Community Member
Community Member
kudos icon +

4.0 Weather Decision Services

Relabel Extreme Rainfall Events so the Public Understands Their Severity

Do not label extreme rain events by recurrence interval (e.g. 100-year storm), implying they are rare. Use a scale that indicates the severity of the storm, just like you do for tornadoes, earthquakes and hurricanes. None of those are called the 100-year tornado, etc. Below is a white paper I wrote on an alternative to this method. I have presented this nationally at ASFPM and at a regional NWS conference. Here is my paper.

 

Relabeling Extreme Rainfall Events so the Public Understands Their Severity

By

Thomas M. Grisa, P.E., F.ASCE

Mr. Grisa is Director of Public Works for the City of Brookfield. Contact him by email at: grisa@ci.brookfield.wi.us or by mail at 2000 N. Calhoun Road, Brookfield, WI 53005.

 

Abstract

 

Why are rain events labeled based on recurrence intervals? The public does not understand how the 100-year rain event can occur more than once in 100 years. In addition, two 100-year events may occur that are completely different in duration and intensity, but are both called the 100-year event. This is confusing to the public and leads to problems for professionals to explain how this could happen.

 

Other natural disasters (e.g. earthquakes, tornadoes, hurricanes) are rated using scales based on other factors than recurrence intervals or probability of these events occurring. The public understands the ratings for these other natural weather events since they use a simple scale to rate the severity of the event, not the rarity of the event.

 

This paper proposes that the industry rerate rainstorms, building on existing science, but changing the designation from a recurrence interval standard to one that is more general in description and more understandable to the public, similar to the rating of other natural disasters.

 

Subject Headings:

1. Floods

2. Hydrology

3. Municipal government

4. Natural disasters

5. Probability

6. Rainfall

7. Rainfall frequency

8. Stormwater management

 

Frustration – It Happened Again

 

The 100-year storm. How many times must I tell a resident in my career that the City has experienced yet another 100-year storm? It has now happened six times in Brookfield in the past 24 years, since 1986. It happened back to back in 1997 & 1998 and back to back to back in 2008, 2009, and 2010. Residents do not understand this, and it can be difficult explaining it in terms they can understand.

 

Professionals in the industry understand the statistics used to develop recurrence intervals and the probability of rain events, given the historic rains that occur in an area. But as a practitioner, this understanding and our explanation of it to residents does not help people understand the severity and magnitude of these events or the statistical probability of these events happening every year. Residents think the professionals are either dumb or lying about the event, thus destroying the credibility of all involved.

 

Why do we as a profession label these events based on their recurrence interval? Some in the industry have changed their approach to this and are using probability of rain event instead. But the public is not easily fooled. They can easily figure out that a one percent storm has a 1 in 100 chance of occurring, and therefore it’s the 100-year storm.

 

This paper offers an alternative rating system for rainstorms. It should be noted that the proposed system does not apply to rainfall runoff, floods, or floodplains, since these are influenced by many other factors than just rainfall.

 

Other Types of Natural Disasters Defined Differently

 

Other professionals label natural disasters in other ways. Seismologists use the Richter scale to rate earthquakes, basing the rating on a measure of the amount of energy released as the strength and duration of the earthquakes seismic waves. The Fujita scale (or F-scale), now the Enhanced Fujita scale (since 2007) uses the intensity and area affected / damage created by tornadoes to rate them. Meteorologists use the Saffir-Simpson scale to rate hurricanes using barometric pressure, wind speeds and storm surge to define a hurricane’s intensity into categories.

 

These professionals do not estimate recurrence intervals for these events or predict probability of these events occurring. They use a simple scale to describe the severity of the event. The public understands that for earthquakes, tornadoes and hurricanes, the higher the number the worse the storm.

