Recent NCAR Results on the Impact of Climate Change on the Water Cycle and

1
Recent NCAR Results on the Impact of Climate
Change on the Water Cycle and its Impact on Local
Users
Roy Rasmussen, David Yates, and Peter
Backlund National Center for Atmospheric
Research,Boulder, CO
Funded by the National Science Foundation
2
NCAR Water System Program
How Does the Water Cycle Change as Climate
Changes?
Diurnal Cycle of Precipitation In the Lee of
Topography
Convective Parameterization
1D cloud model
3
of all the social and natural resource crisis
we humans face, the water crisis is the one that
lies at the heart of our survival and that of
planet Earth U.N World Water Development
Report, Water for People, Water for Life
(2003)
4
ESSL
Water System ProgramHistory
Started as an NCAR initiative in 2001 Graduated
to an NCAR Program in 2005

5
The Climate Problem
80 yr. Temp. Rise CMIP 80 yr. Precipitation
Trend ?
Covey et al. 2003
6
IPCC IV Projected precipitation changes in global
climate models over regions where people live
(/- 50 latitude) mostly show low confidence
(less than 66 of the models agree on the sign of
the precipitation change, white region). This is
especially true in northern hemisphere summer and
southern hemisphere winter (red outlines).
Challenge of Predicting Water Cycle Changes under
Climate Change
7
JJA 1996 2002 Time Radar Precipitation Echo
Carbone and colleagues
8
Propagating convection produced by an
interaction of elevated heating (PV anomalies),
environmental shear and CAPE
Noon
Early next morning
Directional shear
MCS cumulonimbus family
Cumulo- nimbus
Mesoscale downdraft
Moister further east
To first order, elevated solar heating
determines start position start time of
traveling convection
1000 km
9
Organized convection parameterization challenge
Ordinary deep convection (conventional
paradigm)
Organized deep convection (new paradigm)

Spans many grid volumes
Single grid volume
10
Key issues for all models and scales

• Are precipitation intensity, frequency, duration,
• sequence and phase right?
• These aspects all have implications for
• soil moisture, runoff and surface hydrology
• and thus for surface feedbacks.
• The timing and duration of precipitation events
  can be examined systematically by exploiting the
  diurnal cycle of precipitation in the warm season
  over North America and extending results to other
  continents.

11
Water Cycle Global Diagnostic Studies
12
Presentation Overview

• Discussion of CCSP report (SAP 4.3)
• Current and Future Climate Changes
• Impacts of Climate Change
• on water resources
• Impact of climate change on
• snowfall over the Rocky Mountains
• Closing Observations

13
Report on The Effects of Climate Change on
Agriculture, Land Resources, Water Resources, and
Biodiversity in the United StatesA Report by
the U.S. Climate Change Science Program (CCSP)
and the Subcommittee on Global Change Research

• CCSP Synthesis and
• Assessment Product 4.3

14
Charge for the Report

• Build on recent assessments in the peer-reviewed
  literature and other appropriate sources. Not
  new scenarios or modeling.
• Focus on the impacts of climate change on
  agriculture, land, forest, water and
  biodiversity.
• Evaluate the importance of climate and other
  stresses and how they are likely to change in the
  future.
• Timeframe of interest weighted toward next 25-50
  years.
• Does not consider adaptive responses.
• Does not provide advice or recommendations.

15
Convening Authors

• Peter Backlund, Director, Research Relations,
  National Center for Atmospheric Research
• Anthony Janetos, Director, Joint Global Change
  Research Institute
• David Schimel, Chief Executive Officer, National
  Ecological Observatory Network

16
Chapter Lead Authors

• Agriculture Jerry Hatfield, Supervisory Plant
  Physiologist, USDA Agricultural Research Service
• Land (Forests) Mike Ryan, Research Forest
  Ecologist, USDA Forest Service
• Land (Arid Lands) Steve Archer, Prof. School of
  Natural Resources, University of Arizona
• Water Dennis Lettenmaier, Prof. Civil
  Environmental Engineering, University of
  Washington
• Biodiversity Anthony Janetos, Director, Joint
  Global Change Research Institute

