-
Integrating models and observations to understand the hydrology
and water quality impacts from beetle-impacted watersheds
Colorado School of Mines, Colorado State University
Lindsay Bearup, Nicole Bogenschuetz, Brent Brouillard, Stuart
Cottrell, Mike Czaja, Eric Dickenson, Nick Engdahl, Mary
Michael Forrester, Jennifer Jefferson, Andrew Maloney,
Katherine Mattor, Reed Maxwell, John McCray, Kristin Mikkelson,
Adam Mitchell, Alexis Navarre-
Sitchler, Josh Sharp, Colgan Smith, John Stednick students,
postdocs, faculty
-
Quantifying and predicting the impacts of land cover change
presents an interesting challenge in hydrology
Loss + Gain Forest Tree
Cover >80% 0%
Hansen et al Science (2013)
-
Temperature and insect-driven tree mortality is increasing
Edburg et al FEE (2012) Williams et al NCC (2013)
Forest drought stress has increased, increasing beetle
infesta>ons and tree mortality NATURE CLIMATE CHANGE DOI:
10.1038/NCLIMATE1693
ARTICLES
r = 0.83 2
1
0
1
2
1980 1990 2000Year
2010
Wild
fire
area
(km
2 )
r = 0.84
r = 0.82
Bark
-bee
tle a
rea
(km
2 )
1
0
1
PionPonderosa pineDouglas-fir
11 0
5
10
15
20
25
1
0
1
110
1001,000
10,000
2
0
2
10
100
1,000
10,000
2550%
75
a
b
c
d
0.35
0.40
0.45
ND
VI
Perc
enta
ge d
ead
FDSI
FDSI
2-yr FDSI
6-yr FDSI
Figure 2 |Measurements of forest productivity and mortality
overlaid onthe FDSI (red, right y axis). a, The annual average
late-June toearly-August NDVI calculated from satellite (19811999:
AVHRR,20002012: MODIS) imagery. b, Annual forest inventory and
analysismeasurements of the percentage of standing dead trees in
the SWUS forthe three most common conifer species. Error bars
represent standarddeviation of the percentage dead when each years
forest inventory andanalysis measurements are randomly resampled
1,000 times(Supplementary Information). c, Aerial-survey-derived
estimates of thearea where 10 trees per acre were killed by
bark-beetle attack.d, Satellite-derived moderately and severely
burned forest and woodlandarea in the SWUS. See Supplementary
Information and SupplementaryFig. S4 for methods to calculate
burned area. The inset shows thepercentage of years within a given
FDSI class that were top-10% fire-scaryears during AD 16501899 (the
horizontal line is at the expectedfrequency of 10%, bins are 0.25
FDSI units wide). In all panels, the FDSIvalues for 20082012 (open
red squares) were estimated by applyingclimate data to equation
(1). Note the inverted y axes for the FDSI in bd.
drought has been, and remains, a primary driver of
widespreadwildfires in the SWUS.
Given the exponential relationships established between theFDSI
and tree mortality, severe drought events before the observedrecord
probably coincided with widespread tree mortality. Asobserved
climate and mortality data are unavailable for much ofthe past
millennium, we use the FDSI record to identify othersevere drought
events likely to have caused widespread mortalitysince ad 1000
(Fig. 3). A drought event is defined as any periodgreater than
three years when the mean FDSI is negative, the FDSIis not positive
for two consecutive years34, and the FDSI is lessthan two standard
deviation units below the 18962007 mean forat least one year.
Drought-event strength is the sum of the FDSIvalues during the
event. Updating the FDSI for 20082012 withthe FDSI values estimated
from equation (1), three drought eventshave occurred within the
observed climate record: the presentdrought (20002012, the fifth
strongest since ad 1000), 19451964(the sixth strongest) and
18991904 (the seventeenth strongest;Fig. 3). The prolonged 19451964
event was indeed associated withextensive treemortality in the SWUS
as indicated by documentation
FDSI
1.0
0.5
0.0
0.5
1.0
1000 1200 1400 1600Year
1800 2000
Figure 3 | Eleven-year smoothed FDSI for AD 10002012. Black
area: 95%confidence range of the FDSI, representing the range of
FDSI valuesexpected if all 335 chronologies were available.
Vertical grey areas highlightdrought events.
of bark-beetle outbreaks30,35, anomalously large wildfires31,32
andwidespread die-off of conifers30,31,35. The 18991904 drought
wasalso associatedwith forest declines36, although little
documented.
Before the 1900s, the 15721587 event was the most recentevent
exceeding the severity of the present event (Fig. 3).
Thismegadrought event37,38 ranks as the fourth most severe sincead
1000 and the most severe since 1300. Although direct
mortalityobservations are not available for the 1500s event,
studies of forestage structure document a scarcity of trees on
todays landscapethat began growing before the late 1500s (refs
13,31). As lifespansof SWUS conifers often greatly exceed 400
years, the scarcity oftrees preceding the 1500s event indicates
that intense droughtconditions probably led to deaths of a large
proportion of treesliving at the time. Before the late 1500s, the
correspondence betweenrecords of conifer pollen buried at
archaeological sites and tree-ringwidths39 suggests that widespread
tree mortality (indicated bypollen) co-occurred with massive
droughts in the 1200s (indicatedby tree rings). Notably, ancient
Puebloan populations and land-usepractices were in great flux
during this time, confounding theattribution of a dominant cause of
the 1200s forest decline40.
Future forest drought-stressThe ongoing VPD-dominated drought
event (Fig. 4a) is consistentwith climate-model projections (phase
3 of the Coupled ModelIntercomparison Project (CMIP3)) of
increasing warm-seasonVPD (3.6%decade1) throughout the rest of the
twenty-firstcentury in response to business-as-usual greenhouse-gas
emissionsscenarios41 (SRES A2; Fig. 4a and Supplementary Fig. S6
foralternative emissions scenarios: SRES A1B and B1).
Dynamicallydownscaled (0.5 geographic resolution)model projections
indicatesimilar increases in the VPD (Fig. 4a and Supplementary
Infor-mation). Furthermore, most models project a slight decrease
incold-season precipitation during the second half of this century(
1.25%decade1, Fig. 4c). Applying model projections toequation (1),
all models indicate negative FDSI trends throughoutthe twenty-first
century (Fig. 4d). By 2050, the ensemblemean FDSIis consistently
more severe than that of any megadrought since atleast ad 1000
(megadrought conditions are surpassed by 2070 in themost optimistic
B1 emissions scenario, Supplementary Fig. S6d).Notably, projections
of the FDSI are more severe than projectionsof gross water balance
(precipitationevaporation) because water-balance projections are
influenced more by decreased cold-seasonprecipitation than by
increasedwarm-seasonVPD (ref. 15).
