2. Overarching Issue
• Improve climate resilience: What information will foster
resilience to climate extremes
• Requires a projection of the future
• Traditional assumption of climate stationarity (the past is the best
forecast of the future) is less reliable than ever due to changing
composition of atmosphere
• For some extremes, we may have a basis for a better projection
2
3. Focus
• Estimates of extremes risk are incorporated into the design of
much infrastructure
• Intensity-Duration-Frequency Design Values (e.g. 24-hr, 100-yr storm)
• Probable Maximum Precipitation
• Building design values (e.g. 0.4% temperature exceedance, wind
loading)
• Many billions of dollars of investment annually use these values
in various aspects of design
3
4. Science/Statistics Question
• Can we more confidently estimate the contemporary and future risks
of extremes through the application and/or development of new
statistical methods to explore the spatial/temporal characteristics of
the meteorology and climatology of extremes?
oCan statistics shed light on the physical nature of extremes?
5. Three sub groups
• Precipitation extremes
• Networks and extremes
• Semi-parametric approaches to modeling extremes
5
6. Precipitation Extremes SG
Kenneth Kunkel, presenting
WG members: Daniel Cooley, Whitney Huang, Bo Li, Mark Risser, Brook
Russell, Richard Smith, Michael Wehner
7. • Complex temporal and spatial coherence and variability of extreme
precipitation events –
• Individual thunderstorm cells – hour, a few km
• Thunderstorm complexes – a few hours, tens-100+ km
• Spiral rain bands in hurricanes – a few hours, tens-100+ km
• Low pressure wave – day, 100s of km
• Hurricanes – day, 100s of km
• Synoptic low pressure system – days, 1000+ km
• Hemispheric jet stream wave patterns – weeks, 1000s of km
The Challenge
7
8. Science/Statistics Question
• Can we more confidently estimate the contemporary and future risks
of extremes through the application and/or development of new
statistical methods to explore the spatial/temporal characteristics of
the meteorology and climatology of extremes?
oCan statistics shed light on the physical nature of extremes?
oThe scheduling of the climate theme for the academic year right after
Hurricane Harvey provided a serendipitous focal point for the precipitation
extremes sub group
9. Different spatial views
• Local
oOne or a few stations in a cluster
oHouston area
oWhat is the local risk of another Harvey?
• Regional
oSoutheast U.S.
oWhat is the regional risk of a Harvey-like event?
• National
oWhat is the national structure of extreme trends
10. Different ways of identifying “events”
• Station level
oEach station is analyzed individually first to identify the ”events” and statistical
characteristics at that station
• Spatial averages
oFirst construct area averages of precipitation
oThen identify events and statistical characteristics
oIn this approach, we examined area sizes over a range of approximately
10,000 to 100,000 km2
oThis may provide a superior approach for events related to river basin
flooding
11. Working hypotheses
• Risk of extreme rainfall is correlated with sea surface temperatures
(SSTs)
oGulf SSTs affecting Harvey rainfall risk
• Risk of extreme rainfall is correlated with greenhouse gas
concentrations
12. Hurricane Harvey
• Key facts
oLate August 2017
oMulti-day rainfall exceeded 50 inches in some locations
oThis reached Probable Maximum Precipitation levels
oMassive flooding
• What is the exceedance probability of that event?
• Are the probabilities influenced by SSTs and GHGs?
13.
14. Extreme Precipitation – climate analysis
• Define an overlapping grid of cells separated by 1/10° in lat and long over Southeast U.S.
• Consider all possible 2-degree by 2-degree boxes (~40,000 km2).
• Compute daily precipitation for 1949-present as a simple average of all stations in each
box.
• All boxes that are wholly or partly over water are excluded.
• For each grid box, identify top 5-day precipitation totals.
• Pool everything together and identify the top 100 events for 1949–2017 across the
entire region, eliminating those that overlap in time or space with larger events.
• Also did same analysis on other grid sizes from 1° to 3°
19. Motivating Questions
• How unusual was this event?
• What is the probability of observing another event of this
magnitude in the U.S. Gulf Coast region?
• What is the nature of the relationship between GoM SST and
global CO2 concentrations and precipitation extremes in the
U.S. Gulf Coast region
• How can we account for “storm” level dependence using a
relatively simple spatial model?
19
22. Climate Model Data
• Our analyses of Gulf Storms show dependence on SSTs and CO2
• Two questions requiring climate model data
oHow much has global warming changed current risk of extreme precip
oHow much will future warming change future risk (need to assume emissions
scenario)
26. Quantifying uncertainty in the spatial statistics of
extreme precipitation (M. Risser)
• Developing methods to summarize the statistics of extreme
precipitation for a large network of weather stations (5,000+)
• Address heterogeneity of spatial domain
• e.g., CONUS: variable topography; complex physical systems driving
precipitation (ARs, TCs, MCSs, etc.); seasonality
• Account for “storm dependence” (i.e., the spatial coherence of storm
systems) via the nonparametric bootstrap
• Use GHCN station data to provide spatially-complete maps of extreme
statistics of precipitation
28. JJA SON
DJF MAM
−120 −100 −80 −120 −100 −80
25
30
35
40
45
50
25
30
35
40
45
50
0.5
0.9
1.7
3.0
5.5
9.9
18.0
Gridded 20−year return value standard error, 2010 (mm)
29.
30.
31.
32.
33. Science/Statistics Question
• Can we more confidently estimate the contemporary and future risks
of extremes through the application and/or development of new
statistical methods to explore the spatial/temporal characteristics of
the meteorology and climatology of extremes?
• What have I learned
oHarvey recurrence somewhere in Gulf region is of order 103 years, but
getting substantially more probable with GHG and SST increases
oThere are promising statistical tools to analyze the spatial nature of
extremes and produce more robust and spatially coherent estimates of risks
34. Science/Statistics Question
• Can we more confidently estimate the contemporary and future risks
of extremes through the application and/or development of new
statistical methods to explore the spatial/temporal characteristics of
the meteorology and climatology of extremes?
• What are outstanding issues
oUnderstanding the meteorology behind the statistical results
oHow to reconcile differences in observed climate vs model climate (GoM
SSTs)