17
Audience

• Land and resource managers, policy-makers, and
  others interested in relation of climate change
  to agriculture, land resources, water resources,
  and biodiversity
• Organizations and individuals at local, state,
  regional, national, and international levels who
  make ecosystem and resource management decisions
  and establish natural resource policy
• Research scientists who conduct studies of
  climate change impacts on systems and resources,
  and on their potential responses
• State and local governments
• Others who depend on and use the products and
  services provided by systems and resources

18
A Look at Past and Future Climate
19
Multi Model Results for 2030
U.S. Temperature and Precipitation Changes by
2030.
20
Multi-Model Results for 2030
21
Multi-Model Results for 2030
22
Overarching Conclusions (1)

• Climate changes temperature increases,
  increasing CO2 levels, and altered patterns of
  precipitation are already affecting U.S. water
  resources, agriculture, land resources, and
  biodiversity.
• Climate change will continue to have significant
  effects on these resources over the next few
  decades and beyond.
• Many other stresses and disturbances are also
  affecting these resources.

23
Overarching Conclusions (2)

• Climate change impacts on ecosystems will affect
  the services that ecosystems provide, such as
  cleaning water and removing carbon from the
  atmosphere, but we do not yet possess sufficient
  understanding to project the timing, magnitude,
  and consequences of many of these effects
• Existing monitoring systems, while useful for
  many purposes, are not optimized for detecting
  the impacts of climate change on ecosystems.

24
For More Information and Digital Versions of this
Document
Contact Carol Park (park_at_ucar.edu), or visit
Photo courtesy Vanessa Carney

• http//www.climatescience.gov/Library/sap/sap4-3/
  default.php
• http//www.usda.gov/oce/global_change/index.htm
• http//www.sap43.ucar.edu/

25
Observations

• Unfortunately, it appears that many of the most
  severe impacts of climate change in the U.S. are
  occurring and will continue to occur in
  Western regions
• There are numerous adaptation options for
  intensively managed systems water, cities,
  agriculture but many issues around costs,
  policies, trade-offs, etc.
• There are not so many options for lightly managed
  systems, such as wildland areas

26
Water Resources

• Water Dennis Lettenmaier, Prof. Civil
  Environmental Engineering, University of
  Washington

27
Findings Water Resources

• Most of the United States experienced increases
  in precipitation and streamflow and decreases in
  drought during the second half of the 20th
  century. It is likely that these trends are due
  to a combination of decadal-scale variability and
  long-term change.
• Consistent with streamflow and precipitation
  observations, most of the continental United
  States experienced reductions in drought severity
  and duration over the 20th century. However,
  there is some indication of increased drought
  severity and duration in the western and
  southwestern United States.
• Mountain snowpack is declining, and earlier
  spring snowmelt runoff peaks occurring across
  much of the western United States. This trend is
  very likely attributable to long-term warming,
  although some part may have been played by
  decadal scale variability..

28
US Precipitation Change, 1901-2006
29
Projected Temperature and Precipitation Changes
U.S. Temperature and Precipitation Changes by
2030.
30
Findings Water Resources

• Where earlier snowmelt peaks and reduced summer
  and fall low flows have already been detected,
  they are very likely to continue and may have
  substantial impacts on the performance of
  reservoir systems.
• Trends toward increased water use efficiency are
  likely to continue in the coming decades.
  Pressures for reallocation of water will be
  greatest in areas of highest population growth,
  such as the Southwest.
• Declining per capita (and, for some cases, total)
  water consumption will help mitigate the impacts
  of climate change on water resources.

31
Findings Water Resources

• A suite of simulations conducted for the IPCC
  show that the United States may experience
  increased runoff in eastern regions, climate
  simulations gradually transitioning to little
  change in the Missouri and lower Mississippi, to
  substantial decreases in annual runoff in the
  interior of the west (Colorado and Great Basin).