FDSI projections suggest that SWUS forest drought-stress
isentering a new era where natural oscillations such as those
apparentin Fig. 3 are superimposed on, and overwhelmed by, a
trendtowards more intense drought stress. As the VPD diverges from
therange of observed variability, nonlinear effects may alter
droughtimpacts on forests (for example, Fig. 5 in ref. 42). During
theobserved record, equation (1) was a better predictor of the
FDSI
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange 3
SL Edburg et al. Bark beetle-caused tree mortality
The Ecological Society of America www.frontiersinecology.org
biogeochemical impacts include reductions in plant Cuptake,
increases in decomposition, and potential loss ofnutrients. An
example of coupled biogeophysical andbiogeochemical processes is
the influence of canopy struc-ture (leaf area and stem density) on
the amount of precip-itation captured by the foliage (and therefore
on soil mois-ture), the effects of soil moisture on soil
decompositionand plant growth, and the interaction between soil
nutri-ents, decomposition, and plant growth (Figure 2).
Biogeophysical and biogeochemical impacts followingbark beetle
infestation have the potential to severely affectboth natural
resources and economic values. For example,snow from mountain
ecosystems is the major source ofwater for more than 60 million
people in the western USand Canada (Bales et al. 2006); changes in
forest structurefollowing bark beetle epidemics alter the amount,
timing,and partitioning of this resource (Rex and Dub 2006;Pugh and
Small 2012). Post-insect-infestation tree mortal-ity also affects C
and N cycling in forests. Although mostof these forests are net C
sinks (eg Schimel et al. 2002),insect-related disturbances may
cause them to release C tothe atmosphere (Kurz et al. 2008).
Nutrient cycling withinaffected forest ecosystems will also be
modified, withreduced plant uptake increasing water and nutrient
export.As a result, the aggregate impact of insect outbreaks
mayhave consequences for regional and global weather and cli-mate
systems as well as for water supply and C storage.
Here, we present a chronological model of ecosystemimpacts to
help inform future management decisions andto identify future
research areas that will improve under-standing of insect-related
disturbances. Our model focuseson the characteristic time scales of
a mountain pine beetle(Dendroctonus ponderosae) outbreak in
lodgepole pine(Pinus contorta Douglas var latifolia) forests
(Figure 2),beginning in the initial days and weeks after
infestation(Stage 1; Figure 3a), proceeding through a phase in
whichneedles turn red in the months to years following the
out-break (Stage 2; Figure 3b), to the gray phase that occurs
asneedles fall off dead trees within 35 years following
attack(Stage 3; Figure 3c), and finally to tree regeneration
andsnagfall in the decades following the outbreak (Stage 4;Figure
3d). Pine stands that are affected by mountain pinebeetle
infestations are typically dominated by lodgepolepines (> 80% of
the stem density), although spruce (Piceaspp) and fir (Abies spp)
are also found therein. Understoryvegetation may be extensive in
some stands (eg Brown etal. 2010). We anticipate that our
conceptual model willprovide a framework for future investigations
of theimpacts of bark beetles on forest ecosystems.
! Coupled biogeophysical and biogeochemicalimpacts
Stage 1: green attack (days to weeks)
Mountain pine beetles preferentially infest and kill
largerdiameter host trees (pines), leaving smaller diameter
trees
and understory vegetation unaffected (Shore andSafranyik 1992).
Beetles introduce blue stain fungi(Grosmannia clavigera) into tree
xylem, which decreaseand eventually prevent water transport (Paine
et al.1997). Sap flux studies have shown that a drop in
transpi-ration occurs within one month of infestation and thatthe
rate of change is dependent on fungal virulence(Yamaoka et al.
1990). This finding contrasts with exper-imentally girdled trees
(simulating beetles feeding onphloem) that took up to five growing
seasons to die,whereas those inoculated with blue stain fungi died
inone growing season (Knight et al. 1991).
The initial impact of mountain pine beetle-inducedmortality on
lodgepole pine trees is hypothesized to occurin three stages.
First, water transport in the stem shutsdown, which results in the
same response of stomatal clo-sure as tree response to drought.
Stomatal conductanceand plant hydraulics are closely coordinated
(Ewers et al.2007). Second, a drop in stomatal conductance leads to
a
Figure 1. Areas affected by bark beetles from 19972010 (inthe
western US) and 20012010 (in British Columbia,Canada). Color of
each grid cell represents the last year thatbark beetle damage was
observed by aerial surveys.
-
The Mountain Pine Beetle (MPB) is an endemic species
(Dendroctonus ponderosae)
5mm
Green Red Grey
Year
Sin
ce A
ttack
4 3
2
1
0
Summer Fall Winter Spring 2nd Summer
Attacking Brood Adult Egg Larva Larva Pupa Adult
(Figure modified from Wulder et al 2006) (Figure modified from
CSFS 2013)
-
Climate drivers lead to unprecedented infestation
Warmer temperatures and longer habitable summer seasons have
lead to reproductive doubling -Mitton & Ferrenberg (2012)
Drought conditions weaken tree defenses and correlate with
infestation. -Williams et al (2013)
-
Grand Lake, Colorado
-45
-40
-35
-30
-25
-20
-15
-10
1940 1960 1980 2000 2020
Min
. Tem
pera
ture
(Nov
.-Mar
., C
)
Monthly Minimum Temperature
Climate drivers lead to unprecedented infestation locally in
Rocky Mountain National Park (RMNP)
-
How might this impact water?
compared to bark beetle infestation are tree harvestand fire.
Adams et al. (2011) conducted an extensive
review on the ecohydrological similarities and differ-
ences between tree mortality associated with canopydie-off,
forest harvesting practices and fire and
highlighted below are some of the different responses
that are important to distinguish. One of the maindifferences is
that in high severity fires and many tree
harvesting practices, there is entire loss of theoverstory
canopy, while in bark beetle infestations
mortality is not necessarily continuous across the
entire watershed. In general, fire is not as similar tobark
beetle infestations as tree harvesting, as fire
completely alters ground surface vegetation and soil
surface properties. Forest harvesting is similar in manyways to
bark beetle infestations, inasmuch as both
disturbances result in diminished canopy and reduced
vegetative nutrient and water uptake; however, forestharvesting
also often includes soil disturbance, soil
compaction and road construction and maintenance,
all of which contribute to differences in hydrologicaland
biogeochemical processes. Beetle-induced tree
mortality also occurs in phases, with the tree initially
losing its ability to take up water and nutrients,followed by
needle discoloration and several years
later actual needle drop. This slower transitioning
results in less stark changes to forest biogeochemistry,as soil
buffering and surviving vegetation can often
compensate lowering nutrient export and observedhydrological
changes (Griffin et al. 2013). In compar-
ison, forest harvesting occurs on a much shorter
timescale with complete canopy removal associatedwith immediate
tree death. With the above highlighted
differences between bark beetle infestations and other
land disturbances, it becomes apparent that even
withsimilarities we can draw upon from the extensive pool
of literature on non-bark beetle canopy-changing
disturbances, it is necessary to review and synthesizethe
impacts bark beetle infestations can have on
hydrological and biogeochemical processes.
