC)

Step 2: Are any of these air masses ‘offensive’?

Characteristics:

* Statistically significant higher mortality

* Greater within air mass standard deviation in mortality

Step 3: What aspects of the offensive air mass make it most detrimental to human health?

SSC Air Mass Types• DP Dry Polar (cP)• DM Dry Moderate (Pacific)• DT Dry Tropical (cT)• MP Moist Polar (mP)• MM Moist Moderate (Overrunning)• MT Moist Tropical (mT)• MT+ Moist Tropical Plus• TR Transition between air masses

----------------

Sheridan, S.C., 2002: The redevelopment of a weather-type classification scheme for North America.

Int. J. Climatology, 22, 51-68.

Kalkstein, L.S., C.D. Barthel, J.S. Greene and M.C. Nichols, 1996. A New Spatial Synoptic Classification:

Application to Air Mass Analysis. Int. J. Climatology, 16, 983-1004.

Characteristics of WashingtonAir Mass Types

Air mass Air Temp. Dew Pt. Air Temp. Cloud Seasonal

Type 5 pm 5 pm 5 am Cover Freq.*

DP 80 55 64 5 8%DM 87 59 69 5 23%DT 95 64 76 4 5%MP 68 64 67 10 2%MM 79 68 72 9 20%MT 88 69 74 7 27%MT+ 93 73 78 6 6%TR 83 62 72 6 9%

* Mean air mass frequency from 1 June through 31 August

Removing Non-Climatological Noise from the Mortality Data

Phoenix’s Rising Mortality

0

10

20

30

40

50

60

70

75 80 85 90 95

Year

Me

an

Da

ily

Mo

rta

lity

Mean Deaths in Rome by Daystandardization is important

Mean Mortality Increases Within Offensive Air Mass Types

a MT+ does not occur in Seattle or Portland; the moist air mass that is oppressive is MT.* DT+ air mass for Phoenix

Location (Freq) DT MT+

+4.0 (10%)+4.2 (11%)Toronto (7%)

+42.4 (16%)NoneShanghai (11%)

+5.0 (12%)+6.2 (14%)Rome (11%)

None+2.7* (7%)Phoenix (1%)

+3.7 (9%)NoneNew Orleans (2%)

+1.7 (7%)+0.9 (4%)Washington (11%)

+1.1a (4%)+3.0 (9%)Portland (6%)

+4.7a (10%)+3.7 (8%)Seattle (6%)

Development of Forecast Algorithms: Within-Air Mass Category

Mortality Variations

ROME

-8

-6

-4

-2

0

2

4

6

8

10

5/155/31

6/16/15

6/166/30

7/17/15

7/167/31

8/18/15

8/168/31

9/19/15

9/169/30

Me

an a

no

mal

ou

s m

ort

alit

yMT+ mortality decreases through summer

ROME

DT mortality vs. morning temperature

-30-20-10

01020304050

16 18 20 22 24 26 28

5 AM temperature (°C)

Me

an a

nom

alo

us

mor

talit

y

Impact of Consecutive Days of Offensive Air Masses: Philadelphia

0

5

10

15

20

25

30

35

1 2 3 4 5Day in Row

Exc

ess

Dea

ths

DT MT+

Steps That can be Taken When Alerts or Emergencies are Called

1. NWS alerts appropriate city agency (e.g., Department of Health, Office of Aging, etc.)

2. City develops (or already has in place) intervention plans

3. Effectiveness of system can be monitored in three manners,and improvements are implemented if there are shortcomings:--forecasting effectiveness (false positives or negatives)--accuracy in estimating health impacts of offensive weather --noting if we are saving lives.