Office of Planning, Environment, & Realty (HEP)
Planning • Environment • Real Estate
Increases in precipitation or drought events can significantly affect transportation infrastructure and are an important consideration in the design, operation, and maintenance of the transportation system. For example, heavy precipitation events can cause flooding, mudslides, landslides, soil erosion, and adversely high levels of soil moisture. These hazards directly affect the structural integrity and maintenance of roads, bridges, drainage systems, and tunnels.
This section presents the methodology and key findings for several analyses related to observed and projected precipitation.
Additional detail about the precipitation analyses is available in the appendices.
Observed precipitation records for Mobile were analyzed to describe historical and current climate conditions in the region.
This section describes the methodology used to analyze observed precipitation data in the Mobile region.
Historical data from the five NOAA GHCN stations (see Figure 4) in the Mobile region were analyzed to investigate existing climatic trends and baseline conditions.3 Table 13 summarizes the data available for each station and identifies any gaps in the record.
| Station | Station ID# | County | Precip. Start of Data Collection | Precip. Data Gaps* |
|---|---|---|---|---|
| Bay-Minette | USC00010583 | Baldwin | 1914 | 1931, 1937-1941 |
| Coden | USC00011803 | Mobile | 1956 | 1975, 1985-1988 |
| Fairhope | USC00012813 | Baldwin | 1918 | - |
| Mobile Airport | USC00015478 | Mobile | 1948 | - |
| Robertsdale | USC00016988 | Baldwin | 1912 | - |
* Gaps defined as over 80% of data points missing for the year
** Missing minimum temperature only
Each station records total daily precipitation. This study uses observed daily precipitation to represent 24-hour precipitation events. In Alabama, winter storms tend to last at least one day or longer, while summer storms tend to last less than a few hours.4 This suggests that the heaviest 24-hour precipitation event during winter may be split between two consecutive days of measurements, while the heaviest 24-hour precipitation event during summer would likely be captured by daily measurements.
To get monthly, seasonal, and annual averages, the daily data by month, season, and year were averaged across the historical record (1912 to 2009), where data were available. In addition, averages for the present-day climate period (1980 to 2009) were calculated. Daily precipitation totals were summed by month, season, and year, and then averaged for the entire historical record (1912-2009) and present-day climate period (1980-2009).5
Key Findings for Historical Precipitation in the Mobile Region
This section describes key findings from the analysis of observed precipitation.
As with temperature, precipitation patterns demonstrate similar ranges and trends over time across all five stations. While some stations show anomalous dips in precipitation totals, this is likely due to incomplete data, instrument error, or recording error for that year. Figure 28 illustrates total annual precipitation for each station. Averaging across all stations and years, annual precipitation in Mobile was 65.3 inches (165.9 centimeters). Year-to-year variability suggests annual rainfall can vary by as much as 13.4 inches (34.0 centimeters) (20%). This greater variability in the precipitation record compared to the temperature record is expected, as precipitation is more heavily influenced by small-scale phenomena (e.g., coastal breezes).
Figure 28: Annual Precipitation (inches)
Table 14 provides annual precipitation for each station over both the full station record (see Table 13 for each station's full record),6 the present-day climate period (1980 to 2009), and the historical record used to inform the trend analysis (1961 to 2010).7 The table also notes the standard deviation, or the amount of variability, in the data. The table shows relatively consistent average annual precipitation levels from station-to-station and from period-to-period. The only significant trend observed in the annual precipitation is at the Bay-Minette station, where precipitation demonstrates an increasing trend.
| Station | Full Station Record | 1980-2009 | 1961-2010 | Historical Trend? |
|---|---|---|---|---|
| Bay-Minette | 63.2 (14.0) | 69.1 (16.1) | 66.0 (15.0) | increasing* |
| Coden | 63.1 (13.8) | 67.1 (13.4) | 63.8 (14.2.3) | no trend |
| Fairhope | 65.0 (13.1) | 68.0 (12.3) | 66.2 (13.0) | no trend |
| Mobile | 65.2 (11.2) | 66.7 (11.3) | 64.9 (11.5) | no trend |
| Robertsdale | 66.1 (14.2) | 66.3 (12.9) | 65.4 (14.1) | no trend |
| Average | 67.4 (13.2) | 63.1 (15.4) | 65.3 (13.6) |
*Statistically significant at the: *90% Confidence Level, **95% Confidence Level, ***99% Confidence Level
Note: The standard deviation representing variability across the time period is provided in parentheses. Mann Kendall results exploring historical trend are also provided.
