Transcript udel.edu

An Analysis of the Future Climate over the Chesapeake
Bay Watershed
Alexander R.
1
Davies
and Victoria
2
Coles
1. Department of Earth Sciences, Millersville University of PA—Millersville, PA
2. Horn Point Laboratory, University of Maryland Center for Environmental Science—Cambridge, MD.
FUTURE CLIMATE PROJECTIONS
ABSTRACT
The purpose of this study is to analyze the projected climate over the Chesapeake Bay Watershed using the IPCC AR4 models. The A2 and B1 emission scenarios are analyzed
because they bracket a business as usual emissions scenario and a more proactive emissions reduction strategy. The mean model is defined as the average of the individual
models has been shown to yield the most accurate 20th century climate. By removing some poorly performing models, the mean model showed a statistically significant
improvement as compared with 20th century data. With an improved mean model, it is projected that mean annual temperature will increase across the watershed by 4.65
degrees Celsius under the A2 emissions scenario and 2.49 degrees Celsius under the B1 scenario. Additionally, annual precipitation is projected to increase by 5.04% under the
B1 scenario and 7.18% under the A2 scenario with the greatest increases projected for winter and spring. It is possible to speculate that changes in precipitation may alter the
time and duration of the spring freshet. Flooding events and drought conditions may increase as there appears to be an upward trend in the amount of maximum 5 day rain fall,
the number of consecutive dry days, and the number of extreme temperature events—particularly under the A2 scenario
% B1 PR ANOMALIES
and B1 from
TAS Anomalies
TemperatureA2
Increase
Present Day
TEMP ANOM
This study arose out of an NSF funded Research Education for Undergraduates program run by Maryland SeaGrant in partnership with the University of Maryland Center for
Environmental Science. The study took place between June and August 2007.
4
3
2
1
In 1988 the World Meteorological Organization (WMO) and the United Nations Environment Programme founded the Intergovernmental Panel on Climate Change (IPCC) and
gave it the task of providing reliable and non-biased information regarding climate change (IPCC, 2004). Almost twenty years after it was founded, in 2007, IPCC released the
Fourth Assessment Report: “Climate Change 2007” (AR4). AR4 is an expansion upon the findings in the Third Assessment Report (TAR), released in 2001. Both the TAR and
the AR4 include model assessments based on a number of scenarios.
The IPCC Special Report on Emissions Scenarios (SRES) developed 40 greenhouse gas scenarios and grouped them into four “families,” A1, A2, B1, and B2, each with
distinctly different directions for future developments (Nakicenovic et al., 2000). The A2 and B1 scenarios were chosen for this study. These two scenarios represent a range of
emissions paths so the results will be an indication of a business as usual scenario (A2) and a proactive emissions reduction strategy (B1). The A2 scenario shows a continual
and steady increase in emissions through-out the 21st century whereas the B1 scenario features a small increase in emission until mid-century followed by a decrease in
emissions below current level by the end of the 21st century. Using the two scenarios highlights the effect of emissions policies on the emissions placed into the atmosphere and
their impact on anthropogenic climate change over the watershed.
0
1980
2000
2020
2040
Year
2060
2080
2100
A2-BC C R-BC M2.0, Norway
A2-C GC M3.1(T47), C anada
A2-C NRM-C M3, France
A2-GFDL-C M2.1, USA
A2-GISS-ER, USA
A2-INM-C M3.0, Russia
A2-MIROC 3.2(medres), Japan
A2-EC HO-G, Germany/Korea
A2-EC HAM5/MPI-OM, Germany
A2-MRI-C GC M2.3.2, Japan
A2-C C SM3, USA
A2-UKMO-HadC M3, UK
A2-Mean Model
B1-BC C R-BC M2.0, Norway
B1-C GC M3.1(T47), C anada
B1-C NRM-C M3, France
B1-GFDL-C M2.1, USA
B1-GISS-ER, USA
B1-INM-C M3.0, Russia
B1-MIROC 3.2(medres), Japan
B1-EC HO-G, Germany/Korea
B1-EC HAM5/MPI-OM, Germany
B1-MRI-C GC M2.3.2, Japan
B1-C C SM3, USA
B1-UKMO-HadC M3, UK
B1-Mean Model
Figure 3 shows the A2 and B1 scenario temperature anomalies for all 12
individual models and the mean model. The anomalies are in degrees
Celsius
B1 Cons. Dry Days
15
B2-Mean Model
B1-DJF
B1-MAM
B1-JJA
B1-SON
10
5
0
-5
1980
2000
2020
2040
2060
2080
2100
Year
% A2 PRAnomalies
ANOMALIES
% A2 Precip.