Fig. 1 A conceptual image depicting the continuously chang-ing
hydrologic and biogeochemical cycles during the variousphases of
bark beetle infestation. The top portion of the figurecontains the
visual representation of the three primary phases ofinfestation,
with the accompanying elements of the hydrologiccycle. Fluxes are
denoted with arrows and storage reservoirs asrectangles with the
associated increase or decrease in theprocess depicted by the fill
departure above or below the N.S.C.(no significant change) line.
While differences in changes ofeach variable have been observed due
to catchment character-istics, climate and infestation
characteristics, the filled-inportion displays the general trend
even though magnitude may
vary. T transpiration, E ground evaporation, I interception,
SWEsnow water equivalent, h soil moisture, A ablation and Q
wateryield. The bottom of the figure displays the temporal
trendsassociated with the different phases of infestation and
theexpected alterations in biogeochemical cycling. An up
arrowindicates concentrations above baseline, a down arrow
indicatesconcentrations below baseline and a horizontal dash
indicatesno significant change (with the size of the arrow
indicatingmagnitude). Asterisks(*) indicate this trend depicts the
majorityof results published, although occasional studies have
notobserved this trend. See Table 1 and 2 for
additionalclarification
Biogeochemistry
123
Mikkelson, Bearup, Maxwell, Stednick, McCray, Sharp,
Biogeochemistry 2013
Green Red Grey
-
To address hydrologic responses to stress we need integrated
tools that can evaluate managed natural systems
P. Dll et al. / Journal of Geodynamics 59 60 (2012) 143 156
147
to a lack of data on drainage in irrigated areas in many
countries, wehad to combine data on drained irrigated area with
data on drainedarea (without distinction of rainfed and irrigated
agriculture). Val-ues of frgi between 0.2 and 0.4 occur in regions
where irrigatedareas are strongly drained: the Nile in Egypt, the
southern part ofthe Indus, Japan, Philippines, Indonesia and parts
of Australia. Thenorthern part of the Indus as well as Northeastern
China show val-ues between 0.4 and 0.6. In most of the USA, frgi is
between 0.6and 0.7, in India, the value is about 0.75. In the rest
of the world,irrigated areas are not drained much and frgi is close
to 0.8.
2.4. Model runs with WaterGAP 2.1h
For the period 19012002, WGHM and GIM were driven bymonthly
climate data from the Climate Research Unit (CRU) with aspatial
resolution of 0.5 (covering the global land surface). The CRUTS 2.1
data set includes gridded data for the climate variables
tem-perature, cloudiness and number of rain days from 1901 to
2002.This data set is based on station observations and uses
anomalyanalysis for spatial interpolation (Mitchell and Jones,
2005). Torun the model for the GRACE period 20022009, monthly data
ontemperature, cloudiness and number of rain days for 20032009from
the European Centre for Medium-Range Weather Forecasts(ECMWF)
operational forecast system were used. For precipitation,the Global
Precipitation Climatology Centre (GPCC) full data prod-uct v3
provided gridded monthly values for 19512004 (Rudolf andSchneider,
2005), also with a spatial resolution of 0.5, except thatfor GIM
the GPCC full data product v4 was used until 2002. Forthe years
20052009, the so-called GPCC monitoring product witha spatial
resolution of 1 was used, which is based on a smallernumber of
station observations. In WGHM, monthly precipitation isdistributed
equally over the number of rain days within one month.The monthly
precipitation data are not corrected for measurementerrors, but
precipitation, in particular snow is generally underesti-mated by
measurements mainly due to wind induced undercatch.As this has a
strong influence on simulated snow water storage,precipitation was
corrected in WGHM using mean monthly catchratios and taking actual
monthly temperatures into account (Dlland Fiedler, 2008). Water use
in the three sectors households, man-ufacturing, and cooling of
thermal power plants was computed for19012005, and assumed to be
equal to the values for 2005 from2006 to 2009. For livestock, the
year with the last statistics availablewas 2002, and livestock use
in 20032009 was assumed to be thesame as in 2002. For the runs with
WaterGAP 2.1h, the calibrationparameter values of version 2.1g were
used.
2.5. GWS variations developed from measured
groundwaterlevels
For the High Plains aquifer, which covers about 450,000
km2,Strassberg et al. (2009) developed a time series of GWS
vari-ations for the aquifer as a whole from measured
groundwaterlevels in 1989 wells. Four seasonal values were
developed per year(JanuaryMarch, AprilJune, JulySeptember,
OctoberDecember),covering the years 20032006, based on an average
of 983 wellsper season. To convert water level variations to GWS
variations, theformer were multiplied by a constant specific yield
of 0.15, whichrepresent the area-weighted specific yield (McGuire,
2009). Forthe entire Mississippi basin (approx. 3,248,000 km2),
Rodell et al.(2007) developed a monthly time series of GWS from
water levelobservations in 58 wells in unconfined aquifers. These
wells weredistributed almost evenly over the basin. To convert
water levelvariations to GWS variations, the former were multiplied
by spe-cific yield values determined individually for each well,
rangingfrom 0.02 to 0.32, with a mean of 0.14. For our comparison,
we
used an updated data set covering the period January
2002June2006.
2.6. Total water storage variations from GRACE
satelliteobservations
We compared monthly TWS variations simulated with Water-GAP to
values derived from GRACE satellite observations. Tobetter
understand the uncertainty in GRACE data, solutions com-puted by
three different GRACE processing groups were used. Tocompute
ITG-GRACE2010 monthly solutions
(http://www.igg.uni-bonn.de/apmg/index.php?id=itg-grace2010), a set
of sphericalharmonic coefficients for degrees n = 1. . .120 was
estimated foreach month from August 2002 to August 2009 without
applying anyregularization. More detailed information on this GRACE
solutioncan be found in Mayer-Grr et al. (2010). To determine
continentalwater storage variation from the total GRACE signal, the
followingbackground models were taken into account: ocean, Earth
and poletides, atmospheric and oceanic mass variations as well as
glacialisostatic adjustment (as described in Eicker et al., in
press). Degreen = 1 (geo-center variations) was set to zero as
GRACE does notdeliver realistic estimates of these coefficients.
This is justifiableas a comparison using GPS geo-center estimates
has revealed thatthere is no significant contribution of degree 1
in the two investiga-tion areas. In addition, GFZ release 4 monthly
solutions (Flechtneret al., 2010) and CSR monthly solutions
(Bettadpur, 2007) wereconsidered.
All three GRACE solutions were smoothed using the non-isotropic
filter DDK3 (Kusche et al., 2009). To allow a consistentcomparison
to WGHM results, the filtered results were interpo-lated to the
WGHM 0.5 grid such that basin averages of TWS couldbe computed as
averages over the respective WGHM grid cells. Inorder to compare
TWS modeled with WGHM to GRACE-derived
Fig. 2. (a) Total water withdrawals, in mm/year, and (b)
irrigation water with-drawals in percent of total water
withdrawals, for 19982002. The irrigationpercentage is only shown
if total water withdrawals are at least 0.2 mm/year.