 

Severity versus Rarity

 

The public identifies with the ratings for these other natural weather events since their main concern is how severe the event may be. However, the current rating for rain events defines how rare the event is, not how severe it is. It is not uncommon for severe storms to hit an area more than once in a decade and sometimes two years in a row, as has been the case in Brookfield. Defining the storm by its severity impresses on the public that these are huge events and people should do what they can to minimize the storm’s impact to their property. Defining the storms by their frequency unfortunately misleads the public into thinking that once it happens, it will be a long time until it happens again.

 

It is inappropriate to continue to foster this notion among the public that these storms are rare. The profession needs a different method of describing these storms to the public to reinforce the concept of severity of the storm, not rarity of the storm.

 

An Alternate Rating System Proposed

 

The industry should reevaluate how it rates these rainstorms and change its designation from the recurrence interval and probability standard to one that is more general in description and more understandable to the public, similar to how we rate the other natural disasters. This new system can use existing science that is already in place. We can modify what we call these events without necessarily changing the science behind it.

 

Specifically the storms could be rated based on rainfall intensities and total rainfall accumulation. This builds on and uses the data and science already established for recurrence interval storms. So the analysis of these rainfall events and design of infrastructure does not need to change. Essentially this proposal is to add a public user interface on the statistical analysis performed behind the scenes, not unlike adding a graphical user interface to a computer model. The underlying methodology and software for the computer model does not change, but what the user sees does change, making it is easier for the user to perform data entry into and understand output from the model.

 

The proposed rating system identifies categories of storms, based on rainfall intensities and total rainfall accumulation and identifies the recurrence interval storm event associated with each rating. Essentially the category rating is the exponent (called a G-factor) applied to the number two to identify the recurrence interval for that rain event. This can be defined by the Recurrence Interval Conversion formula expressed below:

 

RI = 2^(G-1)

 

Where: RI = Recurrence Interval of Rain Event, and

G = the Category of the Storm

 

Solving for G is accomplished with logarithms, G = {1 + Log(RI)/Log(2)}.

Table 1 depicts the actual G-factor for commonly defined recurrence interval storms.

 

Table 1 – Identifying G-Factors for Specific

Recurrence Interval Storms

Recurrence Interval G-factor

2 year storm 2

5 year storm 3.32

10 year storm 4.32

25 year storm 5.64

50 year storm 6.64

100 year storm 7.64

 

Adjustment for Duration of Storm

 

Different rain events have different effects on runoff and flooding. When two very dissimilar events are both called the 100-year storm, the public gets confused. Short duration 100-year storms may result in culvert and roadside washouts, while long duration 100-year storms can result in widespread flooding.

 

To account for these differences there should be an adjustment factor to address this issue in the rating system. This can be done by considering the relationship of total rainfall by recurrence interval of a given duration to the 24 hour duration storm. This ratio can then be used as an adjustment to the selected category storm. Table 2 shows the rainfall depth in inches for Southeastern Wisconsin by recurrence interval.

 

Table 2 - Recurrence Interval and Depth of Rainfall (inches)

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 1.31” 1.60” 1.84” 2.20” 2.50” 2.82”

2 hour 1.54” 1.93” 2.23” 2.73” 3.16” 3.64”

3 hour 1.68” 2.07” 2.40” 2.93” 3.39” 3.89”

6 hour 1.95” 2.40” 2.79” 3.44” 4.03” 4.70”

12 hour 2.24” 2.74” 3.17” 3.89” 4.53” 5.25”

24 hour 2.57” 3.14” 3.62” 4.41” 5.11” 5.88”

Rainfall data is based on Milwaukee rainfall data for the 108-year period of 1891 to 1998.

Source: Rodgers and Potter (2000)

 

Note: There is also data for rainfall periods exceeding 24 hours, but it is easiest for the public to understand rain storm classifications in terms of one day duration or less rainstorms, so the rest of this data is not provided or included in this system (though it could certainly be extrapolated if needed).