32
High Resolution Simulations of Snowfall over
Colorado and some Climate Impacts
Roy Rasmussen1, Kyoko Ikeda1, Changhai Liu1, Fei
Chen1, Mukul Tewari1, Mike Barlage1, David
Gochis1, Greg Thompson1, David Yates1,Vanda
Grubiic2,Jimy Dudhia1, Kristi Arsenault3, Ethan
Guttman1, and Kathy Miller1 1 National Center
for Atmospheric Research, Boulder, CO, USA 2
University of Vienna, Vienna, Austria 3 George
Mason University, Fairfax, VA, USA
Funded by the NCAR Water System Program (NSF)
33
4th IPCC Model Projections of the Palmer Drought
Severity Index
MOTIVATION

• Snowpack is the most important water source in
  the western US
• Climate models show warm dry conditions in the
  SW US. However, these models do not resolve the
  complex terrain and orographic effects well
• Regional climate models show inconsistent
  snowpack projections

Thank you!

Hoerling (2007)
34
Colorados Headwaters

• Continental-scale river basins whose headwaters
  reside in the Colorado region

• Platte River
• Arkansas River
• Colorado River
• Rio Grande River

(courtesy Col. Div. Water Res.)
Key Questions Will the predicted increase in
snowfall due to a warmer, moister climate be
enough to offset the enhanced melting and
sublimation due to the warmer temperatures? Will
this be sufficient to maintain river flow at
current levels, or is it expected to decrease?
How high resolution of the regional climate model
do we need to answer these questions?
35
January
36
February
37
March
38
April
39
Model Domain
Sub-domain
Full Domain
40

Garvert et al (2007)
41
(No Transcript)
42
Model Domain
Sub-domain
Full Domain
43
Verification with SNOTEL data

• Brooklyn Lake, WY
• SNOTEL site snowgauge

- Site run on batteries recharged by solar
panels on tower - Data transferred once per
day using meteor burst technology - Sensors
at site Snow pillow Snow gauge
Snow depth (ultrasonic)
44
Outline

• Study of impact of varying model resolution
• Ten day model runs at 20, 10, 4, and 2, km
  resolution
• High-res regional current WRF climate model
  simulations driven by NARR boundary conditions
  compared to SNOTEL snowfall.
• Six month snowfall and snowpack simulations for
  an average, above average, and below
  average winter season.
• Climate sensitivity run based on perturbing the
  current runs with a mean climate signal from the
  IPCC4 CCSM A1B simulation.

• Computer Time Award
• 500,000 GAUs on IBM Power 575 (Bluefire) as part
  of the Accelerated Science Discovery competition

45
Ten Day WRFV3 Model Run with NARR23 November
2002 4 December 2002
2 km
4 km
10 km
20 km
46
WRFV3 simulations at 20, 10, 4, and 2 km
horizontal resolutionAverage precipitation at
high elevation SNOTEL sites
25-50 more snowfall at 2 km vs 20 km resolution
10 km
2 km
20 km
Simulations were performed with WRFV3 using NARR
model data for lateral boundary conditions.
Thompson microphysics scheme was used for all the
simulations. Model data points are averages of
values at four grid points nearest to the
individual SNOTEL sites. There were 100
operational SNOTEL sites in the subdomain (shown
right) in 2002.
47
Full Water Year Historical Simulations

• 6 month period from Nov. 1 May. 1
• Four years simulated
• 2001/2002 (dry)
• 2003/2004 (average)
• 2005/2006 (average)
• 2007/2008 (wet)
• Model setup and design
• WRF Model (version 3)
• A single domain 1200x1000 km2 2 km grid
  spacing 45 levels
• PBL scheme MYJ
• Noah land-surface model
• CAM longwave shortwave scheme
• Thompson et al. cloud microphysics scheme
• The 3-hourly, 32-km NARR data provide the initial
  and lateral boundary conditions

