What I just learned by watching Loco 4....

[Bobbert]

In all the storm reporting, has anyone mentioned sump pumps? They should start doing summer storm damage prevention stories every spring just like they do winter preparation stories every November. I keep seeing story after story about all the storm damage and people wondering how to pay for it but I never see any reporting on how people can prepare for such situations.

[SolidGoldDancers]

Well, Paul Gross is a complete moron then, if that is indeed what he said and meant. I didn't see that part of the newscast. 6 inches of rain is not a "500 year event." Some portions of Metro Detroit saw similar rainfall amounts in August 2014. Areas west of Big Rapids saw 7 inches of rain in 24 hours in early October 2019. I guess he didn't want to let an opportunity to exaggerate / over-sensationalize go to waste.

Sorry to hear about your car, Rate This. That sucks.

I felt bad for Andrew Humphrey last night. WXYZ and Fox 2 both had team coverage (two meteorologists) during the Livingston County tornado warning last night. Andrew - who had been working since the crack of dawn - was left to do coverage all by himself, including running all the graphics, from the look of things.

A 500 year event doesn't mean it happens once every 500 years, nor N times in 500 years, it means its Q (the quantity of whatever they are measuring) fits in a Bayes' curve at the point of 500 years. They use this as opposed to percentiles/quartiles because the time scope is implicit.

Paul Gross is one of the few true meteorologists in the Detroit market. His chops go back to working with Mal Sillars, graduate of UofM and all the calculus based physics and thermodynamics that go with it. He's not a moron.

[MWmetalhead]

A 500 year event doesn't mean it happens once every 500 years, nor N times in 500 years, it means its Q (the quantity of whatever they are measuring) fits in a Bayes' curve at the point of 500 years. They use this as opposed to percentiles/quartiles because the time scope is implicit.

The average viewer interpreted his statement as meaning that type of rainfall event, on average, occurs 1 time every 500 years for a given location in Metro Detroit.

I definitely prefer Paul's work to that of his counterparts on WDIV in the vast majority of cases. I overreacted earlier. However, on this particular matter, I do find his claims to be dubious.

[Stebo48021]

I agree with Paul Gross's assertion that it was a 500 year event, the problem is that we are having these events more frequently to the point that it doesn't seem like a rare occurrence, but it is. To put it into perspective too, if we had that 6.8" at DTW for example, it would have crushed our daily record for any calendar day by over 2", and those records go back to the 1870s, which is already 150 years.

[MWmetalhead]

The problem is - those records only measure discrete daily events, correct?

It is possible to have 6+ inches of rain within a 24 hour time block, part of which falls on one calendar day and the remainder of which falls on the immediately subsequent calendar day.

[jadednihilist]

The problem is - those records only measure discrete daily events, correct?

It is possible to have 6+ inches of rain within a 24 hour time block, part of which falls on one calendar day and the remainder of which falls on the immediately subsequent calendar day.

The charts I provided do quite a bit of work to treat each rainfall event as a discrete object, rather than rely exclusively on total daily rainfall detailed here. Table 4.4.2 in page 14 of the document shows that they used 1652 stations with precip data at 15 minute (or variable) time intervals, 2657 stations recording precip at hourly intervals, and 11,918 stations reporting at 1-day intervals.

I have only skimmed this document, but generally these products tend to estimate uncertainty, in this case expressed as a confidence interval, as a function of measurement error and artifacts of the numerical methods used to integrate all of the datasets into one they can analyze.

To your specific question, you can make a rough upper threshold estimates of consecutive 24-hour precip by plotting at two-day consecutive rainfall records. Keep in mind you'd be aliasing multiple events outside that 24-hour window, but even then I would imagine reaching 6" in that time period would be pretty rare. Here are monthly precip extremes in Detroit (source):

Code: Select all

Month		Max/Year	Average
January		5.02"/1932	1.96"
February	6.41"/1881	2.08"
March		5.63"/1913	2.43"
April		6.89"/1947	3.26"
May		8.46"/2004	3.72"
June		8.31"/1982	3.26"
July		8.76"/1878	3.51"
August		8.33"/1926	3.26"
September	7.52"/1986	3.22"
October 	7.80"/1954	2.53"
November	5.68"/1982	2.57"
December	5.19"/1967	2.25"

As Stebo has pointed out, no singular day in Detroit's official climate records exceeds 6" dating back to 1874. Any month in which we record 6" or more pushes us near record territory. We just don't typically see that kind of rain around here, with the exception of nearly 3 events in the past decade. 6" in 24 hours at minimum represents 160 to 300% of any given total monthly average precipitation.

[MWmetalhead]

I appreciate your diligent insight, as always.