Annual and seasonal impacts of future climate on water regime:
To obtain data required for future water regime, the CMIP5 multi-model
ensemble LOCA was integrated with SWAT model. We then derived
precipitation, surface runoff, water yield, ET, and discharge for
Bellwood (USGS2361500) monitoring point. We analyzed the data at
monthly, seasonal, and annual scales (Figure 10-13). Table 10 represents
the projected mean annual changes to hydrological components for the
entire simulation period and each decade. Mean annual change to the
discharge at Bellwood station has an increase of 30.45% under moderate
scenario and 29.67% increase under the severe scenarios during the
entire simulation period. Similarly, mean annual surface runoff during
the period has significantly increased with 337.4% and 325.66% under
moderate and severe scenarios, respectively. Water yield also has shown
increase of 18.34% and 18.08% under moderate and severe scenarios
through the entire period. Slight decreases of 0.8% under moderate
scenario and 2.46% under severe scenario were observed to mean annual
ET during the entire simulation.
Table 10 indicates the simulated mean annual changes to water balance
components for mid-century and late-century period. During mid-century
mean annual discharge at Bellwood station was estimated to increase by
24.2% under RCP4.5 and 32.93% under RCP6.0. for late century. However,
the mean annual discharge at the station shows 36.72% increase under
RCP4.5 and 26.4% increase under RCP6.0. Under the severe scenario,
despite the increased urbanization, the discharge amount at the station
is projected to decrease towards the end of the century. It can be
suggested that the last decade of water balance variables has been
affected dramatically (Table 10). Average annual surface runoff during
the mid-century is estimated to increase by 286.3% under RCP4.5 and
315.5% under RCP6.0. Mean annual surface runoff continues to increase
during the late-century by an average increase of 388.5% under moderate
scenario and 335.9% under the severe emission scenarios. These changes
indicate the significant impact of the land use change on water balance
variables. Increases to mean annual water yield were observed under both
scenarios of moderate and severe emissions by 12.55% and 21.42% during
mid-century. During late-century, average annual water yield is expected
to increase by 24.14% under RCP4.5. For RCP6.0, however, average annual
water yield is estimated to increase by 14.73% compared to baseline
period. Increase in mean annual water yield under RCP6.0 compared to the
increase under RCP4.5 is much smaller during the late-century. This is
partially due to the dramatic drop of the hydrologic components during
the 90s. Unlike the other variables, slight decreases occur to mean
annual ET during the mid and late century. For mid-century, a decrease
of 0.63% and 0.01% respectively under the moderate and severe
scenarios, is estimated. During late-century, average annual ET
decreases further with 0.95% and 4.92% under RCP4.5 and RCP6.0,
respectively.
Figure 10 shows the mean annual trend of water balance variables
(surface runoff, water yield, and ET) during the simulation period based
on the models and under the scenarios. The regression line for the
surface under both scenarios indicate the increase during the entire
simulation period. For water yield under moderate scenario the
regression lines for all models are slightly steep. The regression
slope, however, increases under the severe scenario. Annual trend
towards end of the century, shows obvious decrease for ET under sever
emission scenarios.