Precipitation associated with the maximum 24-hour precipitation event recorded each year has fluctuated over the historical record, exhibiting no significant trends. Historically, precipitation during the maximum annual 24-hour precipitation event in the Mobile region has ranged between 1.3 inches (3.3 centimeters) at Bay-Minette in 1931 to 17.5 inches (44.5 centimeters) at Robertsdale in 1917. Averaged across all stations, precipitation associated with the maximum 24-hour precipitation event has ranged between 2.4 inches (6.1 centimeters) in 1938 and 14 inches (35.6 centimeters) in 1917. Precipitation associated with the average maximum 24-hour precipitation event across all stations from 1912 to 2009 was 5.2 inches (13.2 centimeters) (see Figure 29). A few extreme precipitation events can be traced to a hurricane event.8 This suggests future changes in high precipitation return periods are likely to be partially driven by changes in hurricane activity. This figure may underestimate actual maximum precipitation due to gage failures that can occur during periods of heavy rainfall.9
Figure 29: Maximum 24-Hour Precipitation Events Recorded Each Year (inches)
Figure 30 and Figure 31 illustrate seasonal and monthly precipitation, respectively, for each GHCN station in the Mobile region, 1912 to 2009. All stations demonstrate a similar distribution of precipitation over the year. The summer is the wettest season. July is the wettest month with approximately 8.1 inches (20.6 centimeters) of rainfall. October is the driest month, with approximately 3.7 inches (9.4 centimeters) of rainfall. Over the historical record, monthly precipitation has increased significantly (p<0.10) at all five stations in January, October, and November. Summer precipitation has increased significantly at all five stations.
Figure 30: Average Seasonal Precipitation Totals (inches), 1912-2009
Figure 31: Average Monthly Precipitation Totals (inches), 1912-2009
Table 15 provides a summary of the present-day (1980 to 2009) averages for the precipitation variables, based on observed data. This table serves as a comparison for the projected precipitation discussion in Section 4.2 (see Table 16 for a description of how these variables were developed).
| Precipitation Event | 1980-2009 | Precipitation Event | 1980-2009 |
|---|---|---|---|
| Maximum 3-Day Winter Precipitation | 15.3 inches | 1% exceedance probability for the 2-Day Storm Event | 5.5 inches |
| Maximum 3-Day Spring Precipitation | 15.7 inches | 1% exceedance probability for the 4-Day Storm Event | 6.9 inches |
| Maximum 3-Day Summer Precipitation | 20.2 inches | ||
| Maximum 3-Day Fall Precipitation | 14.2 inches |
This section describes the methodology and key findings for the analysis of future precipitation in the Mobile region.
Terminology
Precipitation projections for Mobile were statistically downscaled using the same methodology as was used for temperature projections. This methodology is described in detail in Section 3.2. See Appendix C.3 for detailed methodology specific to the precipitation variables. Projected precipitation was estimated using daily downscaled daily precipitation data from up to ten climate models (see Table 11) under three emission scenarios (B1, A2, A1FI). Note that climate models do not simulate regional precipitation as well as regional temperature due to the higher spatial and temporal variability associated with precipitation.10
Table 16 provides a list of the precipitation-related weather hazards and climatic averages that are investigated in this study. The methods used to develop these datasets vary by precipitation variable. For example, the 24-hour precipitation projections are developed by applying a Gumbel extreme value distribution (which is traditionally used for this purpose) to annual duration data to obtain the probability of a precipitation event occurring in a given year (see textbox entitled, "Storm Event Probabilities").11 Other precipitation variables are developed by applying a quantile distribution to the 30-year simulation; this procedure is not intended to extrapolate beyond the 30-year dataset (e.g., this distribution will not provide the precipitation of an event that has a 1% chance of occurring in any given year). In Table 16, asterisks denote the variables and percentiles that do not provide robust quantitative results (per communication with Dr. Katharine Hayhoe) and their use should be limited to qualitatively informing the impact assessment. In general, these asterisks denote cases where extreme events are based on a small sample size—e.g., the 24-hr precipitation event with a 0.2% chance of occurring per year (i.e., a 1-in-500-year event) is based on fitting a theoretical distribution curve to only 30 data points and extrapolating to the tails of the curve. In contrast, precipitation projections based on a large sample size are considered robust—e.g., the two-day storm event with a 0.2% chance of occurring within the 30-year period is based on a running two-day sum over the entire 30-year period for a total of more than 10,000 data points.
