534
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
Impact of Vegetation Feedback on the Response of Precipitation to Antecedent Soil
Moisture Anomalies over North America
YEONJOO KIM
AND
GUILING WANG
Department of Civil and Environmental Engineering, and Center for Environmental Sciences and Engineering, University of
Connecticut, Storrs, Connecticut
(Manuscript received 6 July 2006, in final form 29 January 2007)
ABSTRACT
Previous studies support a positive soil moisture–precipitation feedback over a major fraction of North
America; that is, initial soil moisture anomalies lead to precipitation anomalies of the same sign. To
investigate how vegetation feedback modifies the sensitivity of precipitation to initial soil moisture conditions over North America, a series of ensemble simulations are carried out using a modified version of the
coupled Community Atmosphere Model–Community Land Model (CAM–CLM). The modified CLM includes a predictive vegetation phenology scheme so that the coupled model can represent interactions
between soil moisture, vegetation, and precipitation at the seasonal time scale. The focus of this study is on
how the impact of vegetation feedback varies with the timing and direction of initial soil moisture anomalies. During summer, wet soil moisture anomalies lead to increase in leaf area index and, consequently,
increase in evapotranspiration and surface heating. Such increases tend to favor precipitation. Therefore,
under wet summer soil moisture anomalies, the soil moisture–induced precipitation increase is reinforced
when predictive phenology is included. That is, the vegetation feedback to precipitation is positive. The
response of vegetation to dry soil moisture anomalies in the summer months, however, is not significant due
probably to a dry bias in the model, so the resulting vegetation feedback on precipitation is minimal. To soil
moisture anomalies in spring, the leaf area index (LAI) response is delayed since LAI is still limited by cold
temperature at that time of the year. During the summer following wet spring soil moisture anomalies,
vegetation feedback is negative; that is, it tends to suppress the response of precipitation through the
depletion of soil moisture by vegetation.
1. Introduction
Soil moisture–precipitation coupling may result in
persistence of climate anomalies, making soil moisture
a potentially useful predictor in seasonal predictions.
The slowly varying soil moisture records past and present precipitation anomalies; as the resulting soil moisture feeds back to influence precipitation, this may lead
to the persistence of soil moisture and precipitation
anomalies. Where vegetation growth is limited by
water, this soil moisture–precipitation coupling is
modified by vegetation feedback, with uncertain impact on the land-induced precipitation persistence. For
example, wetter-than-normal soil tends to promote
precipitation and through the soil moisture–precipitation feedback may lead to persistence of higher-than-
Corresponding author address: Dr. Guiling Wang, Department
of Civil and Environmental Engineering, University of Connecticut, 261 Glenbrook Road, Storrs, CT 06269.
E-mail: [email protected]
DOI: 10.1175/JHM612.1
© 2007 American Meteorological Society
JHM612
normal precipitation. As a result of the wetness, vegetation grows denser, which has two consequences:
first, the increase of vegetation favors more precipitation through its impact on albedo and Bowen ratio,
which enhances the wetness persistence, leading to a
positive feedback (e.g., Bounoua et al. 2000; Buermann
et al. 2001); second, the enhanced transpiration leads to
faster depletion of soil moisture, which may reduce the
persistence of wet anomalies, leading to a negative
feedback (e.g., Pielke et al. 1998; Wang et al. 2006).
Whether the net impact is positive or negative is uncertain. Such competing mechanisms or feedbacks are
further elaborated using the diagram in Fig. 1.
Numerous studies have tackled the issue of how initial soil moisture anomalies impact climate conditions
(e.g., Shukla and Minz 1982; Oglesby and Erickson
1989; Bosilovich and Sun 1999; Pal and Eltahir 2001;
Kim and Wang 2007). Most of these studies agreed
upon a positive feedback between soil moisture and
precipitation: wet (dry) soil tends to enhance (suppress)
precipitation through soil moisture’s impact on evapo-
JUNE 2007
KIM AND WANG
535
FIG. 1. Diagram for the impact of vegetation feedback on how precipitation responds to
initial soil moisture anomalies. Here ⌬P⬘ indicates changes in precipitation due to vegetation
feedback, and dashed lines indicate negative feedback.
transpiration. However, none of these studies considered the impact of the feedback from the dynamically
varying vegetation, although several studies examined
the impact of different prescribed vegetation on seasonal and interannual climate (e.g., Dirmeyer 1994).
Recently, remotely sensed vegetation indices such as
normalized difference vegetation index (NDVI) and
NDVI-derived leaf area index (LAI) have been used to
prescribe vegetation conditions in land models and to
study the impact of vegetation on climate (e.g., Chase et
al. 1996; Bounoua et al. 2000; Buermann et al. 2001;
Guillevic et al. 2002). Bounoua et al. (2000) found that,
as a result of global vegetation increase, both evapotranspiration and precipitation increase, and evapotranspiration increases more than precipitation does.