Basin-wide monthly average of the hydrological components is informative
in investigating the water balance behavior of the watershed. The
hydrological response to projected climate data in UCS changed for each
month. Table 11 and Figure 11 illustrate these changes based on the
models under both scenarios. During mid-century, under moderate GHG
emission the largest changes to precipitation is projected in May
(during flooding season) with increase of 48.5% compared to mean
rainfall in the same month during the baseline period. For January,
February, and March, it increases 18.9%, 22.6%, and 27.4%,
respectively. The largest decrease in rainfall, however, is estimated in
August and October with 23% and 26.7% respectively (RCP4.5,
Mid-century). During the late-century under RCP4.5, the largest increase
and decrease are projected in March with 36.1% and in August with
26.1%, respectively. For the severe emission scenario, the largest
rainfall increase and decrease during mid-century are in January
(41.8%) and December (21.1%), respectively. For the late century,
however, precipitation decreases overall. The largest increase is
expected to happen in September (29.9%) and the largest decrease in
October (39.2%). Overall, August, October, and December are expected to
be drier and January, March, and May are expected to be significantly
wetter through the entire simulation period under both scenarios (Table
11 and Figure 11). Similar to mean annual behavior of surface runoff,
monthly estimates are also projected to increase dramatically. Under
both scenarios, June has the highest increase of up to 5 times baseline
period in monthly basin-wide surface runoff. The second highest increase
is expected to happen in January with up to 4 times of the baseline
period. The smallest increases in monthly mean surface runoff is
projected in December with the lowest increase under RCP6.0 during
late-century (36.2%) (Table 11 and Figure 11). Monthly behavior of
water yield amount differs from rainfall and surface runoff. Under
RCP4.5, overall water yield is projected to be higher than that of
mid-century. The largest changes under the moderate scenario is
estimated in February (+58.6%) and March (+62.1%) during late century
and in June (+42.9%) and July (+44.4%) during mid-century (RCP4.5).
Under the moderate scenario water yield is estimated to decrease during
mid-century in December by 12.4% and in August by 6.3%. Mean water
yield for each month indicates different behavior under the sever
emission scenario than expected. Under RCP6.0 and during mid-century,
January and June, and October have the increase of 58.4%, 62.7%, and
53.7%, respectively. Through the late-century, however, November has
the largest increase of 99.4% in water yield amount. Under RCP6.0,
water yield decreases only in December though the entire simulation
period (Table11 and Figure 11). Overall ET is expected to decrease
slightly. For all months mean ET drops except for April and May. Under
both scenarios the largest decrease is estimated in November with close
to 20% drop. From the Figure 11, one can notice the level of decrease
in ET during Summer and Fall. Monthly discharge projection is shown in
Figure 12 and Table 11. Under both scenarios discharge is estimated to
decrease in April and December at Bellwood station with the largest drop
of 31.2% in December during the mid-century under sever emissions. In
other months, discharge is expected to increase. During the mid-century,
the largest increase is observed in July (106.7%) and September
(115.7%) under RCP4.5 and RCP6.0, respectively. During the
late-century, February, July, and November have almost the same
discharge under the moderate emissions. For the severe emission
conditions, however, September and November have the highest increase of
133.7% and 186.3%, respectively. The discharge projections indicate
increases in the months in which rainfall is expected to decrease. This
can be attributed to the land use change.
Seasonal variations are also expected in hydrological response to future
climate data. Interquartile range (IQR) can be used to express this
variability (Figure 13). Outliers can be attributed to extreme weather.
Also, larger IQRs indicate more frequent severe weather. For Spring
precipitation during the simulation period, under moderate emissions,
the largest IQR ranging from -27.1% to 46.9% is projected at the end
of mid-century (70s); under sever emissions the largest IQR ranging from
-40.4% to 34.8% is estimated at the beginning of the mid-century
(40s). Under RCP4.5 the medians for percent changes increase through the
mid-century. Through the late-century, however, the medians for each
decade and their IQR show modest changes. This means the most frequent
extreme rainfall in Spring is expected during 60s under moderate
emissions. These extreme behaviors, however, is expected 2 decades
earlier (40s) under the severe emissions. Under RCP6.0, the medians
increase from 10.2% to 33.2% during the mid-century. Under RCP6.0,
medians drop at the start of the late-century and then fluctuate during
the last century meaning that Spring rainfall peaks in 80s with the same
frequency behavior. Under moderate emissions 50s has the wettest Spring
and 80s has the driest. Under severe emissions, 60s and 70s have the
wettest and driest Spring, respectively (compared to the baseline
period). IQR ranges for Summer rainfall are noticeably smaller than that
of Spring, meaning smaller changes for Summer rainfalls. Overall IQR
under RCP6.0 is longer indicating more changes (decrease and increase)
under severe emissions. Moderate emissions, however, reflects more
outliers indicating greater likelihood for heavy rainfalls during
Summer. Under RCP4.5 the largest IQR ranging from -21.9% to 28.8%
occurs at the start of late-century (70s), making the decade with
wettest Summer (compared to baseline) under moderate emissions. The
driest Summer is expected to happen at the beginning of the midcentury
under RCP4.5. Under RCP6.0 the largest IQR ranging from -13.2% to 29%
is expected at the end of mid-century (60s) making the period with
wettest Summer compared to the baseline. Under RCP6.0, UCS is expected
to experience the driest Summer in the middle of late-century period
(80s) with the IQR ranging from -24.9% to-8.2%. Similar to Summer
rainfall, Fall rainfall also has larger overall IQR under RCP6.0 than
RCP4.5, indicating more changes (decrease and increase) under sever
emissions.