| Variable | Transportation Mode | Methodology |
|---|---|---|
| Annual, seasonal, and monthly total precipitation for each 30-year time period | Multi (pavement design) | Daily precipitation corresponding to each month, season, or year was summed for each year, station location, climate model, and emission scenario. Then the 30-year average of each sum was determined. Averages and standard deviations were calculated across climate models for each station location and emission scenario. For purposes of discussion, the results were averaged across station locations to produce an average for the Mobile region. |
| Precipitation for 24-hour period with a 0.2%*, 1%*, 2%, 5%, 10%, 20%, and 50% probability of occurrence per year | Multi (drainage, liquid storage) | The day with the maximum total daily precipitation for each year was found for each emission scenario, climate model, and station location. This produced a total of 30 data points for each time period. Across the 30 data points, the daily precipitation representing each probability of occurrence was estimated for each emission scenario, climate model, and station location by applying a Gumbel extreme value distribution. Averages and standard deviations were calculated across climate models for each station location and emission scenario. For purposes of discussion, the results were averaged across station locations to produce an average for the Mobile region. |
| Occurrence of precipitation for 24-hour period based on today's 0.2%*, 1%*, 2%, 5%, 10%, 20%, and 50% probability of occurrence per year | Multi (drainage) | For the 1980 to 2009 time period, the value of the occurrence probabilities using the maximum total daily precipitation was identified using the results of the variable above for each climate model, emission scenario, and station location. For each of the future time periods, the day with the maximum total daily precipitation for each year was found for each emission scenario, climate model, and station location. This produced a total of 30 data points. Across these 30 data points, the occurrence probabilities were determined by applying a Gumbel extreme value distribution. These fitted distributions provided the new probabilities associated with the historical value of each baseline occurrence probabilities. Averages and standard deviations were calculated across climate models for each station location and emission scenario. For purposes of discussion, the results were averaged across station locations to produce an average for the Mobile region. |
| Exceedance probability of precipitation across four consecutive days for each 30-year period: 0.2%, 1%, 2%, 5%, 10%, 20%, 50%; Exceedance probability of precipitation across two consecutive days for each 30-year period: 0.2%, 1%, 2%, 5%, 10%, 20%, 50% |
Pipeline | For each time period, a sum of daily precipitation was calculated for every four consecutive days. This produced a total of 10,950 data points. The data was ranked from high to low, and the exceedance probabilities of 0.2%, 1%, 2%, 5%, 10%, 20%, and 50% were then determined for each climate model, emission scenario, and station location by applying a quantile distribution. Averages and standard deviations were calculated across climate models for each station location and emission scenario. For purposes of discussion, the results were averaged across station locations to produce an average for the Mobile region. This was repeated for the two-day exceedance probabilities. |
| Largest three-day total precipitation each season | Multi | The maximum three-day total precipitation for each season was identified for each year. This produced 30 data points for each of the four seasons. The 30 data points were averaged to produce the average maximum three-day total for each season. For purposes of discussion, the results were averaged across station locations to produce an average for the Mobile region |
*Variables and percentiles that do not provide robust quantitative results (per communication with Dr. Katharine Hayhoe). Their use should be limited to qualitatively informing the impact assessment.
Overall, there is a high degree of variability in the precipitation results. Only certain projections show significant change from simulated baseline conditions and many variables (such as total annual precipitation) are not projected to change significantly. Generally, the variability across downscaled climate models is much greater for projections of precipitation than it is for projections of temperature. Findings suggest that total annual precipitation may not change significantly but the timing of that precipitation may change.