Dll et al JoG (2012) Hansen et al Science (2013)
-
Observations are valuable but dont tell the whole story
Local measurements are difficult to scale
hBp://triplemlandfarms.com/ hBp://nasa.gov
Remote sensing cant see everything
-
We use the integrated hydrologic model ParFlow which is a tool
for computational hydrology
Saturated(Subsurface(
Vadose(Zone(
Land(Surface(
No(Flow(Boundary(
Overland)Flow)
Lateral)Subsurface)Flow)
Exfiltra8on)
Infiltra8on)
Z=0(
P2)
z2)
H2)H1)
z1)
P1)
1)
2)
dz)dx)
dL)x)
Recharge)Overland))Flow)
Variably saturated groundwater flow Fully integrated surface
water Parallel implementa,on Coupled land surface processes
Maxwell (2013); Kollet and Maxwell (2008); Kollet and Maxwell
(2006);Maxwell and Miller (2005); Dai et al. (2003); Jones and
Woodward (2001); Ashby and Falgout (1996)
-
Saturated(Subsurface(
Vadose(Zone(
Land(Surface(
No(Flow(Boundary(
Overland)Flow)
Lateral)Subsurface)Flow)
Exfiltra8on)
Infiltra8on)
Z=0(
P2)
z2)
H2)H1)
z1)
P1)
1)
2)
dz)dx)
dL)x)
Atmospheric)forcings)
Water))Energy)Balance)
Vegeta;on(
Root(zone(
ParFlows coupling with land surface processes (CLM) allows for
simulation of interactions and connections
Maxwell (2013); Kollet and Maxwell (2008); Kollet and Maxwell
(2006);Maxwell and Miller (2005); Dai et al. (2003); Jones and
Woodward (2001); Ashby and Falgout (1996)
Land-energy balance Snow dynamics Driven by meteorology
-
Models can be useful tools to provide insight
Controlled numerical experiments elucidate process interactions
under change
A single perturbation (e.g. temperature increase) can be tracked
through the entire nonlinear system
Connections we see in simulations can provide insight and guide
observations
-
We can use models to propagate tree-scale, beetle impacts to the
hydrologic cycle at the hillslope scale
How do changes to stomatal resistance and leaf area index impact
snow, runoff, storage?
Bedrock is assumed to be 12.5m below the ground surfaceand is a
no flow boundary. Once water reaches the confinesof the domain as
either subsurface lateral flow or overlandflow, it crosses the
boundary and is no longer considered tobe part of the domain,
leading to subsurface storage lossesor gains depending on the
scenario.This hillslope-scale study was simulated with a
hypothe-
tical domain, 500m (x-direction) by 1000m (y-direction) by12.5m
(z-direction). The domain was discretised usingx= 100m, y= 200m and
z=0.5m. Three differentsimulations were run with topographic slopes
of 0.01, 0.08and 0.15m/m (Figure 1) that were selected to represent
thetopography found both in high-elevation,
mountainous,MPB-infested watersheds in the Rocky Mountain
West,along with the flatter MPB-infested watersheds found
insouthernWyoming. The saturated hydraulic conductivity wasset to
0.1m/h. Porosity and vanGenuchten parameters wereassigned for the
watershed. Porosity was held constantthroughout at 0.390 (!) and
the van Genuchten parameterswere set as follows: a =3.5 (1/m), n=2
and Sres = 0.01 torepresent an idealised, relatively fast draining,
sandy clayloam soil. Soils in mountain hillslopes are of course
veryheterogeneous with great uncertainty about both point valuesand
spatial distribution. The assumption of homogeneityinvoked here is
used to isolate the impacts of land-coverchanges on the hydrology
of the system, whereas minimisingother confounding signals.
Approaches that directly includeheterogeneity at the hillslope
scale (Rihani et al., 2010;Atchley and Maxwell, 2011) are beyond
the scope of thiscurrent study and are excellent topics for future
work.Hourly meteorological forcings were taken from the
North American Land Data Assimilation System(NLDAS), forcing
dataset (Cosgrove et al., 2003) for the2008 water year (1 September
2007 to 31 August 2008) atPennsylvania Gulch, Blue River, Colorado.
The climateduring 2008 was typical of an average year for this
regionby comparing precipitation with seasonal averages,
andtherefore a good representation of what a climatologicalweather
pattern will be in the Rocky Mountains, yet still
including high-frequency (hourly) variability. The modelwas run
for each phase for 3 years, re-running the 2008meteorological data
for each year, to minimise theinfluence of initial conditions (but
not running the modelto equilibrium) on simulated results with
results focusingon the third year.
In the model simulations, we defined the process ofmortality in
an affected tree to have four distinct phases: (1)green phase: the
tree is alive and transpiring; (2) red phase:the tree has been
attacked and has ceased transpiring andinterception has slightly
decreased; (3) grey phase: the treeis dead, has no remaining
needles, transpiration has ceased,and interception is significantly
decreased; and (4) diebackphase: the tree has fallen to the ground
and begundecomposing as new vegetation begins to take its
place.During an actual MPB infestation, the vegetation
distribu-tion throughout the four phases is likely to be
hetero-geneous; however, for the sake of simplicity and
tounderstand the magnitudes of difference in the hydrologicand
energy regimes between each phase, we are assuming ahomogeneous
distribution of vegetation during each phase.
The hydrologic and land-energy impacts of each of thefour phases
of MPB infestation were simulated byperturbing two vegetative
parameters: stomatal resistanceand leaf area index. Table 1 shows
how we defined eachphase of MPB infestation, along with its
correspondingland surface classification, maximum and minimum
LAI(depending on season, recall that LAI is dynamicallycalculated
in CLM) and the minimum stomatal conduc-tance. When the MPB infects
a stand of trees, it introducesa blue-stain fungi (Ceratostomella
montia and Europhiumclavigerum) that essentially clogs the xylem
and phloemtubes (Amman, 1978). This renders the tree unable to
takeup water and nutrients from its roots. Stomatal resistancewas
manipulated to represent this phenomenon duringthe red and grey
stages of infestation by increasing theminimum stomatal conductance
until it was maximised andtranspiration approached zero. During the
red and greyphases, it is assumed that there is no new undergrowth
in
Figure 1. Displays the governing processes in the three
simulated watersheds. Arrow lengths indicate flux magnitudes. P is
precipitation, ET isevapotranspiration, O is overland flow, and I
is infiltration.
67PINE BEETLE IMPACTS ON THE WATER AND ENERGY BUDGET
Copyright 2011 John Wiley & Sons, Ltd. Ecohydrol. 6, 6472
(2013)
Mikkelson, Maxwell, Ferguson, McCray, Stednick, Sharp,
Ecohydrology 2013
-
ET decreases with MPB infestation
because increased radiation penetration through the canopycauses
earlier and faster snowmelt. This phenomenon isseen during both
periods of transition, fall and spring, inassociation with the snow
season.Surface saturation increases through each phase of
infestation, but the differences are most apparent duringtimes
of moisture stress (Figure 3). In steeper watersheds,the difference
in surface saturation is not as drastic becausesoils are drier
owing to increased baseflow runoff andsubsequent drainage of
shallow soils. Saturation also drivesmany of the other hydrologic
processes. For example,during November, ET decreases in the red and
greenphases partly because of the dry top surface layer (Figures
2(A) and 3) with a minimum in mid-winter months. In the1% slope
case, the dieback phases ET is able to rebound inthe summer months,
surpassing the grey phase ET. This isnot seen for the steeper
slopes and is due to the fact that the
shallow, surface soil moisture is higher for the 1% slope(Figure
3(A)) than for the other two slopes, allowing moreET to occur in
the dieback phase. This relationship betweendieback rebound and
slope exists because the deeper rootsof the phreatophytic lodgepole
pine are able to accessgroundwater whereas the shrub vegetation
cannot. Theshrub vegetation used in the dieback phase has roots
thatextend less than 3.1m into the subsurface. For the
steeperslopes, the water table is deeper than 3.1m, therebylimiting
the amount of water the plants can transpire.