 

Using this data, a simple ratio between the total rainfall for a given duration event as compared to the 24-hour duration event becomes the duration adjustment factor. The 24-hour duration is the basis for comparison since most people think of weather in terms of days and civil engineers most commonly use this duration for design storms. The duration adjustment factor formula is:

 

Duration Adjustment Factor (DAF) = Total rainfall for X-year Y hour duration storm/Total rainfall for X-year 24 hour duration storm

 

The ratio of total rainfall by duration as compared to the 24-hour duration event for each storm identified in Table 2 is shown in Table 3.

 

Table 3 –

Recurrence Interval and Ratio of Rainfall by Duration to the 24 hour Storm

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 51 % 51 % 51 % 50 % 49 % 48 %

2 hour 60 % 61 % 62 % 62 % 62 % 62 %

3 hour 65 % 66 % 66 % 66 % 66 % 66 %

6 hour 76 % 76 % 77 % 78 % 79 % 80 %

12 hour 87 % 87 % 88 % 88 % 89 % 89 %

24 hour 100 % 100 % 100 % 100 % 100 % 100 %

 

To refine the rating of the storm for duration, multiply this duration adjustment factor by the G-factor for all storms based on duration. For example, a 100-year storm is defined as a Category G-7.64 storm in accordance with Table 1. For a 100-year storm that has a 1 hour duration, the duration adjustment factor is 48%. Multiplying 7.64 by 48% results in 3.67, round to G-4. Therefore, this 1 hour duration 100-year storm would be defined as a Category G-4 storm.

 

Using Table 1 to rate the category of storms, apply the duration adjustment factor from Table 3 to each storm. This refinement of the storms shown in Table 1 is shown in Table 4.

 

Table 4 – Duration Adjusted Category Storms (G-factor)

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 1 2 2 3 3 4

2 hour 1 2 3 3 4 5

3 hour 1 2 3 4 4 5

6 hour 2 3 3 4 5 6

12 hour 2 3 4 5 6 7

24 hour 2 3 4 6 7 8

 

At first glance, this rating of storms looks reasonable. The larger and longer duration storms rate the highest on this scale and smaller storms with lower durations ranks lowest. But when one overlays the G-factor rating against the actual total rainfall for each storm by duration, there is an inconsistency that shows up that requires an additional modification to the system.

 

Substituting the total rainfall for the Milwaukee area from Table 2 into the respective cells in Table 5 shows that some storms have similar rainfall totals over different durations, shown by way of example in the bold and italicized numbers in Table 5. Unfortunately, this system rates them all the same (G-6) as represented by the color yellow in the table.

 

Table 5 – Duration Adjusted Category Storms with Total Rainfall (inches)

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 1.31” 1.60” 1.84” 2.20” 2.50” 2.82”

2 hour 1.54” 1.93” 2.23” 2.73” 3.16” 3.64”

3 hour 1.68” 2.07” 2.40” 2.93” 3.39” 3.89”

6 hour 1.95” 2.40” 2.79” 3.44” 4.03” 4.70”

12 hour 2.24” 2.74” 3.17” 3.89” 4.53” 5.25”

24 hour 2.57” 3.14” 3.62” 4.41” 5.11” 5.88”

 

The larger amount of rain in a shorter duration cannot rate the same as a longer duration event with a lower total rainfall. This will continue to confuse the public, as it does not make sense. It must rate higher on this scale. To accomplish this, we must adjust the rating using an intensity adjustment factor.

 

Using the rainfall data in Table 2, a simple ratio between the total rainfall for a given duration event divided by the total rainfall for the 100-year storm of the same duration becomes the intensity adjustment factor. The 100-year event is the basis for this adjustment since this is typically the most common extreme event for which the public is familiar and is used for design storms. The intensity adjustment factor formula is:

 

Intensity Adjustment Factor (IAF) = Total rainfall for X-year Y hour duration storm/Total rainfall for 100-year Y hour duration storm

 

The ratio of total rainfall by intensity as compared to the 100-year event for each storm identified in Table 2 is shown in Table 6.