The amount of projected change in precipitation was not proportional to the projected change in emissions. The most notable changes in annual precipitation occurred under the low (B1) emission scenario, including an increase of nearly 7 inches (17.8 centimeters) by mid-century and 8.4 inches (21.3 centimeters) by end-of-century. Meanwhile, projections under the moderately-high (A2) and high (A1FI) emission scenarios do not demonstrate a significant change from simulated baseline conditions for any future time period.
The sections below present the results of the projected precipitation analysis, including a discussion of:
Tables including more detail about the projected changes are available in Appendix E.2.
Key Findings for Annual Average Precipitation
Only two projections for total annual precipitation are considered statistically significantly different from the simulated baseline. Under the low (B1) emission scenario, total annual precipitation is projected to increase 6.9 inches (17.5 centimeters) by mid-century and 8.4 inches (21.3 centimeters) by end-of-century. Figure 32 illustrates how total annual precipitation is projected to change over time in the Mobile region, as a function of emission scenario. As illustrated in the figure, the projected changes indicate that total annual precipitation under the three emission scenarios will be similar to the simulated baseline, particularly when variability is taken into account.
Figure 32: Projections of Total Annual Precipitation
The uncertainty associated with the climate model ensemble mean grows with time, which suggests some disagreement between climate models in the magnitude and direction of projected changes in precipitation. In general, precipitation changes are more notable at the seasonal and monthly scale. In other words, total annual precipitation may not change dramatically, but the timing of precipitation may shift.
Key findings for projected change in total annual precipitation relative to simulated baseline (1980-2009) are as follows:
Key Findings for Average Seasonal and Monthly Precipitation
Very few projections for seasonal and monthly precipitation demonstrate statistically significant changes from the simulated baseline. Under the low (B1) emission scenario, winter precipitation is projected to increase significantly both in the near-term and by mid-century, by 1.6 and 1.7 inches (4.1 and 4.3 centimeters), respectively. Meanwhile, fall precipitation under the low emission scenario is projected to increase 2.2 inches (5.6 centimeters) by mid-century. Figure 33 illustrates the projected total monthly precipitation of the climate model ensemble at the end-of-century, as a function of emission scenario.
Figure 33: Projected End-of-Century Change in Total Monthly Precipitation, Averaged Across Climate Models and Station Locations and Relative to Simulated Baseline (1980-2009)
Key findings for the projected change in total seasonal and monthly precipitation relative to simulated baseline (1980-2009) are as follows:
Key Findings for Precipitation Events
Precipitation events covered in this section include:
Note that climate models may underestimate changes in precipitation events. This is because climate models tend to produce rainfall events that are less intense than observations, in part due to the models' low spatial resolution (see textbox titled, "Underestimating Precipitation Events").12 However, as discussed in the methodology section above, the simulated precipitation events for baseline conditions (1980 to 2009) tended to be overestimated compared to observations.
Underestimating Precipitation Events
Scientists have relatively high confidence in the ability of climate models to simulate changes in mid-latitude storms and jet streams. However, climate models may not do a good job of capturing precipitation events in the Mobile region, particularly extreme events such as 1-in-100-year events (i.e., 1% probability of occurring in any given year). In part, this is because tropical storms and hurricanes may represent a sizeable portion of extreme storms in the area, and small-sized tropical storms and hurricanes are not reliably simulated by climate models. Other events, such as summertime convective thunderstorms, are too small in scale to be well represented and require the use of parameterization schemes. Overall, future changes of the very extreme storms that are developed from model projections (i.e., changes in 1% and 0.2% probability of occurrence) are very uncertain.
Maximum seasonal three-day precipitation is projected to increase across all seasons, emission scenarios, and time frames. However, not all increases are statistically significant.
Under the low (B1) emission scenario, projected increases are statistically significant in the winter in the near-term; in the winter, summer, and fall by mid-century; and in the winter and fall by end-of century. Under the moderately-high (A2) emission scenario, only the projected increase in winter by the end-of-century is statistically significant.