48
Nov. 2007-May 2008
Full Water Year Simulation
49
Total Precipitation December 2008
WRF simulation
SNOTEL observations
(mm)
50
Total Precipitation January 2008
WRF simulation
SNOTEL observations
(mm)
51
Total Precipitation January 2008
WRF simulation
SNOTEL observations
(mm)
52
Total Precipitation February 2008
WRF simulation
SNOTEL observations
(mm)
53
Total Precipitation March 2008
WRF simulation
SNOTEL observations
(mm)
54
Total Precipitation April 2008
WRF simulation
SNOTEL observations
(mm)
55
(No Transcript)
56
2 km
18 km
36 km
57
Histogram of percent differences between model
and SNOTEL by grid resolution 2007/2008
2 km
18 km
36 km
58
Comparisons Total Precipitation for 1 Nov.
2007-1 May 2008
18 km
6 km
2 km
36 km
59
Difference in Total Precipitation for 1 Nov.
2007-1 May 2008
2 km - 18 km
2 km - 6 km
2 km - 36 km
2 km simulation has higher snowfall over the
peaks, less in the valleys (No cumulus
parameterization for the simulation results with
18 and 36 km grid resolutions shown here.)
60
Nov. 2001-Apr. 2002
Full Water Year Simulation
61
Nov. 2005-May. 2006
Full Water Year Simulation
62
Nov. 2003-May. 2004
Full Water Year Simulation
63
Future Climate Sensitivity RunMotivationProhib
itively expensive to run high-resolution models
at the desired resolutions for decades to
generate a statistically meaningful future
climate (time-slice method)Approach Add a
climate signal to the current-day high resolution
simulations. Primary impact of the climate
signal is to warm and moisten the troposphere.
Signal derived from future climate model runs.
64
Future Climate Sensitivity Run Setup

• Climate sensitivity run performed using modified
  boundary conditions to the NARR analysis of a
  current water year following the approach by
  Kawase et al. 2008 and Hara et al. 2008 (so
  called Pseudo Global Warming (PGW) approach).
• Modified initial and boundary conditions obtained
  by subtracting the 10 year average monthly
  conditions of 10 2050s CCSM3 A1B scenario runs
  from the average of 10 1990s CCSM3 present
  climate runs averaged over the month and added to
  the NARR initial and boundary conditions from a
  current water year (temperature, water vapor
  mixing ratio, geopotential height, and wind).
• Modified initial and boundary conditions show a
  1.5 C temperature increase over Colorado, and an
  increase of water vapor mixing ratio on the order
  of 10. RH in the simulation similar to the
  control.
• WRF model run using the new boundary conditions
  for high resolution simulations of two full water
  years

65
500mb-Temperature
Current
Future
Future-Current
66
500mb-RH
Current
Future
Future-Current
67
Height and Wind Vector500-mb
Current
Future
Future-Current
68
Sensitivity Run ResultsNovember 2007 May 2008
(wet year)
69
Total Precipitation
70
Model Domain
Sub-domain
Full Domain
71
Precipitation in the CNTRL and 2050 PGW runs at
SNOTEL Sites1 Nov. 2007-1 May. 2008
PGW
ORG
Average Total Precipitation model values are
the average of four nearest grid points from each
SNOTEL site.
72
Average Precipitation in the CNTRL and
sensitivity runsat SNOTEL Sites1 Nov. 2007-1
May. 2008
73
Sub-domain Average of Precip. in the CNTRL and
2050 PGW runs1 Nov. 2007-1 May. 2008
74
Total Precipitation Nov 2007 - MAY 2008
WRF PGW Run
WRF Control Run
PGW Run Control Run
75
Sub-domain Average of Precip. in the CNTRL and
2050 PGW runs1 Nov. 2007-1 May. 2008
76
Precipitation rates 1-2 December 2007
Current Climate Simulation
PGW 2050 Simulation
77
Difference in precipitation rates 1-2 December
2007
PGW - Current Climate Simulation
78
Sub-domain Average of Precip. in the CNTRL and
PGW runs in various elevation brackets
79
Mean precipitation in the CTRL and sensitivity
runs over the sub domain Nov 1 2007 - 1 May 2008
80
Difference in Sub-domain Average of Precip. in
various elevation brackets
81
Domain Average of Precip. in the CNTRL and PGW
runs1 Nov. 2007-1 May. 2008
82
Precipitation rates 1-2 December 2007
Current Climate Simulation
PGW Simulation
83
Total Precipitation Nov 2005 - MAY 2006
WRF PGW Run
WRF Control Run
PGW Run Control Run
84
Temperature at 2 m