 

Table 6 –

Recurrence Interval and Ratio of Rainfall by Intensity to the 100-year Storm

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 46 % 57 % 65 % 78 % 89 % 100 %

2 hour 42 % 53 % 61 % 75 % 87 % 100 %

3 hour 43 % 53 % 62 % 75 % 87 % 100 %

6 hour 41 % 51 % 59 % 73 % 86 % 100 %

12 hour 43 % 52 % 60 % 74 % 86 % 100 %

24 hour 44 % 53 % 62 % 75 % 87 % 100 %

 

To adjust the rating of the rain event one last time, multiply the intensity adjustment factor in Table 6 by the refined rating of the storm in Table 5. The category storm determined from the original formula identified above will then change based on these two adjustment factors.

 

The adjusted Recurrence Interval Conversion formula becomes:

 

GADJ = {1 + Log(RI)/Log(2)} x (DAF) x (IAF)

 

Where:

GADJ = Adjusted Category of Storm

RI = Recurrence Interval

DAF = Duration adjustment factor, where:

DAF = Total rainfall for X-year Y hour duration storm/

Total rainfall for X-year 24 hour duration storm

IAF = Intensity adjustment factor, where:

IAF = Total rainfall for X-year Y hour duration storm/

Total rainfall for 100-year Y hour duration storm

 

For example, a 50-year storm is a Category G-6.64 storm in accordance Table 1. For a 50-year storm that has a 12 hour duration, the duration adjustment factor is 89% from Table 3. The intensity adjustment factor is 86%. Using the adjustment formula yields:

 

6.64 x 89% x 86% = 5.08, round to 5.

 

Therefore, this 12 hour 50-year storm would be defined as a Category G-5 storm, not the G-6 as rated using solely the duration adjustment factor.

 

Using Table 1 to rate the category of storms and then applying the adjustment factors for duration and intensity to each storm (or using the adjusted Recurrence Interval Conversions formula) yields the category ratings for the rain events as shown in Table 7.

 

Table 7 – Fully Adjusted Category Storms (G-factor)

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 0 1 1 2 3 4

2 hour 1 1 2 3 4 5

3 hour 1 1 2 3 4 5

6 hour 1 1 2 3 4 6

12 hour 1 2 2 4 5 7

24 hour 1 2 3 4 6 8

 

Checking the results of this revised table to see if there are storms rated the same that should not results in a reasonable rating for each storm as shown in Table 8.

 

Table 8 – Fully Adjusted Category Storms with Total Rainfall (inches)

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 1.31” 1.60” 1.84” 2.20” 2.50” 2.82”

2 hour 1.54” 1.93” 2.23” 2.73” 3.16” 3.64”

3 hour 1.68” 2.07” 2.40” 2.93” 3.39” 3.89”

6 hour 1.95” 2.40” 2.79” 3.44” 4.03” 4.70”

12 hour 2.24” 2.74” 3.17” 3.89” 4.53” 5.25”

24 hour 2.57” 3.14” 3.62” 4.41” 5.11” 5.88”

 

The intensity adjustment factor corrected the aforementioned inconsistency. A more careful evaluation of each storm shows that there are minor inconsistencies at the smaller storms. This is a result of rounding of the actual adjusted G-factor. It is not worth further modification of this system as each category has a range associated with it. However, for those who find even this slight discrepancy unacceptable, Table 9 shows the actual adjusted but unrounded G-factor associated with each storm, thus resolving this inconsistency associated with rounding to whole numbers.