Key findings for projected change in seasonal three-day precipitation events relative to simulated baseline (1980-2009) are as follows:
Figure 34: Projected Change in Maximum Three-Day Precipitation, from Baseline (1980-2009) to Mid-Century (2040-2069)
Figure 35: Projected Change in Maximum Three-Day Precipitation from Baseline (1980-2009) to End-of-Century (2070-2099)
The frequency and magnitude of 24-hour precipitation events are projected to increase significantly in the future under both the low (B1) and moderately-high (A2) emission scenarios (see textbox titled, "Storm Event Probabilities" for definition of storm events).13 The projections of storm events with low probability are less robust than other projections due to the small sample from which they are drawn. Additional research is needed to investigate the patterns suggested here for the storm events with a 1% or 0.2% chance of occurring in any given year. This section provides additional detail on the components contributing to the projected change in low probability storm events.
Storm Event Probabilities
Storm events, defined by 24-hour total precipitation, are classified by their likelihood of occurrence in any given year. For example, a storm with a 20% probability has a 20% chance of occurring in any given year and is likely to occur about once every five years. Based on the model simulations, a 20% storm in Mobile corresponds to a storm with 6.7 inches (17.0 centimeters) of precipitation. Projections of these storm event probabilities were developed by fitting a Gumbel Extreme Value distribution to 30 years of data for each time period and extrapolating to the distribution tails for the extreme values. This method mirrors the approach currently applied when working with observed data. For use in impact assessments, it is the projected change compared to today's observations that is considered.
|
||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a This database provides information on extreme rainfall for Alabama by applying a Gumbel Extreme Value distribution to over 10,738 station-years of record representing varying time periods, temporal resolution, and locations around the state; the values provided in this table are for Mobile, Alabama (Durrans and Brown, TRB Paper No. 01-0125).
b This technical paper provides contour maps of extreme precipitation events for the continental United States based observation data from 1940 to 1958. These values were derived based on the empirical data for the period of occurrences of 2 to 10 years, and applying a Gumbel procedure for the period of occurrences greater than 20 years (Hershfield 1961).
c To investigate how the precipitation events based on the simulated baseline (1980-2009) might represent precipitation events based on a longer period of record, precipitation observations at all five stations for the Mobile region were used to compare present-day climate (1980-2009) to the entire historical record. The results suggest the present-day record is representative of the historical record with similar percentages of occurrence for each bin of maximum precipitation per year (bins were: 0 to 3 inches, 3 to 6 inches, 6 to 9 inches, 9 to 10 inches, 10 to 11 inches, and greater than 11 inches). A pattern was noticeable where the present-day climate had slightly less occurrences of rain below 6 inches and slightly more occurrences of rain above 6 inches.
As part of this analysis, model simulations of 24-hour precipitation events from 1980 to 2009 were compared to historical observations to investigate the accuracy of the downscaled climate model simulations. This investigation indicates that the 24-hour precipitation events simulations are not likely to replicate the timing and magnitude of every observed event. However, the simulations are likely to replicate the nature of the precipitation events over an entire 30-year time period.14
For example, Figure 36 presents the maximum 24-hour precipitation event for each year from 1980 to 2099 as simulated by a single downscaled climate model, GFDLCM2.0, at the Bay-Minette location under the low (B1) emission scenario. For this time period, the simulation captures similar magnitude and variability as the events recorded at Bay-Minette (see Figure 29). Under this simulation, a number of extreme precipitation events are projected to occur towards the end of the century.
Figure 36: Simulated Maximum 24-Hour Precipitation Events of a Downscaled Climate Model under the Lower (B1) Emission Scenario at Bay-Minette (inches)
Figure 37 presents the maximum 24-hour precipitation events for each year from 1980 to 2099 as simulated by all 10 climate models at Bay-Minette under the low (B1) emission scenario. This figure indicates that downscaled maximum 24-hour precipitation events increase substantially in intensity after about 2015. This noticeable increase is particularly evident when comparing the baseline simulations (1980 to 2009) to the projected simulations (2010 to 2099).
Figure 37 also demonstrates the variability across climate models. For example, some models simulate extreme precipitation events in the near-term while other models simulate extreme precipitation events towards the end-of-century. The projected changes in extreme 24-hour precipitation events presented in this section are based on averages of the collection of climate model simulations, where each distinct simulation was fitted to a Gumbel extreme value distribution (see Table 16).