Under RCP4.5, very small changes are projected to happen to medians
during Fall. Under RCP6.0, however, medians increase during the entire
simulation period after large drop in the start of the mid-century.
Under RCP4.5, the largest IQR is projected during 50s ranging from
-23.3% to 34.9% change compared to baseline mean Fall. 70s also shows
the same variations. However, the smallest IQR ranging from -19.5%
to-5.1% (under RCP4.5) is estimated during 60s where change variations
indicate noticeable number of outliers meaning, strong Fall
precipitations compared to baseline period. Under severe emissions, the
largest IQR for Fall precipitation occurs in the end late-century (90s)
ranging from -12% to 51.2%. This variation is followed by 50s IQR
ranging from -47% to 10.6%. Big portion of the 90s’ IQR indicates
increase, but for 50s a decrease is observed. This makes the end of the
late-century to have wettest Fall and middle of the mid-century to have
the driest Fall (under RCP6.0). Early late-century is projected to have
the smallest range changes (IQR ranging from -15.7% to 12.2%) under
RCP6.0. Projected Winter rainfall has the shortest IQRs (smallest
changes) compared to the other three seasons (both emission scenarios).
For both scenarios, during the entire simulation period, IQRs and
medians fall below zero line, meaning decreased amount of precipitation
during Winter. More outliers under RCP4.5 indicates more extreme changes
under moderate than sever scenario during Winter. Under RCP4.5, medians
increase towards the end of century, starting from -21.7% and ending at
-1.9%. This indicates overall less changes towards the end the
simulation period. Under RCP4.5, the largest IQR for Winter rainfall is
projected during the late-century (80s) ranging from -32.9% to 1 8%
with the median of -27.7% (close to the 25thpercentile). Under moderate emission the smallest variation is projected
to be during the end of the late century (90s) with IQR ranging from
-16.5% to 1.2% (median of -1.9%). Under sever emissions, however, end
of the late-century period has the largest IQR ranging from -7.9% to
38.7% with median close to 25th percentile, making
the decade wettest for the Winter rainfalls. The smallest variations of
Winter precipitation are estimated to occur at the start of the late
century (70s). the direst Winter under severe emissions is projected to
end of the mid-century (60s) where the median is -21.5% (figure 13).
Comparing the percent change variations for projected precipitation
under both scenarios shows the largest changes in Summer and Winter when
switching from RCP4.5 to RCP6.0. similar results were observed by Sunde
et al. (2017). Surface runoff variations under future climatic data,
shows dramatic changes (figure 13). For all seasons and under both
scenarios, surface runoff is projected to increase up to several folds.