• WRF output variable T2 (2D var)

85
Comparison of 2-m Temperature in the
Sub-domain1 Nov. 2007-1 May 2008
(a) Daily mean 2-m temperature (C)
(b) Percent area with T gt 0 oC in the sub-domain
86
Sensitivity Run ResultsNovember 2005 May 2006
(average year)
87
Total Precipitation
88
Precipitation in the CNTRL and sensitivity runs1
Nov. 2005-1 May. 2006
Average Total Precipitation model values are
the average of four nearest grid points from each
SNOTEL site.
89
Average Precipitation in the CNTRL and
sensitivity runsat SNOTEL Sites1 Nov. 2005-1
May. 2006
90
Sub-domain Average of Precip. in the CNTRL and
PGW runs1 Nov. 2005-1 May. 2006
91
Sub-domain Average of Precip. in the CNTRL and
PGW runs in various elevation brackets
92
Difference in Sub-domain Average of Precip. in
various elevation brackets
93
Domain Average of Precip. in the CNTRL and PGW
runs1 Nov. 2005-1 May. 2006
94
Monthly precipitation 2005-2006
95
Mean precipitation in the CTRL and sensitivity
runs over the sub domain 1 May 2006
96
Snow Water Equivalent from SNOTEL compared to WRF
97
SWE in the CNTRL and sensitivity runs at SNOTEL
sites1 Nov. 2005-1 May. 2006
Average SWE model values are the average of four
nearest grid points from each SNOTEL site.
98
Preliminary Results of Pseudo Climate Warming
Simulations for SWE and Runoff

• April 1 snowpack reduced by 25 due to warmer
  climate (increased melting and sublimation)
  despite increased snowfall.
• Runoff likely reduced by a similar amount.

99
Preliminary Results of Pseudo Climate Warming
Simulations for SWE and Runoff

• April 1 snowpack reduced by 25 due to warmer
  climate (increased melting and sublimation)
  despite increased snowfall.
• Runoff likely reduced by a similar amount.

100
Colorado Headwaters Summary

• Comparison of WRF high resolution simulations of
  annual snowfall to SNOTEL observations over the
  Colorado Headwaters regions show very good
  agreement if resolutions below 6 km are used.
• WRF model water year accumulated snowfall at
  SNOTEL sites are within 20 of the observations
  for 80 of the 112 Colorado Headwaters sites
• Some areas of disagreement due to the WRF model
  creating too much snow to the lee of steep
  topography
• Other disagreements under investigation

101
Colorado Headwater Summary

• Comparison of WRF high resolution simulations of
  annual snowfall to SNOTEL observations over the
  Colorado Headwaters regions show very good
  agreement if resolutions below 6 km are used.
• WRF model water year accumulated snowfall at
  SNOTEL sites are within 20 of the observations
  for 80 of the 112 Colorado Headwaters sites
• Some areas of disagreement due to the WRF model
  creating too much snow to the lee of steep
  topography
• Other disagreements under investigation

102
Colorado Headwaters Summary

• High resolution simulations (2 km horizontal) of
  annual snowfall suggests that current global and
  regional model estimates of snowfall at the
  ground (18 km resolution and higher)
  underestimate high elevation snow by 25-50, and
  over estimate low elevation snow fall by a
  similar amount.
• A sensitivity run simulating the impact of
  enhanced atmospheric moisture due to global
  warming from an ensemble of CCSM A1B runs
  centered on 2050 suggests that high elevation
  snowpack increases by 10-25 under this
  scenario.
• Due to increased melting and sublimation, runoff
  is estimated to decrease by 25 (preliminary
  results)
• Future plans
• Improving the hydrology of the model runs,
  including SWE and streamflow
• Applying the verified high resolution model to
  future climate impacts using a Time Slice
  approach (collaboration with NARCAPP)