 

Table 9 – Specific G-factor Fully Adjusted Category Storms

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 0.47 0.96 1.43 2.20 2.88 3.66

2 hour 0.51 1.08 1.63 2.62 3.56 4.73

3 hour 0.56 1.16 1.77 2.82 3.84 5.05

6 hour 0.63 1.30 1.98 3.22 4.49 6.11

12 hour 0.74 1.51 2.28 3.69 5.08 6.82

24 hour 0.87 1.77 2.66 4.23 5.77 7.64

 

Defining the Numbers

 

From National Weather Service furnished items, the author developed descriptions for the numbers on the G-factor scale to provide additional meaning to the public, similar to the earthquake and hurricane scales. Below in Table 10 is the Rain Storm Severity Index. This index describes the scale using simple language instead of numbers as some may find that more understandable. Proposed descriptions are given below for ratings on this scale:

 

Table 10 – Rain Storm Severity Index

G-Factor Description of rain event

1 to 2 Minor

3 to 4 Moderate

5 to 6 Major

7 to 8 Extreme

9 to 10 Catastrophic

 

These descriptions provide a sense of severity of the storm, confirming the public’s understanding with other scales that the bigger the number the more severe the storm.

 

Does this System Work Elsewhere?

 

Engineers, meteorologists and other professionals can use this method throughout the country. A check of two significantly different areas, from a hydrometeorology perspective, shows this system works everywhere. Las Vegas, NV is very dry and receives little and infrequent precipitation and accordingly has low rainfall totals for the high recurrence interval storms (less than 3 inches of rain in its 100-year 24 hour duration storm). Washington D.C. on the other hand is much wetter, receiving over 8 inches of rain in its 100-year 24 hour duration storm. Precipitation data was found on the National Oceanic and Atmospheric Administration (NOAA) website (Hydrometeorological Design Studies Center - HDSC) for these two cities.

 

The proposed rating system for rainstorms categorized each of the recurrence interval storms with 24 hour durations or less from the NOAA database. This did not appear to result in any difficulty or material change in relationships using the proposed rating system. Large and long duration storms rated highest and the ratings proceeded in a logical manner decreasing with duration and intensity. Using this system in both regions yields similar results to that for southeastern Wisconsin as far as the relationship of storms one to another.

 

The results of this analysis are shown in Tables 11 and Table 12.

 

Table 11 – Category Storms for Las Vegas, NV

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 0 1 1 2 3 4

2 hour 0 1 1 3 4 5

3 hour 0 1 2 3 4 6

6 hour 1 1 2 3 5 7

12 hour 1 1 2 4 5 7

24 hour 1 2 3 4 6 8

 

Table 12 – Category Storms for Washington D.C.

Recurrence Interval

Duration 2 year 5 year 10 year 25 year 50 year 100 year

1 hour 0 1 1 2 2 3

2 hour 0 1 1 2 3 4

3 hour 1 1 2 2 3 4

6 hour 1 1 2 3 4 5

12 hour 1 1 2 3 5 6

24 hour 1 2 3 4 6 8

 

Criticism and Defense of this Proposal

 

Some have criticized this system for being too simplistic. That is, however, precisely the point. It should be simple for the public to understand, the bigger the number the more severe the storm, not the more rare the storm.

 

Another criticism is that it is inherently difficult to categorize all variety of storms into this simplistic system given the variety of rainfall events, the changes in rainfall intensity that occur during a rainfall event, and the subjective method of defining the duration of the rainfall event to identify the category above. However, this is also true of the current system, so this proposed change is no worse than the use of recurrence intervals in this regard.

 

The G-factor scale does not have a continuous curve of events like that for the recurrence interval system. However, if the durations of the event are shorter or longer than those shown in the tables above, projections or interpolations can be made to rate the storm into the associated category. This should be sufficiently accurate to present to the public.

 

This proposal introduces a more appropriate scaling factor to the storm rating system by defining these in categories that increase numerically by one. Many people are surprised to hear that the total rainfall from a 2-year storm is almost 50% of the total rainfall from a 100-year event for a given duration. Many wrongly assume that the 2-year event has 2% of the rain that a 100-year storm has, or that the 50-year event has 50% of the rain as a 100-year event. That is not the case. Using this proposed method it is more clearly understood that the Category G-6 storm is almost as severe as the Category G-7 storm.