20
15
% PR ANOM
6
20
% PR ANOM
7
5
BACKGROUND
% B1 Precip. Anomalies
A2-Mean Model
A2-DJF
A2-MAM
A2-JJA
A2-SON
10
5
0
-5
1980
2000
2020
2040
2060
2080
2100
Year
Figure 4 shows the B1 and A2 Yearly, DJF, MAM, and SON
precipitation anomalies at 2025, 2050, and 2100. The anomalies are in
% change and a negative sign represents a percent decrease in
precipitation. Notice the % anomalies for the winter and spring months
are circled. This time period is critical for the timing and duration of the
spring freshet
A2 Cons. Dry Days
Winter, Spring, and Yearly Snow Cover Anomalies
in 2025, 2050 and 2100
QUESTIONS OF INTEREST AND
HYPOTHESIES
B1 Heat Wave Index
A2 Heat Wave Index
Season
Year (right)
Scenario (below)
2025
2050
2100
Yearly Mean
B1
-40.08
-38.81
-52.95
A2
-23.77
-39.93
-78.91
B1
-34.96
-33.13
-49.44
A2
-20.24
-32.01
-75.74
B1
-56.65
-58.20
-64.81
A2
-32.81
-68.50
-90.78
DJF
1. Can the uncertainty in model projections be reduced by selectively choosing models that accurately simulate the 20th Century? The hypothesis is that
eliminating poorly performing models will yield a more accurate representation of the 21st Century.
MAM
2. What do the models run under different SRES scenarios suggest the physical forcing of the Chesapeake Bay watershed will be in 2025? 2050? 2100? The
hypothesis is that the greatest anomalies will be seen in 2100 with anomalies being significantly smaller in 2025 and 2050. Influenced by the finding in the IPCC AR4
(IPCC, 2007), the prediction would be that air temperature will increase along with the frequency heat waves and spells, and precipitation will slightly increase with a
greater frequency of heavy precipitation events.
B1 5 Day Rain Max
A2 5 Day Rain Max
Table 1 shows the mean yearly, winter (DJF), and spring
(MAM) snow cover anomalies the years 2025, 2050, and 2100
for the B1 and A2 emission scenarios. The anomalies are in %
change and a negative sign represents a decrease in snow
cover. Snow cover was calculated by multiplying the variables
snc and snd for the grid points over the watershed.
REDUCING MODEL UNCERTAINTY
While the purpose of this study is to analyze the climate over the Chesapeake Bay watershed over the next 100 years using models and emission scenarios, first the uncertainty
in the model solutions must be characterized for the region. To estimate some of the model uncertainty we use two techniques; the first is to construct a mean model to better
estimate the future climate and the second is to eliminate poorly performing individual models from the mean model average. The mean model is built by averaging all the data
output from all the individual models that contribute to this study. The reasoning for using the mean model is because it has been shown that on a global basis the mean model
run has been the most effective at simulating 20th century climate (AchutaRao et al., 2004).
To improve the mean model, we test whether eliminate poorly performing models is effective at reducing model-data error. The first step is to analyze how each individual model
performs from the beginning to the end of the 20th century for a number of carefully selected variables. Raw error scores were normalized and weighted then, the final error
score was calculated. The graph below on the left shows the normalized and weighted error scores. Since the mean model is constructed from the individual models, we
hypothesize that removing individual model with high error scores from the mean model will improve its error score. The graph on the right shows that this is exactly what
happened and in-fact , once a threshold was reached (by removing 3 models) the mean model error score began going back up suggesting that removing any more individual
models reduces the accuracy of the mean model.
Weighted Score
10
12
Normalized
and Weighted
Error
TAS+PR
Weighted
Score
IPSL-CM4, France
GFDL-CM2.0, USA
CSIRO-Mk3.0, Australia
UKMO-HadCM3, UK
INM-CM3.0, Russia
CGCM3.1(T47), Canada
GISS-ER, USA
MRI-CGCM2.3.2, Japan
1
CNRM-CM3, France
GFDL-CM2.1, USA
ECHAM5/MPI-OM, Germany
BCCR-BCM2.0, Norway
CCSM3, USA
ECHO-G, Germany/Korea
MIROC3.2(medres), Japan
ALL 15 Models
IPSL- CM4, Franc e
GFDL- CM2.0, USA
Figure 5 shows the extreme induces (those which are calculated on a yearly basis, not a monthly). General trends were only evaluated for these variables and
the mean model is represented by the black line on each graph. The top graphs are the B1 and A2 Consecutive Dry Days anomalies with anomalies in days per
year. The middle graphs are for the Heat Wave Index. The anomalies are in days in which the temperature exceeds 5 degrees Celsius above the average daily
temperature calculated from 1960 to 1990. The extreme temperatures must persist for a minimum of 5 days to be considered under this index. The bottom
graphs are of the maximum amount of precipitation over a five day period and the anomalies are in kg/m2.
