A Closer Look at Extreme Precipitation Events in Michigan

The Natural Resources Defense Council (NRDC), working with the Rocky Mountain Climate Organization (RMCO), have released a new report on “Extreme Storms in Michigan.”

According to the NRDC/RMCO:

The report documents an 89 percent increase in the annual frequency of extreme precipitation events – defined as 2 inches or more in a day – across the state over the last 50 years.

In contrast, using the NOAA National Weather Service (NWS) database, there is not a single climate sub-region in the state (among the Detroit, Flint, Grand Rapids, Lansing, Muskegon, Alpena, Houghton Lake, Sault Ste Marie, and Marquette County areas) that has a statistically significant increasing trend in “the annual frequency of extreme precipitation events – defined as 2 inches or more in a day” over the last 50 years.  Not one.

In the Appendix of the NRDC/RMCO report, the authors describe their research methodology and list the “37 analyzed stations” that were the source of their data.  Looking at the NOAA-NWS database for the completeness of the annual historical record at some of these stations reveals substantial problems.

For example, at the “Grand Ledge 1 NW” station the NRDC/RMCO used, there was a lack of a complete annual record for the following years between 1964 and 2013 (i.e., the range of the NRDC/RMCO's study): 1991, 1997, all years from 1999 through 2008 inclusive, 2011, 2012, and 2013.  A total of 55 months with incomplete precipitation data exist over this period, and consequently, it is unclear how one could derive a reliable determination and trend analysis of the number of precipitation events of “2 inches or more in a day” for this station.

The following table shows how many incomplete years are present for a subset of just 12 of the stations the NRDC/RMCO examined in Michigan.

The results are striking.  Among this set of 12 of the 37 stations the NRDC/RMCO analyzed, only two of the 12 stations had complete precipitation datasets.  The remainder had large gaps – the South Haven station fails to have a complete annual precipitation series for the entire period of 1991 through 2013, inclusive.

The NRDC/RMCO state that “missing data for particular days were addressed by converting the values for all reported days of a year to equivalent values for 365.25 days.”  This sounds as though the method for filling in missing data assumes that the days with missing data had the same probability of an extreme rainfall event as the days with data.  That is not necessarily a valid assumption, especially where heavy precipitation events themselves can negatively impact gauge performance.  The best approach is only to deal with stations having complete annual datasets.

From a quick scroll of some other stations the NRDC/RMCO used, the missing data problem appears to be ubiquitous throughout their dataset.  By comparison, the nine NOAA-NWS climate sub-region datasets I analyzed – and for which I found no significant trends – all have effectively complete annual data series.

If this report were merely being distributed into the background noise of climate change, the concerns would be minimal.  But it is being used for policy making in Michigan.  Take this quote from a story at the MLive/Grand Rapids Press website:

'There's more water coming and we need to be prepared for it,' Mayor George Heartwell said during a Thursday, Dec. 4, teleconference about a new report from The Rocky Mountain Climate Organization and Natural Resources Defense Council. 'The changes in climate are accelerating and the need to address them is critical. Creating resilient communities has to be our top priority. I only pray (our awareness) hasn't arrived too late to save us from the most devastating impacts of climate change.'

Here is the last three decades of the “annual frequency of extreme precipitation events” for Grand Rapids.

The “changes in climate are accelerating”?  That dashed orange line is the best-fit linear regression over the past 30 years of “extreme precipitation” in Grand Rapids.  It is a flat line (actually, it is angled slightly downward – not upward).

In the Detroit climate region, there has not been a significant trend in these types of “extreme precipitation” events since records began in 1874.  The correlation is negative, toward fewer events per year over the last 140 years.  Over in the Flint region, no significant trend since records began in 1921.  Same applies in the Muskegon region since records began 1898, and in the Houghton Lake area since records began 1918, and in the Sault Ste. Marie area since records began in 1889.  Records begin in 1864 for the Lansing area, and no significant trend (yet again, a negative correlation toward fewer events, not more).  Same applies in the Alpena sub-region since records began in 1918, and in the Marquette County area since records began in 1872.  No significant trend in the Grand Rapids area since continuous records began in 1903, either.

In other words, none of the individual climate sub-regions in Michigan have significant increasing trends in the number of annual extreme precipitation events since records began, or over the past 50 years.  Hopefully the policymakers in this state will consider these details when reading the NRDC/RMCO report.