Figure 37: Simulated Maximum 24-Hour Precipitation Events of Downscaled Climate Models under the Lower (B1) Emission Scenario at Bay-Minette (inches)
Stationarity Versus Non-Stationarity
"Stationarity" assumes that while weather varies, the climate mean does not change over time. Infrastructure is currently designed under this assumption. However, changes observed in the hydroclimatic record suggest that "stationarity is dead." This is due to factors including humans changing the natural landscape and greenhouse gas emissions affecting climate.
"Non-stationarity" is already evident in climate observations and can be accounted for by extreme value analyses. However, there is much debate amongst scientists on how best to account for non-stationarity. Two key requirements to adequately account for non-stationarity are:
Although this is an area characterized by significant uncertainty, climate projections suggest that it would not be prudent to design infrastructure assuming stationarity.
For all time periods, precipitation associated with high-probability/low-impact storms (e.g., a storm with a 50% probability of occurrence in any given year) is projected to increase by 1 to 3 inches (3 to 8 centimeters). Meanwhile, precipitation associated with low-probability/high-impact storms (e.g., a storm with a 1% probability of occurrence in any given year) is projected to increase more, by up to 4 to 8 inches (10 to 20 centimeters). This suggests extreme storms will become more intense and potentially damaging.
By mid-century, precipitation associated with a 1% probability of occurrence is projected to increase by 4.9 inches (12.4 centimeters) under both the low (B1) and moderately-high (A2) emission scenarios. By the end of the century, precipitation associated with a 1% probability of occurrence is projected to increase by 6.0 inches (15.2 centimeters) under the moderately-high (A2) emission scenario. Figure 38 illustrates the projected change in precipitation totals for storms with a 1% probability of occurrence. Please see Appendix E.2 for tables of projected changes under all emission scenarios, for all time periods, and all storm event probabilities.
Figure 38: Projected Precipitation Totals for a Storm with a 1% Probability of Occurrence
In addition, under the low (B1) and moderately-high (A2) emission scenarios, storms experienced today across all return intervals are projected to be more likely to occur. For example, a storm with a 1% probability of occurrence, defined as approximately 12 inches (30 centimeters) of rainfall in 24 hours, is projected to increase up to 6 inches (15 centimeters) by the end of the century under both scenarios. Figure 39 illustrates the projected change in probability of historic storms with a 1% probability of occurrence. This figure illustrates that a storm with approximately 12 inches (30 centimeters) of rainfall in 24 hours is projected to increase in frequency by the end of the century to have roughly a 10% probability of occurring in any given year. Please see Appendix E.2 for tables of projected changes under all emission scenarios, for all time periods, and all storm event probabilities.
Figure 39: Projected Probability of the Historical Storm that has a 1% Probability of Occurrence
Key findings for projected change in 24-hour precipitation events relative to simulated baseline (1980-2009) are as follows (note that the 24-hour precipitation events described here do not include events associated with tropical cyclonic activity):
Two-Day and Four-Day Precipitation Events15
Rainfall during maximum two-day and four-day precipitation events is projected to increase significantly by mid- and end-of-century under all emission scenarios. The projections associated with the low (B1) emission scenario show the most significant changes from baseline conditions, but there are also some significant projected changes associated with the moderately-high (A2) and high (A1FI) emission scenarios. Projected changes in total precipitation for two-day and four-day precipitation events with a 1% exceedance probability are shown in Figure 40 and Figure 41, respectively.
Figure 40: Projected Total Precipitation for Two-Day Precipitation Events with a 1% Exceedance Probability
Figure 41: Projected Total Precipitation for Four-Day Precipitation Events with a 1% Exceedance Probability
Two-day and four-day precipitation events that are currently uncommon in the Mobile region are projected to become more frequent by mid-century and end-of-century (near-term projections suggest very little significant change from baseline conditions).
Key findings for projected changes in two-day and four-day precipitation events relative to simulated baseline (1980-2009) are as follows:
While minor changes in the total annual levels of precipitation are not likely to affect transportation, increases in the magnitude and frequency of precipitation events can have significant local impacts. These include the near-term consequences of heavy downpours as well as the longer-term damages associated with these events. More frequent and intense heavy precipitation events can cause flooding, mudslides, landslides, soil erosion, and result in high levels of soil moisture. These hazards can cause immediate damage during a rainfall event, necessitating emergency response. They also can undermine the structural integrity and maintenance of roads, bridges, drainage systems, and tunnels, necessitating more frequent repairs and reconstruction. The design of culverts and water receiving areas in vulnerable locations may need to accommodate a greater capacity than current design practice. Interestingly, an intense rain event after a period of very dry conditions can cause as much or more damage to assets and services as an intense rain event following a period of very wet conditions. In the first case, the dry ground cannot absorb the water quickly enough and it runs off or pools, while in the second case, the ground is already saturated and the additional precipitation runs off or pools.