For example, when looking at the monthly differences inET during
the late fall, ET does not follow the typicalyearly pattern (Figure
2(A)): the grey phase exhibits anincrease in ET whereas all the
other phases experience adecrease in ET. To further explore this
phenomenon, wecompared the daily average ground temperature, SWE
andET for the green and grey phases during a 2-month period
Figure 2. The complete water balance and average monthly snow
water equivalent for the four phases of infestation at 1%, 8% and
15% slopes. Row Ais total monthly ET, row B is total monthly
overland flow, row C is the change in storage from the beginning to
the end of that month, and row D is the
monthly average snow water equivalent. Values are all in mm.
Figure 3. Top layer saturation, averaged hourly over the entire
domain for (A) 1%, (B) 8%, and (C) 15% slopes.
69PINE BEETLE IMPACTS ON THE WATER AND ENERGY BUDGET
Copyright 2011 John Wiley & Sons, Ltd. Ecohydrol. 6, 6472
(2013)
because increased radiation penetration through the canopycauses
earlier and faster snowmelt. This phenomenon isseen during both
periods of transition, fall and spring, inassociation with the snow
season.Surface saturation increases through each phase of
infestation, but the differences are most apparent duringtimes
of moisture stress (Figure 3). In steeper watersheds,the difference
in surface saturation is not as drastic becausesoils are drier
owing to increased baseflow runoff andsubsequent drainage of
shallow soils. Saturation also drivesmany of the other hydrologic
processes. For example,during November, ET decreases in the red and
greenphases partly because of the dry top surface layer (Figures
2(A) and 3) with a minimum in mid-winter months. In the1% slope
case, the dieback phases ET is able to rebound inthe summer months,
surpassing the grey phase ET. This isnot seen for the steeper
slopes and is due to the fact that the
shallow, surface soil moisture is higher for the 1% slope(Figure
3(A)) than for the other two slopes, allowing moreET to occur in
the dieback phase. This relationship betweendieback rebound and
slope exists because the deeper rootsof the phreatophytic lodgepole
pine are able to accessgroundwater whereas the shrub vegetation
cannot. Theshrub vegetation used in the dieback phase has roots
thatextend less than 3.1m into the subsurface. For the
steeperslopes, the water table is deeper than 3.1m, therebylimiting
the amount of water the plants can transpire.
For example, when looking at the monthly differences inET during
the late fall, ET does not follow the typicalyearly pattern (Figure
2(A)): the grey phase exhibits anincrease in ET whereas all the
other phases experience adecrease in ET. To further explore this
phenomenon, wecompared the daily average ground temperature, SWE
andET for the green and grey phases during a 2-month period
Figure 2. The complete water balance and average monthly snow
water equivalent for the four phases of infestation at 1%, 8% and
15% slopes. Row Ais total monthly ET, row B is total monthly
overland flow, row C is the change in storage from the beginning to
the end of that month, and row D is the
monthly average snow water equivalent. Values are all in mm.
Figure 3. Top layer saturation, averaged hourly over the entire
domain for (A) 1%, (B) 8%, and (C) 15% slopes.
69PINE BEETLE IMPACTS ON THE WATER AND ENERGY BUDGET
Copyright 2011 John Wiley & Sons, Ltd. Ecohydrol. 6, 6472
(2013)
Mikkelson, Maxwell, Ferguson, McCray, Stednick, Sharp,
Ecohydrology 2013
-
Snow Water Equivalent (SWE) increases with MPB Infestation
As infestation progresses we see a greater snowpack and a
shorter snow season
because increased radiation penetration through the canopycauses
earlier and faster snowmelt. This phenomenon isseen during both
periods of transition, fall and spring, inassociation with the snow
season.Surface saturation increases through each phase of
infestation, but the differences are most apparent duringtimes
of moisture stress (Figure 3). In steeper watersheds,the difference
in surface saturation is not as drastic becausesoils are drier
owing to increased baseflow runoff andsubsequent drainage of
shallow soils. Saturation also drivesmany of the other hydrologic
processes. For example,during November, ET decreases in the red and
greenphases partly because of the dry top surface layer (Figures
2(A) and 3) with a minimum in mid-winter months. In the1% slope
case, the dieback phases ET is able to rebound inthe summer months,
surpassing the grey phase ET. This isnot seen for the steeper
slopes and is due to the fact that the
shallow, surface soil moisture is higher for the 1% slope(Figure
3(A)) than for the other two slopes, allowing moreET to occur in
the dieback phase. This relationship betweendieback rebound and
slope exists because the deeper rootsof the phreatophytic lodgepole
pine are able to accessgroundwater whereas the shrub vegetation
cannot. Theshrub vegetation used in the dieback phase has roots
thatextend less than 3.1m into the subsurface. For the
steeperslopes, the water table is deeper than 3.1m, therebylimiting
the amount of water the plants can transpire.
For example, when looking at the monthly differences inET during
the late fall, ET does not follow the typicalyearly pattern (Figure
2(A)): the grey phase exhibits anincrease in ET whereas all the
other phases experience adecrease in ET. To further explore this
phenomenon, wecompared the daily average ground temperature, SWE
andET for the green and grey phases during a 2-month period
Figure 2. The complete water balance and average monthly snow
water equivalent for the four phases of infestation at 1%, 8% and
15% slopes. Row Ais total monthly ET, row B is total monthly
overland flow, row C is the change in storage from the beginning to
the end of that month, and row D is the
monthly average snow water equivalent. Values are all in mm.
Figure 3. Top layer saturation, averaged hourly over the entire
domain for (A) 1%, (B) 8%, and (C) 15% slopes.