For Spring surface runoff, under RCP4.5, the largest IQR is expected by
end of the mid-century (60s) ranging from 76.9% to 526.5%. The second
largest variation is estimated in middle of the late-century period
(80s) ranging from 155.22% to 601.7% (under RCP4.5). The smallest
variation (IQR) under the moderate emissions, however, is observed in
beginning of the mid-century ranging from 229.4% to 330.9%. The
medians change barely through the entire simulation period. Spring
surface runoff, under RCP6.0, has overall length of IQR shorter than the
that of RCP4.5 indicating less variations in percent changes except for
beginning (40s) and end (80s) of the simulation period. The largest IQR
under the severe scenario is estimated for the end of the late-century
period (90s) with variations ranging from 278.8% to 731.2%. The
smallest IQR, however, is estimated during the end of the mid-century
(60s) ranging from 272.4% to 517.3%. Medians, under the RCP6.0, follow
the general pattern of increasing towards the end of the century. Figure
13 indicates the expected difference between moderate and severe
emissions. Comparison of Spring surface runoff under both scenarios
shows significant differences during the beginning of the simulation
where moderate emissions make shortest IQR while severe emissions make
one the largest variations. Figure 13 indicates more outliers are
projected under moderate emissions. The variations for Summer surface
runoff can be as large as Spring’s.
Through the entire simulation period the overall IQR for Summer surface
runoff under RCP6.0 is larger than that of Rcp4.5 indicating more
variations under severe emissions. Under RCP4.5 medians increase towards
the end the century with the highest of 465% during 80s and the lowest
of 221% during the 50s. under RCP4.5, the largest IQR is projected for
the middle of the mid-century (50s) period and the shortest IQR for
Summer surface runoff is estimated for the beginning pf the late-century
period (70s). Summer surface runoff under sever emissions has higher
variations. The largest IQR, under RCP6.0, is estimated during the end
of the late-century period ranging from 207.6% to 615.6% and the
shortest IQR ranging is projected during 50s ranging from 146.7% to
342%. For Fall surface runoff variation, under both emission scenarios,
the medians follow the increasing pattern towards the end of the century
(slight increase under RCP4.5 and more accelerated increase under
RCP6.0). Under severe emissions, however, more outliers are observed
indicating more extremes. The overall length for IQR varies through the
simulation for both scenarios. The percent change values for Fall are
also as high as Spring’s and Summer’s indicating several fold surface
runoffs during Fall too. Under RCP4.5 the largest IQR is projected at
the beginning of the late-century period (70s) ranging from 138% to
510%. The shortest IQR, however, is observed for a decade after that
during 80s. Under RCP6.0, the largest and shortest IQR ranging from
44.7% to 434.4% and from 239.9% to 387% is estimated during the
beginning (40s) and the end (60s) of the mid-century period,
respectively. For Winter surface runoff the overall length of IQR under
moderate emissions is greater than that of sever emissions indicating
more variation compared to the baseline period. Under both scenarios,
medians follow slight increase pattern towards the end of the simulation
period. Under RCP4.5, the largest IQR ranging from 133% to 605.1% is
projected at the middle of the late-century period (80s) and the
shortest IQR (smallest variations) is estimated for 50s. Under RCP6.0,
however, the largest variations (IQR) ranging from 103% to 349% is
expected during the beginning of the mid-century period, and the
shortest IQR is observed during the middle of the late-century period.
Water yield percent change variations from figure 13 indicates both
increasing and decreasing amounts through all seasons.
Under moderate emissions, medians appear to follow slight decreasing
pattern towards the end of the century and under sever emissions the
pattern turns into slight increasing trend by end of the simulation
period. Since the surface runoff and water yield are related, water
yield follows the same pattern as the surface runoff does (Arnold,
Kiniry et al. 2013). For Spring water yield, under RCP4.5, the largest
IQR ranging from -38.2% to 68.6% is projected to happen during the
middle of the mid-century period and the shortest IQR ranging from -14%
to 9.7% is observed during the start of the mid-century indicating the
smallest changes of water yield amount compared to the baseline period.