 

Some have suggested including an adjustment factor for things like antecedent moisture condition. This, they argue, would account for the increased runoff from the rain events that occur during periods of saturated soil conditions. However, that would make this a rating of the runoff from the storm and not the storm itself. Other rating systems rate the natural disaster or weather event, not necessarily the impact of the event. Impacts from earthquakes are different depending on soil conditions; stiff clays respond differently than sands and silt (e.g. liquefaction). Other factors that affect the impact of an earthquake include the type of construction materials and the building code standards used during construction. However, these things do not change the Richter scale number of the earthquake itself. Similarly, hurricanes have different affects in the United States where building codes are much stricter than those in the Caribbean, for example; yet the rating is not changed based on its impact.

 

Things that affect runoff from major storms include slope, soil type, topography, snow cover and land cover with impervious surfaces, etc., some of which can change block by block. These are factors that engineers and hydrologists use to determine runoff, but do not enter into the rating for the rain storm as they are more akin to the aforementioned construction variables that affect impacts from earthquakes. Accordingly, this proposed method does not include these types of modifications.

 

The recurrence interval method of rating storms leads to common misperceptions regarding floodplains and their association with the rainfall events with similar sounding names. The public mistakenly believes the 100-year floodplain fills only when there is a 100-year storm and the 100-year storm will always fill the 100-year floodplain. Not necessarily so; yet try explaining that to a crowd of angry residents who already experienced the 100-year flood twice in as many years.

 

So not only should we consider modifying the 100-year storm rating system, but that could lead to a relabeling of the 100-year floodplain, which will have a positive impact on how the public perceives and understands the flood risk inherent to their property. If this is what it takes to get the public to take positive steps to protect themselves during these major rain events then that is a step in the right direction and could have lasting and significant impacts throughout areas that are prone to or more likely to flood.

 

Conclusion

 

The current system for rating extreme rain events does not work for relation of the event to the public. It is too hard to explain and confusing to the public. The current system does not serve the public well by calling these storms something that the public perceives they are not and that they do not understand. Too much work has gone into defining these complex natural events only to see it rejected by the public.

 

A revised method, building on the existing science that is already in place but using a similar method as used for rating other natural disasters, will work better. The method proposed in this paper develops a simple scale that the public understands, the bigger the number, the bigger the storm. Changing the rating of these storms is imperative for public acceptance. This will require the cooperation of all professionals in this industry. Anyone with me?

 

Acknowledgements

 

The author would like to express gratitude to the engineers, floodplain managers, hydrologists, and practitioners who reviewed early versions of the white paper proposing a change in the rating of the extreme rains. Their comments, questions and encouragement were instrumental in modifying this method and addressing issues that improved the overall concept. Specific thanks go to: Theresa Caven, Rick Eilertsen, Dave Fowler, Mike Hahn, Victor Hom, Larry Larson, Mike Lemens, Tom Simmons, and Janet Thigpen

 

References

 

Loucks, E. et. al. (2000). “Technical Report No. 40 – Rainfall Frequency in the Southeastern Wisconsin Region.”, Chapter 4 – Frequency Analysis of Extreme Rainfall in the Southeastern Wisconsin Region, Table 18, Southeastern Wisconsin Regional Planning Commission, Waukesha, WI (50).

 

National Oceanic and Atmospheric Administration (NOAA) National Weather Service website (2009). “Hydrometeorological Design Studies Center Precipitation Frequency Data Server.” http://hdsc.nws.noaa.gov/hdsc/pfds/sa/nv_pfds.html and http://hdsc.nws.noaa.gov/hdsc/pfds/orb/md_pfds.html. (Nov. 2, 2009)

Tags

Voting

168 votes
Active
Idea No. 82