69PINE BEETLE IMPACTS ON THE WATER AND ENERGY BUDGET
Copyright 2011 John Wiley & Sons, Ltd. Ecohydrol. 6, 6472
(2013)
because increased radiation penetration through the canopycauses
earlier and faster snowmelt. This phenomenon isseen during both
periods of transition, fall and spring, inassociation with the snow
season.Surface saturation increases through each phase of
infestation, but the differences are most apparent duringtimes
of moisture stress (Figure 3). In steeper watersheds,the difference
in surface saturation is not as drastic becausesoils are drier
owing to increased baseflow runoff andsubsequent drainage of
shallow soils. Saturation also drivesmany of the other hydrologic
processes. For example,during November, ET decreases in the red and
greenphases partly because of the dry top surface layer (Figures
2(A) and 3) with a minimum in mid-winter months. In the1% slope
case, the dieback phases ET is able to rebound inthe summer months,
surpassing the grey phase ET. This isnot seen for the steeper
slopes and is due to the fact that the
shallow, surface soil moisture is higher for the 1% slope(Figure
3(A)) than for the other two slopes, allowing moreET to occur in
the dieback phase. This relationship betweendieback rebound and
slope exists because the deeper rootsof the phreatophytic lodgepole
pine are able to accessgroundwater whereas the shrub vegetation
cannot. Theshrub vegetation used in the dieback phase has roots
thatextend less than 3.1m into the subsurface. For the
steeperslopes, the water table is deeper than 3.1m, therebylimiting
the amount of water the plants can transpire.
For example, when looking at the monthly differences inET during
the late fall, ET does not follow the typicalyearly pattern (Figure
2(A)): the grey phase exhibits anincrease in ET whereas all the
other phases experience adecrease in ET. To further explore this
phenomenon, wecompared the daily average ground temperature, SWE
andET for the green and grey phases during a 2-month period
Figure 2. The complete water balance and average monthly snow
water equivalent for the four phases of infestation at 1%, 8% and
15% slopes. Row Ais total monthly ET, row B is total monthly
overland flow, row C is the change in storage from the beginning to
the end of that month, and row D is the
monthly average snow water equivalent. Values are all in mm.
Figure 3. Top layer saturation, averaged hourly over the entire
domain for (A) 1%, (B) 8%, and (C) 15% slopes.
69PINE BEETLE IMPACTS ON THE WATER AND ENERGY BUDGET
Copyright 2011 John Wiley & Sons, Ltd. Ecohydrol. 6, 6472
(2013)
Mikkelson, Maxwell, Ferguson, McCray, Stednick, Sharp,
Ecohydrology 2013
-
Decreased ET and more snow increases in runoff and earlier
timing
because increased radiation penetration through the canopycauses
earlier and faster snowmelt. This phenomenon isseen during both
periods of transition, fall and spring, inassociation with the snow
season.Surface saturation increases through each phase of
infestation, but the differences are most apparent duringtimes
of moisture stress (Figure 3). In steeper watersheds,the difference
in surface saturation is not as drastic becausesoils are drier
owing to increased baseflow runoff andsubsequent drainage of
shallow soils. Saturation also drivesmany of the other hydrologic
processes. For example,during November, ET decreases in the red and
greenphases partly because of the dry top surface layer (Figures
2(A) and 3) with a minimum in mid-winter months. In the1% slope
case, the dieback phases ET is able to rebound inthe summer months,
surpassing the grey phase ET. This isnot seen for the steeper
slopes and is due to the fact that the
shallow, surface soil moisture is higher for the 1% slope(Figure
3(A)) than for the other two slopes, allowing moreET to occur in
the dieback phase. This relationship betweendieback rebound and
slope exists because the deeper rootsof the phreatophytic lodgepole
pine are able to accessgroundwater whereas the shrub vegetation
cannot. Theshrub vegetation used in the dieback phase has roots
thatextend less than 3.1m into the subsurface. For the
steeperslopes, the water table is deeper than 3.1m, therebylimiting
the amount of water the plants can transpire.
For example, when looking at the monthly differences inET during
the late fall, ET does not follow the typicalyearly pattern (Figure
2(A)): the grey phase exhibits anincrease in ET whereas all the
other phases experience adecrease in ET. To further explore this
phenomenon, wecompared the daily average ground temperature, SWE
andET for the green and grey phases during a 2-month period
Figure 2. The complete water balance and average monthly snow
water equivalent for the four phases of infestation at 1%, 8% and
15% slopes. Row Ais total monthly ET, row B is total monthly
overland flow, row C is the change in storage from the beginning to
the end of that month, and row D is the
monthly average snow water equivalent. Values are all in mm.
Figure 3. Top layer saturation, averaged hourly over the entire
domain for (A) 1%, (B) 8%, and (C) 15% slopes.
69PINE BEETLE IMPACTS ON THE WATER AND ENERGY BUDGET
Copyright 2011 John Wiley & Sons, Ltd. Ecohydrol. 6, 6472
(2013)
Mikkelson, Maxwell, Ferguson, McCray, Stednick, Sharp,
Ecohydrology 2013
-
Aspen Water Treatment Plant
LETTERSPUBLISHED ONLINE: 28 OCTOBER 2012 | DOI:
10.1038/NCLIMATE1724
Water-quality impacts from climate-inducedforest die-offKristin
M. Mikkelson1,2*, Eric R. V. Dickenson1,3, Reed M. Maxwell2,4, John
E. McCray1,2and Jonathan O. Sharp1,2
Increased ecosystem susceptibility to pests and other stres-sors
has been attributed to climate change1, resulting in un-precedented
tree mortality from insect infestations2. In turn,large-scale tree
die-off alters physical and biogeochemicalprocesses, such as
organic matter decay and hydrologic flowpaths, that could enhance
leaching of natural organic matterto soil and surface waters and
increase potential formationof harmful drinking water disinfection
by-products3,4 (DBPs).Whereas previous studies have investigated
water-quantityalterations due to climate-induced, forest
die-off5,6, impactson water quality are unclear. Here,
water-quality data setsfrom water-treatment facilities in Colorado
were analysedto determine whether the municipal water supply has
beenperturbed by tree mortality. Results demonstrate higher to-tal
organic carbon concentrations along with significantlymore DBPs at
water-treatment facilities using mountain-pine-beetle-infested
source waters when contrasted with thoseusing water from control
watersheds. In addition to this dif-ferentiation between
watersheds, DBP concentrations demon-strated an increase within
mountain pine beetle watershedsrelated to the degree of
infestation. Disproportionate DBP in-creases and seasonal
decoupling of peak DBP and total organiccarbon concentrations
further suggest that the total organiccarbon composition is being
altered in these systems.
The mountain pine beetle (MPB, Dendroctonus
ponderosae)infestation has reached epidemic proportions and is
generatinggrowing concern for regional water resources with little
knownabout potential impacts. In the Rocky Mountains, warmer
winterminimum temperatures and persistent drought conditions
havecontributed to an ongoing MPB epidemic7 that has affectedmore
than 4 million acres of lodgepole pine forests in Coloradoand
Wyoming (Fig. 1). Changes in hydrology following bark-beetle
infestation such as decreased interception, increased erosionand
particulate transport8, increased soil moisture and
increasedradiation to the forest floor5 can lead to altered
degradation andtransport of soil organic matter in both particulate
and dissolvedforms9. Total organic carbon (TOC, which comprises
particulateand dissolved organic carbon) increases have been
observed acrosslarge areas of the Northern Hemisphere, and although
there isno scientific consensus on the driving mechanisms of
increasedTOC a number of different factors have been proposed
includingchanges in acid deposition10, variability in climate11 and
land-usechanges12. We propose that the recent bark-beetle epidemic
isanother mechanism that may alter TOC loading and compositionin
surface and groundwaters.