DISCUSSION AND CONCLUSION
The average of the IPCC AR4 ensemble is shown to have the lowest error in simulating 20th century climate, and the average can be improved by elimination of
a few model outliers. The high emissions scenario projects an annual temperature increase of 4.65 degree Celsius by 2100 while the lower emissions scenario
projection is 2.49 degree Celsius temperature increase. With that said, it is interesting to note the wide range of individual model projections as seen in Figure
3. The model projecting the lowest temperature anomaly is the GISS-ER under the B1 scenario. It projects a 1.62 degree Celsius annual increase in
temperatures while the MIRCO3.2-medres under A2 conditions projects a 6.16 degree Celsius increase. This represents a 4.54 degree Celsius range of
variability which illustrates the uncertainty and variability in the projections of regional climate change.
CSIRO- Mk3.0, Australia
UKMO- HadCM3, UK
INM- CM3.0, Russia
CGCM3.1(T 47), Canada
GISS- ER, USA
MRI- CGCM2.3.2, Japan
CNRM- CM3, Franc e
GFDL- CM2.1, USA
ECHAM5/MPI- OM, Germany
BCCR- BCM2.0, Norway
CCSM3, USA
ECHO- G, Germany/Korea
MIROC3.2(medres), Japan
Precipitation is important to the region because its implications for water supply and agriculture; however here we consider its effect on stream flow. While there
is only a weak correlation between precipitation and stream flow, the literature suggests enough of a relationship between the two that precipitation can give a
rough idea as to the amount of stream flow entering the bay. When assessing the fresh water influx into the bay, a critical time is during the spring freshet. The
Chesapeake Bay usually receives a peak in fresh water influx roughly around the beginning of April. From Figure 4 it would be reasonable to speculate that the
spring freshet might come sooner in the year due to the increases in precipitation in not only spring, but also winter. Also because of the general increase in
precipitation in both winter and spring, it may also be reasonable to speculate that the freshet will be more gradual instead of having a pronounced peak at some
point in the year. The results shown in Table 1 are also critical for the freshet however a full hydrologic model is need to gain a full understanding.
ALL 15 Models
The conclusion to be made here is that we have a choice. The A2 scenario represents a world where little is done to retard the amount of emissions placed into
the atmosphere. The B1 scenario represents a world where policy changes take place and emissions decrease. As a result, the 2100 projections under the B1
scenario are not as extreme or severe as those under the A2 scenario. Now is the time to make the choice as a global society as to which path we will take.
Mean Model
0
2
4
6
8
10
12
Acknowledgments:
Figure 1 shows the normalized weighted scores each individual model
got along with the control score for the mean model that includes all 15
individual models. As a reminder, the best possible score is zero and
the worst possible score is 14. The models used in the project are
along the left hand side and the colors correspond with the error bars
on the graph.
Figure 2 shows how the error associated with the mean model changed
as individual models were removed from the mean model. The models
were removed in the order based upon their error scores as shown in
Figure 1 to the left.
David Kimmel (UMCES)
Maryland Sea Grant REU Program
Horn Point Laboratory, University of Maryland Center for Environmental Science
National Science Foundation
Sepi Yalda and Richard Clark (MU), Ander Muller (USNA)
MDSG REUs Class of 2007
We acknowledge the modeling groups, the Program for Climate Model Diagnosis and
Intercomparison (PCMDI) and the WCRP's Working Group on Coupled Modeling (WGCM) for
their roles in making available the WCRP CMIP3 multi-model dataset. Support of this dataset is
provided by the Office of Science, U.S. Department of Energy
Works Cited:
AchutaRao K, Covey C, Doutriaux C, Fiorino M, Gleckler P, Phillips T, Sperber K, and Taylor K. 2004. An Appraisal of Coupled Climate Model Simulations. U.S.
Department of Energy and University of California, Lawrence Livermore National Laboratory. Contract W-7405-Eng-48
Intergovernmental Panel on Climate Change (IPCC). 2004 Dec [cited 2007 June 5]. 16 Year of Scientific Assessment in Support of the Climate Convention.
Available from: http://www.ipcc.ch/about/anniversarybrochure.pdf
Nakicenovic N, Alcamo J, Davis G, de Vries B, Fenhann J, Gaffin S, Gregory K, Grübler A, Jung TY, Kram T, Lebre La Rovere E, Michaelis L, Mori S, Morita T,
Pepper W, Pitcher H, Price L, Riahi K, Roehrl A, Rogner H, Sankovski A, Schlesinger M, Shukla P, Smith S, Swart R, Van Rooijen S, Victor N and Dadi Z. 2000
[cited 2007 June 5]. Intergovernmental Panel on Climate Change: Special Report on Emissions Scenarios. Available from:
http://www.grida.no/climate/ipcc/emission/