The Natural Resources Defense Council (NRDC), working with the Rocky Mountain Climate Organization (RMCO), have released a new report on “Extreme Storms in Michigan.”

According to the NRDC/RMCO:

The report documents an 89 percent increase in the annual frequency of extreme precipitation events – defined as 2 inches or more in a day – across the state over the last 50 years.

In contrast, using the NOAA National Weather Service (NWS) database, there is not a single climate sub-region in the state (among the Detroit, Flint, Grand Rapids, Lansing, Muskegon, Alpena, Houghton Lake, Sault Ste Marie, and Marquette County areas) that has a statistically significant increasing trend in “the annual frequency of extreme precipitation events – defined as 2 inches or more in a day” over the last 50 years.  Not one.

In the Appendix of the NRDC/RMCO report, the authors describe their research methodology and list the “37 analyzed stations” that were the source of their data.  Looking at the NOAA-NWS database for the completeness of the annual historical record at some of these stations reveals substantial problems.

For example, at the “Grand Ledge 1 NW” station the NRDC/RMCO used, there was a lack of a complete annual record for the following years between 1964 and 2013 (i.e., the range of the NRDC/RMCO's study): 1991, 1997, all years from 1999 through 2008 inclusive, 2011, 2012, and 2013.  A total of 55 months with incomplete precipitation data exist over this period, and consequently, it is unclear how one could derive a reliable determination and trend analysis of the number of precipitation events of “2 inches or more in a day” for this station.

The following table shows how many incomplete years are present for a subset of just 12 of the stations the NRDC/RMCO examined in Michigan.

The results are striking.  Among this set of 12 of the 37 stations the NRDC/RMCO analyzed, only two of the 12 stations had complete precipitation datasets.  The remainder had large gaps – the South Haven station fails to have a complete annual precipitation series for the entire period of 1991 through 2013, inclusive.

The NRDC/RMCO state that “missing data for particular days were addressed by converting the values for all reported days of a year to equivalent values for 365.25 days.”  This sounds as though the method for filling in missing data assumes that the days with missing data had the same probability of an extreme rainfall event as the days with data.  That is not necessarily a valid assumption, especially where heavy precipitation events themselves can negatively impact gauge performance.  The best approach is only to deal with stations having complete annual datasets.

From a quick scroll of some other stations the NRDC/RMCO used, the missing data problem appears to be ubiquitous throughout their dataset.  By comparison, the nine NOAA-NWS climate sub-region datasets I analyzed – and for which I found no significant trends – all have effectively complete annual data series.

If this report were merely being distributed into the background noise of climate change, the concerns would be minimal.  But it is being used for policy making in Michigan.  Take this quote from a story at the MLive/Grand Rapids Press website:

'There's more water coming and we need to be prepared for it,' Mayor George Heartwell said during a Thursday, Dec. 4, teleconference about a new report from The Rocky Mountain Climate Organization and Natural Resources Defense Council. 'The changes in climate are accelerating and the need to address them is critical. Creating resilient communities has to be our top priority. I only pray (our awareness) hasn't arrived too late to save us from the most devastating impacts of climate change.'

Here is the last three decades of the “annual frequency of extreme precipitation events” for Grand Rapids.

The “changes in climate are accelerating”?  That dashed orange line is the best-fit linear regression over the past 30 years of “extreme precipitation” in Grand Rapids.  It is a flat line (actually, it is angled slightly downward – not upward).

In the Detroit climate region, there has not been a significant trend in these types of “extreme precipitation” events since records began in 1874.  The correlation is negative, toward fewer events per year over the last 140 years.  Over in the Flint region, no significant trend since records began in 1921.  Same applies in the Muskegon region since records began 1898, and in the Houghton Lake area since records began 1918, and in the Sault Ste. Marie area since records began in 1889.  Records begin in 1864 for the Lansing area, and no significant trend (yet again, a negative correlation toward fewer events, not more).  Same applies in the Alpena sub-region since records began in 1918, and in the Marquette County area since records began in 1872.  No significant trend in the Grand Rapids area since continuous records began in 1903, either.

In other words, none of the individual climate sub-regions in Michigan have significant increasing trends in the number of annual extreme precipitation events since records began, or over the past 50 years.  Hopefully the policymakers in this state will consider these details when reading the NRDC/RMCO report.