1Department of Civil and Environmental Engineering, Colorado
School of Mines, Golden, Colorado 80401, USA, 2Hydrological Science
and EngineeringProgram, Colorado School of Mines, Golden, Colorado
80401, USA, 3Water Quality Research and Development Division,
Southern NevadaWaterAuthority, Henderson, Nevada 89015, USA,
4Department of Geology and Geological Engineering, Colorado School
of Mines, Golden, Colorado 80401,USA. *e-mail:
[email protected].
Changes in TOC characteristics and increased loading can leadto
human health concerns as humic and fulvic fractions of
naturalorganic matter (NOM) have been correlated with the
formationof DBPs, such as trihalomethanes (THMs, known
carcinogens),during chlorination3,13,14. Hence, the potential for
exceedance ofregulatory limits, human health impacts and increased
treatmentcosts are potential concerns for water-treatment
facilities associatedwith bark-beetle-infested watersheds. The
objective of this studywas to collect and analyse archived,
publicly available water-qualitydata from water-treatment
facilities located in the Rocky Mountainregion of Colorado.
Water-quality data were compared betweenMPB-infested watersheds and
regionally analogous facilities locatedin watersheds that did not
experience the same degree of MPBinfestation (control
watersheds).
Archived water-quality data were collected from nine
differenttreatment plants in Colorado, which included four control
(Aspen,Carbondale, Glenwood Springs and Gypsum) and five
MPB-impacted facilities (Kremmling, Steamboat Springs, Winter
Park,Dillon and Granby). The average level of infestation at the
controlsites was about a quarter (0.8 0.2 trees killed per hectare)
thatof the MPB-impacted sites (3.0 0.8 trees killed per hectare)and
sites were as geographically similar as the available datawould
allow (Supplementary Fig. S4). Most treatment facilitiescollect
their water from surface-water sources; however, WinterPark and
Dillon primarily use a groundwater supply whereasCarbondale, Aspen
and Steamboat Springs use both groundwaterand surface-water
supplies (Supplementary Table S2). Water-quality samples were
collected and analysed by the treatmentfacilities in quarterly
intervals in compliance with EnvironmentalProtection Agency (EPA)
standardized procedures15 and spanthe years of 20042011, during
which the impacted lodgepolepine forests experienced heavy MPB
infestation (Fig. 1). Althoughfacilities were analysed individually
(Supplementary Table S3),trends were more significant owing to
larger sample sizes whenanalysed as a group (that is, MPB or
control) and thus aggregatedresults will be presented.
Quarterly TOC samples taken before water treatment weregrouped
for these distinct facilities (Fig. 2a). Our analysis demon-strates
significantly more TOC in the MPB watersheds versusthe control
watersheds with respect to both mean and maximumconcentrations
(p< 0.0001 fromMannWhitney test). As is typicalin most
watersheds, surface-water sources in impacted watershedshad
significantly higher TOC concentrations than groundwatersources;
however, this interpretation is limited as the only
availableTOCdata from an impacted groundwater source was from the
town
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange 1
What can observations tell us about carbon cycle and water
quality?
-
Legend
14050001
14010004
14010003
14010001
MutipleClip
SurfText
channery loam
clay loam
coarse sandy loam
cobbly loam
fine sandy loam
gravelly loam
gravelly sandy loam
loam
sandy loam
very cobbly loam
very cobbly sandy loam
Multiple
!!!!!
!"#$%!&'()*+'!
"#$%&'(%#!")*+,-.!
/*%,'0!
1+,#$*!!2%*3!
4+55(,!
6*$&&5+,-!
/5$,7((8!!")*+,-.! /0).9&!
:%*'(,8%5$!
;.)$,!
!"#$%&'$()*+,$%-%$$.*
/0%$)*123$(*/0%$ )*($,$405$(*
6,$%"-$*)*7%-".28*1"##$%*2.*9024
6,$%"-$*:$"(*;%$$&*5$%*/ 5?24 520 62? 0243
526,"="$#@"(A9#@B#;9$:(C($/ ?624 327 62F 02?G
024,"="$#@"(A9#@B#;9$:(C(89:;(+>H/ 0623 F23 625 0236
62FIJJ9$(K#LJ#(+H/ 572G 321 62F 0257 02FM=N9(+O/ ??2? ?20 024 02?0
42G
Water treatment facilities in the Rocky Mountains are already
experiencing MPB impacts
*Beetle kill was the only sta>s>cally significant variable
between MPB and control watersheds. (Mikkelson et al NCC 2013)
-
0"
2"
4"
6"
8"
10"
12"
MPB" Non.MPB"
TOC$(
mg/l)$
min"
median"
max"
0"
20"
40"
60"
80"
100"
MPB" Non"MPB"
HAA5
$(ug/l)$
b"
a"
c"
0"
40"
80"
120"
160"
MPB" Non"MPB"
TTHM
$(ug/l)$
Control"
TOC (m
g/l)
HAA5
(ug/l)
TTHM
(ug/l)
Higher TOC and DBP concentrations are observed in MPB-impacted
facilities than at control facilities.
But what is the mechanism?
(Mikkelson et al NCC 2013)
-
0"
2"
4"
6"
8"
10"
12"
MPB" Non.MPB"
TOC$(
mg/l)$
min"
median"
max"
0"
20"
40"
60"
80"
100"
MPB" Non"MPB"
HAA5
$(ug/l)$
b"
a"
c"
0"
40"
80"
120"
160"
MPB" Non"MPB"
TTHM
$(ug/l)$
Control"
TOC (m
g/l)
HAA5
(ug/l)
TTHM
(ug/l)
Higher TOC and DBP concentrations are observed in MPB-impacted
facilities than at control facilities.
MPB
Control
But what is the mechanism?
(Mikkelson et al NCC 2013)
-
Our conceptual model links late summer groundwater uptake and
tree mortality
Bearup, Maxwell, Clow and McCray Nature Climate Change,
2014.
-
Big T.
N. Inlet
We use a paired-watershed approach combined with historical
observations
(Bearup et al NCC 2014)
-
We use end-member mixing to determine contributions to the
hydrograph
End Member Mixing Analysis(EMMA)
Three end-member hydrograph separation
15 10 5 0 5 10
50
510
15
U1
U2
12
3
4
5
Baseflow *.
................
.... .. .