103
Colorado Headwaters Summary

• High resolution simulations (2 km horizontal)
  of annual snowfall suggests that current global
  and regional model estimates of snowfall at the
  ground (18 km resolution and higher)
  underestimate high elevation snow by 25-50, and
  over estimate low elevation snow fall by a
  similar amount.
• A sensitivity run simulating the impact of
  enhanced atmospheric moisture due to global
  warming from an ensemble of CCSM A1B runs
  centered on 2050 suggests that high elevation
  snowpack increases by 10-25 under this
  scenario.
• Due to increased melting and sublimation,
  runoff is estimated to decrease by 25
  (preliminary results)
• Future plans
• Improving the hydrology of the model runs,
  including SWE and streamflow
• Applying the verified high resolution model to
  future climate impacts using a Time Slice
  approach (collaboration with NARCAPP)

104
Closing Observations

• The big picture is increasingly clear, but
• There is still considerable uncertainty regarding
  fine scale changes within Western regions we
  can gain insight through improved observations,
  experimentation, and modeling
• Successful adaptation will depend on improved
  partnerships among scientists and local,
  regional, state, and federal decision-makers

105
Closing Observations

• Competition for water (between agriculture,
  industry, cities) will be a major factor in
  adapting to climate change throughout the West
• Climate changes at the interface of urban and
  suburban areas and wildlands in the West will be
  challenging to manage
• Many traditional management and planning
  activities that have relied on history as a proxy
  for the future will need to change

106
Integrated Water Resource Management
A Decision Support System for Sustainable Water
Supply Planning WEAP (http//weap21.org)David
Yates
http//weap21.org
107
WEAPs Integrated Hydrology/ Water Management
Framework
Natural Watershed
Pre-development
Water Cycle
Over 100 users across the U.S.
Managed System
Developed Watershed
108
WEAP of the Colo Headwaters
109
Thank you.
110

• Add analysis of the source of the precipitation
  increase
• (snow, rain, graupel, etc.)

111

• Add analysis of the source of the precipitation
  increase
• (snow, rain, graupel, etc.)

112
WRF Simulation with NARR data 2007-2008

• Comparisons between 2, 6, 18, and 36 km grid
  resolutions
• 30 January 2009

113
Comparisons Accum. Precipitation (mm)

• Grid Resolution 2, 6, 18, and 36 km with no
  cumulus parameterization, KF, and BMJ
  parameterizations.
• Time history shows comparisons between average
  SNOTEL observations, PRISM data, and WRF
  simulations at 2, 6, 18, and 36 km. PRISM and
  WRF outputs are average of four grid points
  closest to each SNOTEL site.

2 and 6 km
SNOTEL
18 km
36 km
114
Comparisons Elevations
18 km
6 km
2 km
36 km
115
An example of terrain profile
Stronger updraft at 2 and 6 km
Gradual upward motion over a broad region at 18
and 36 km
Terrain features are smoothed out with 18 and 36
km grid resolution. This has a significant
impact on flow over the mountain and the
resulting precipitation. For example, weak
upward motion occurs over a broad region upwind
of the topographic peak in the 18 and 36 km
simulations as compared with the 2 and 6 km
simulations. This leads to more precipitation
over the relatively lower terrains (valleys) at
coarse resolutions. On the other hand, stronger
upward motions are associated with well-defined
topographic peaks at 2 and 6 km resolutions.
Consequently, more precipitation are received at
higher elevations with the finer grid
resolutions. Cross-sectional analysis of w, etc.
and quantitative analysis of precip diff. vs
elevations will be provided later.
116
Terrain profile (m)
Precipitation over 3 hours (mm)
Surface Temperature (C)
117
Total Precipitation Nov 2005 - MAY 2006
WRF Sensitivity Run
WRF Control Run
Sensitivity Run Control Run
118
Corridors of Convection
Time Radar Precipitation Echo

1 9 9 7
1 9 9 9
July 1998
Carbone and colleagues
119
MM5, NCEP MRF boundary layer surface exchange,
Noah LSM, GSFC microphysics (1-km), 3-km, 10-km,
30-km, 60-km grid-resolution Betts-Miller
convective parameterization, NCEP global analysis
for lateral boundary conditions etc BAMEX 3-10
July 2003
120
Precipitation 3-km grid resolution
Mountains
Plains
121
Airflow organization
Mesoscsale downdraft
C
122
Phase of diurnal cycle
Knievel et al. (2004)