While the beginning of the simulation period has the smallest variations
of water yield under RCP4.5, severe emission projects largest IQR
ranging -50.2% to 32.9%. Under RCP6.0, the shortest IQR ranging from
-0.5% to 37.2% is observed during 70s. For Summer water yield,
variations under RCP6.0 appears to be greater. This discrepancy is more
obvious in the beginning and at the end of the simulation period. Under
RCP4.5, the largest IQR ranging from -28.7% to 38.8% is projected at
the beginning of the late-century period (70s). The shortest IQR (from
-15% to 7.2%), however, is expected at the beginning of the
mid-century period (40s). RCP4.5 has resulted in more outliers than
RCP6.0 indicating higher chance of extreme amounts. Summer water yield,
under sever emissions, shows longer IQR and less outlier. The largest
IQR is observed during the middle of the simulation period (60s and 70s)
ranging from -23% to 62%. The smallest variations (-16.5% to 24.7%),
however, is estimated during the middle of the late century. For Fall
water yield variations, medians under both scenarios show mild changes
towards the end the century. The overall length of IQR under RCP6.0 is
longer than that of RCP4.5, indicating more variations under severe
emissions. The largest IQR under RCP4.5 is estimated during 50s (ranging
from -24.6% to 63%). The shortest IQR, however, is observed during 80s
with ranging from -21.9% to 14.21%. Under severe emissions, the
largest variations are projected during 50s (middle of the mid-century
period) ranging from -62.2% to 30.4%. The shortest IQR is estimated in
the beginning of the simulation period (40s) with ranges of -10.2% and
32.9%. Winter water yield projections indicate decrease during the
mid-century period and slight increase by end of the century under both
emission scenarios. Under RCP4.5 medians increase slightly by end of the
century. Under RCP6.0, however, the trend shows no change. Under RCP4.5,
the largest and the shortest IQR ranging from +41.8% to 23.6% and from
-18.8% to 4.3% are estimated during 80s and 90s, respectively. During
Winter, water yield for late-century period, RCP4.5 projects wide
variations while RCP6.0 projects small changes compared to the base
line.
ET has the smallest changes and variations compared to other hydrologic
variables during all seasons. Positive changes of ET during Spring show
general increase. Through the mid-century period ET decreases (RCP4.5)
while during the late-century it increases towards the end of the
period. Under RCP6.0 during the late-century decreases are observed
while RCP4.5 shows ET increase during the period. Under RCP4.5, the
largest and shortest IQR ranging from -0.7% to 8.5% and from -1.5% to
3.3% are estimated during 90s and 80s, respectively. Under RCP6.0,
however, the largest (ranging from -3.8% to 6.7%) and the shortest
(ranging from 0.9% to 4.7%) IQR are observed in the beginning of the
simulation period (40s and 50s). During Summer, ET is expected to
decrease always under RCP6.0. Since the medians also decrease towards
the end the century, it indicates accelerated decrease with approaching
to the end the simulation period. Under RCP4.5, however, Et increases
sometimes. Under RCp4.5 medians follow slight decreasing pattern. The
largest (ranging from -4.6% to 3.9%) and the shortest (ranging from
-1.99% to -0.02%) IQR is projected during middle of the late century
(80s) and middle of the mid-century (50s) respectively. Under RCP6.0 the
largest (ranging from -9.2% to 0.5%) variation is expected during 60s
and smallest (ranging from (ranging from -4.9% to -1.6%) variation of
ET compared to baseline is estimated at the beginning of the simulation
(40s). ET during Fall shows decrease for both scenarios and for the
entire period. The projections indicate that the largest decrease is
estimated during Fall and under RCP6.0 up to -23.5% during 50s. During
Fall, the overall length of IQR under RCP6.0 is longer than that of
RCP4.5 indicating more extreme variations under severe emissions. Under
RCP6.0 the largest (ranging from -22.3% to -3.7%) and the shortest
(ranging from -19.1% to 11%) IQR is estimated during 60s and 70s,
respectively. Projection for ET during Winter follows the same pattern
as Summer. Under moderate emissions the medians during the mid- and
late-century increases. The overall length of IQR under RCP4.5 is
greater than that of RCP6.0 meaning more variation under moderate
emissions. Under sever emission decreases are estimated during the
late-century while moderate emission projections show lesser decrease or
increase. Under RCP6.0 the largest (ranging from -2.1% to 11.3%) and
the shortest (ranging from -0.9% to 1.3%) IQR are observed during
mid-century in the middle (50s) and the end (60s) of the period,
respectively.