Flooding can render a route temporarily impassable, and require maintenance to clear mud and debris. The connectivity of intermodal systems – including goods movement to and from ports - can be disrupted even if short segments of roadways are flooded. Severe precipitation can cause delays in air travel as aircraft are grounded or rerouted. Transportation agencies may need to fortify their emergency management and traffic management capabilities in anticipation of more frequent instances of heavy rainfall and associated response measures.
While these impacts are not new to transportation agencies, the frequency and severity of these problems are likely to increase as the incidence of extreme precipitation events rises. Managing damage and service disruption in real time may take more agency resources and require new communication channels and coordination protocols. Preventive adaptation measures may be considered to increase the resilience of infrastructure (e.g., through design, operational improvements, and/or altered maintenance practices) and to prepare for additional emergency response associated with projected changes in precipitation patterns.
The implications of the precipitation findings detailed in this report on transportation assets and services in Mobile will be investigated in the next task of this study (Task 3: Vulnerability Screen and Assessment).
1 Historical data was provided by Dr. Katharine Hayhoe of Texas Tech, who also conducted the precipitation trend analysis.
2 Daily downscaled projections were provided by Dr. Katharine Hayhoe of Texas Tech.
3 Dr. Katharine Hayhoe provided the historical data (see Hayhoe and Stoner 2012 for a discussion of data quality and additional data filtering).
4 Durrans and Brown
5 Annual precipitation totals that were more than three standard deviations from the mean of the observation record were considered erroneous and removed from the analysis.
6 The full record available was used to compute the "full station record" (see Table 3). An analysis was conducted to determine if the data gaps would affect the long-term precipitation as recommended by Dr. Kelly Redmond of the Western Regional Climate Center. A station with the most complete data record, Fairhope, had existing data removed from its record to replicate the station with the least complete data record, Coden. The removal of these days did not affect the Fairhope station average totals. Hence, it was determined that the full record available for each station, respectively, would provide the best estimated precipitation average totals.
7 The 1961 to 2010 analysis is based on full data except as noted previously and is for a partially complete 2010 year (does not include October, November, and December).
8 Hurricane Katrina is not the most extreme precipitation event in 2005. The measured daily precipitation totals for Hurricane Katrina range from 1.2 inches to 3.25 inches on August 29 and August 30, 2005 across the five observation stations.
9 Durrans and Brown
10 For days in the 1980 to 2009 record with low amounts of precipitation, the local precipitation simulations tend to underestimate observed data by 5% to 15%. For days with the highest amounts of precipitation (i.e., for the 99th percentile of precipitation), the local precipitation simulations tend to overestimate observed data by 20% to 30%. (Hayhoe and Stoner, 2012)
11 The 24-hour exceedance probabilities described in Hayhoe and Stoner (2012) are derived by applying a quantile distribution to the annual duration of maximum 24-hour precipitation events over each 30-year period; hence, those results describe a different analysis than that provided in this report.
12 USCCSP, 2008c
13 Daily precipitation data was used as a substitute for 24-hour precipitation.
14 The observed data represent all forms of extreme precipitation events that affect Mobile, such as mid-latitude storms, tropical storms and hurricanes, and summer-time thunderstorms. Though climate simulations do not capture tropical storms and hurricanes, the statistical downscaling is based on observation data that does. Projections of extreme precipitation events represent an area of large uncertainty.
15 The peak four-day precipitation event identifies longer lasting storms which may be impacted by a strong slow-moving mid-latitude storm. These results are constrained to events occurring within each thirty year period but are statistically robust given the large number of data points used in the analysis. See previous section for methodology description, and Hayhoe and Stoner (2012) for description of the quantile distribution applied to obtain the exceedance probabilities. These results do not reflect annual return periods as that requires analyzing just the maximum event for each year in the time period.