56 7 89 10111213141516
17181920212223
Rain
Snow
EC & 18O
Snow
Rain
Groundwater
Streamwater
EC & 18O
Snow
Rain
Groundwater
Streamwater
EC & 18O
Snow
Rain
Groundwater
Streamwater
EC & 18O
Snow
Rain
Groundwater
Streamwater
Stream Flow
Snow
Rain
Ground-water
(Bearup et al NCC 2014)
-
0.0
0.2
0.4
0.6
0.8
1.0 Big Thompson
Frac
tiona
l Con
tribu
tion
to S
tream
flow
rainsnowgroundwater
Jul Aug Sept Oct
a)
2012
0.0
0.2
0.4
0.6
0.8
1.0 North Inlet
Jul Aug Sept Oct
b)
0.0
0.2
0.4
0.6
0.8
1.0
Frac
tiona
l Con
tribu
tion
to S
tream
flow
Jul Aug Sept Oct
c)
1994
0.0
0.2
0.4
0.6
0.8
1.0
Jul Aug Sept Oct
d)1994 Big
T
2012 Big T
2012 N. Inlet
Tempo
ral
Spa>al
We found an increase in GW contributions for impacted
watersheds
(Bearup et al NCC 2014)
-
Tree Scale Sap flux: 16 L/day
(Hubbard et al 2013, CO) Stand Scale Potometers: 3.4 mm/day
(Knight et al 1981, WY) Hillslope to Watershed Scale ParFlow ET:
- 20-35%
(Mikkelson et al 2013, CO) Watershed Scale MODIS ET:
(Maness et al 2013, BC)
Eddy Covariance: 0.7 mm/day (Brown et al 2014, BC; Biederman et
al 2014, CO/WY; Reed et al 2014; WY)
020
4060
80
ET(mm)
July Aug Sep Oct
1% Slope
020
4060
80
ET(mm)
July Aug Sep Oct
8% Slope
020
4060
80
ET(mm)
July Aug Sep Oct
15% Slope
Estimating evapotranspiration is challenging across scales
-
0 20 40 60 80 100
0.0
0.5
1.0
1.5
Percent of net trees killed in impacted area
Flux
Cha
nge
(mm
/day
)
MODIS
Comp
arison
(Manes
s et al
2013)
Sap F
lux C
ompa
rison
(Hub
bard
et al
2013
)
Temporal Control
Spatial Control
Temporal Control(Constant EM)
a)
0.0
0.5
1.0
1.5
T
T
T TT
T
T
T
TT
C
C
C CC
C
C
C C
C
S
SS
SS
S
S
S S S
Jul Aug Sept Oct
b) TCS
Temporal ControlConstant EMSpatial ControlModel Grey PhaseModel
Red Phase
Which allowed a scale-up of ET fluxes to the watershed
(Bearup et al NCC 2014)
-
Using models to predict streamwater age and composition is an
important topic in hydrology
What are the physical processes and material properties that
control transit time distribution? How and why do these processes
vary with time, ambient conditions, and place? How can we deal with
the effects of ET partitioning in predicting transit time
distributions
-
Integrated hydrologic models may be used to attribute source and
to study the effects of disturbances such as ET
Outflow
Sublimation
Snowfall
Interception
Transpiration
Stream Flow
Snow
Rain
Ground-water
(Bearup et al, in review)
-
Outflow
Evaporation
Rainfall
Interception
Transpiration
Stream Flow
Snow
Rain
Ground-water
(Bearup et al, in review)
Integrated hydrologic models may be used to attribute source and
to study the effects of disturbances such as ET
-
0.0
0.2
0.4
0.6
0.8
1.0 Big Thompson
Frac
tiona
l Con
tribu
tion
to S
tream
flow
rainsnowgroundwater
Jul Aug Sept Oct
a)
2012
0.0
0.2
0.4
0.6
0.8
1.0 North Inlet
Jul Aug Sept Oct
b)
0.0
0.2
0.4
0.6
0.8
1.0
Frac
tiona
l Con
tribu
tion
to S
tream
flow
Jul Aug Sept Oct
c)
1994
0.0
0.2
0.4
0.6
0.8
1.0
Jul Aug Sept Oct
d)1994 Big
T
2012 Big T
2012 N. Inlet
Model Results Field Observations
(Bearup et al, in review)
Transient model simulations allow a virtual hydrograph
separation and show an increase in groundwater contribution and
demonstrate similar behavior to observations
(Bearup et al NCC 2014)
-
Groundwater-generated outflow is greater in infested watersheds
at early times, but shows less memory
Living HillslopeDead Hillslope
1 year 10 years3 months 100 years
(Bearup et al, in review)
STEADY STATE RESULTS
-
Big Thompson Model: 100 m Resolu>on
1 km2 Forested Domain: ET at Variable Resolu>on
for 8% slope
2 m resolu>on
Colorado Model: 1 km Resolu>on 100 m resolu>on 500 m
resolu>on
Denver
East Inlet Model: 10 m Resolu>on
We are using a multi-scale modeling approach
-
We are using an integrated hydrologic model to study scaling
implications of beetle infestation
Big T.
Green Phase June Depth (m)
0
1
2
3
4
Difference (Grey Green)
1.0
0.5
0.0
0.5
1.0
Green Phase August Depth (m)
0
1
2
3
4
Difference (Grey Green)
1.0
0.5
0.0
0.5
1.0
(Penn et al, in review)
-
Green Phase June Depth (m)
0
1
2
3
4
Difference (Grey Green)
1.0
0.5
0.0
0.5
1.0
Green Phase August Depth (m)
0
1
2
3
4
Difference (Grey Green)
1.0
0.5
0.0
0.5
1.0
(Penn et al, in review)
Green Phase June Depth (m)
0
1
2
3
4
Difference (Grey Green)
1.0
0.5
0.0
0.5
1.0
Green Phase August Depth (m)
0
1
2
3
4
Difference (Grey Green)
1.0
0.5
0.0
0.5
1.0
Depth to water table difference (grey green)
-1.0
-0.5
0.0
0.5
1.0
Models indicate higher groundwater tables in infested areas
-
05
10152025303540455055
A) Transpiration
Green PhaseGrey Phase
05
10152025303540455055
B) Intercepted Evaporation
Nov Jan Mar May Jul Sep05
10152025303540455055
C) Soil Evaporation
Transpiration or Evaporation (mm)
Nov Jan Mar May Jul Sep0
102030405060708090
100D) Total Evapotranspiration
Models exhibit compensation in evapotranspiration
(Penn et al, in review)
Transpira>on Intercepted Evapora>on
Soil Evapora>on Total Evapotranspira>on
-
Modeled streamflow response is muted
0
2
4
6
8
10
12
14
16
18
20
22
Outflow (m3/s)
Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Oct0123456789
10
Cumulative Runoff (x107 m3)
Grey PhaseGreen Phase
A)
B)11%
(Penn et al, in review)
-
Bridging scales allows us to help quantify the cascade of
impacts from the mountain
pine beetle epidemic
The mountain pine beetle infestation of North America is the
first observable climate change impact on water quality and helps
us quantify transpiration
We see increased groundwater contributions from beetle-killed
watersheds which allow us to estimate transpiration
We can use hydrologic models to predict source contribution and
water age
Models allow us to scale impacts from the hillslope to the
watershed
-
46
Thank You!
This material was based upon work supported by the Na>onal
Science Founda>on (WSC-1204787) and U.S. Geological Survey
(G-2914-1). Any opinions, findings, and conclusions or
recommenda>ons expressed in this material are those of the
authors and do not necessarily reflect the views of these
organiza>ons.
-
https://xkcd.com/1007/ xkcd FTW