Correlation of weather ‘feel’ (heat index/windchill)

For some health and weather research I'm doing in Minnesota, I was interested in how closely related the weather is at any given day across the state: In a mountainous state like Oregon, I would expect high variance. Less so in Minnesota. It bears out. Looking at Duluth, Fargo/Moorhead, Minneapolis-St.Paul, Albert Lea,St-Cloud, Granite Falls.

Duluth and Fargo, which have slightly higher latitude, have the 'cooler' lines (unsurprisingly) but the degree of alignment is interesting.

I'll repeat for a mountainous state where I might think greater stochasticity would be evident.

ASOS weather data from Iowa MESONET, visualized in Stata v 16.

Below is a map of the sites listed above (I know they need labels 🙂 ). The shaded area is 85% of the state. For my purposes, I just use the MSP site for exposure. What I gain from this exercise is that that choice is a fairly safe bet to approximate exposure across the shaded area... which in turn is a fairly good approximation for all state case data where specific spatial information is missing.

created using ArcGIS, station location from Iowa MESONET, zip shape files from ESRI w/ 2017 population estimates.

Heat, Education, Race

Heatwaves are associated with preterm birth. Usually education is seen as protective in health outcomes. Births to women who are black are at higher risk of preterm birth.

But in examining the 3-way interaction, we find that heatwaves of 7 day duration, defined as mean high heat index in degrees C, are associated with higher risk of PTB in births to college educated women who are Black. The effect is exacerbated for heatwaves of higher heat index.

Caveats, small N (cases), rare exposure (heatwaves).

Research done in Minnesota, 1998-2005, and submitted for publication to the IJERPH.

via Imgflip