Guillevic et al. (2002), however, found that the interannually varying vegetation influences evapotranspiration, but its influence on large-scale climate dynamics is
very weak. Because of the prescribed vegetation variations in the models used, these studies did not directly
tackle the issue of soil moisture–vegetation–precipitation coupling.
Recently, vegetation phenology schemes simulating
the response of vegetation at the seasonal time scale to
hydrometeorological and other environmental conditions have been incorporated into land surface and climate models (Dickinson et al. 1998; Lu et al. 2001;
Tsvetsinskayaet al. 2001; Kim and Wang 2005). These
models provide useful tools for studying seasonal vegetation–climate interactions. For example, Lu et al.
(2001) coupled the CENTURY ecosystem model with
the Regional Atmospheric Modeling System (RAMS),
and performed simulations with both the offline and
coupled models over the United States. Based on spatial averages over the central United States, lower
simulated LAI in the coupled model than prescribed in
the offline RAMS leads to more precipitation due to
larger vegetation transmissivity, resulting in greater radiation at the land surface, and finally more convective
precipitation in the coupled model. However, at one
grid cell where winter wheat is the dominant vegetation, lower LAI in the coupled model than in the offline
RAMS due to harvest leads to less precipitation in the
coupled model.
In this study we use the coupled Community Atmosphere Model–Community Land Model (CAM–CLM).
The model has been modified to include the predictive
vegetation phenology scheme of Kim and Wang (2005),
which allows us to study soil moisture–vegetation–precipitation feedbacks at the seasonal time scale. We focus on North America, a region of strong land–atmosphere coupling (Koster et al. 2004; Wang et al. 2007)
identified by many GCMs including the CAM–CLM
model. In our previous study (Kim and Wang 2007), we
investigated the impact of soil moisture anomalies on
subsequent precipitation using the coupled CAM–CLM
with prescribed vegetation phenology. The present
study focuses on how vegetation feedback modifies the
sensitivity of precipitation to initial soil moisture conditions.
2. Model and methodology
a. Model description
The model used in this study is version 3 of the National Center for Atmospheric Research (NCAR)
536
JOURNAL OF HYDROMETEOROLOGY
CAM (CAM3) (Collins et al. 2004) coupled with version 3 of CLM (CLM3) (Dai et al. 2003; Oleson et al.
2004). Oceanic boundary conditions in this coupled
land–atmosphere model are prescribed with the climatological monthly varying sea surface temperature and
sea ice coverage. The level of atmospheric CO2 is assumed to be 355 ppm. Among the three dynamics
schemes available in CAM [Eulerian spectral, semiLagrangian dynamics, and finite volume (FV) dynamics], we choose the FV dynamical core (Lin and Rood,
1996; Lin 2004) with a horizontal resolution of 2° latitude by 2.5° longitude and a total of 26 levels in the
vertical direction. The land model CLM3 has 10 unevenly spaced soil layers, up to 5 snow layers, and 1
vegetation layer. Land surface within each grid cell is
represented by the fractional coverage of four types of
patches (glacier, lake, wetland, and vegetated), and the
vegetation portion of the grid cell is represented by the
fractional coverage of up to 4 out of 16 different plant
functional types (PFTs) available in the model. In this
study, the default leaf phenology scheme in CLM3 is
replaced with a predictive scheme that has been validated against the latest Moderate Resolution Imaging
Spectroradiometer (MODIS) observational data over
North America (Kim and Wang 2005).
In the predictive phenology scheme, the PFT-specific
leaf area index is updated daily by scaling down the
annual maximum leaf area index (LAImax) with a predictive phenology factor (D):
LAIdaily ⫽ LAImaxD,
共1兲
where LAIdaily is the PFT-specific daily LAI, and
LAImax is derived from the monthly PFT-specific
MODIS LAI at 0.5° ⫻ 0.5° resolution (Tian et al. 2004;
see Fig. 2). In Fig. 2, only information for the five primary PFTs that exist in North America is presented.
Over this region, needleleaf trees (Fig. 2a) are mostly
temperate evergreen trees; broadleaf trees (Fig. 2b) are
mostly temperate deciduous trees, which are cold-deciduous; shrubs (Fig. 2c) consist of winter deciduous
and evergreen shrubs; and grasses and crops (Figs. 2d
and 2e) are both cold- and drought-deciduous, and are
further divided into C3 and C4 types in the model according to the photosynthetic pathway. The phenology
factor (D), ranging from zero to one, is simulated for
cold-deciduous plants and drought-deciduous plants
separately. For plants responding to both coldness and
drought (e.g., grasses and crops), the phenology factor
is determined based on the multiplicative effect of cold
and drought stresses. For evergreen trees, their LAI
seasonality is prescribed based on the MODIS LAI observations. For crops, their climatological plantation
and harvest times are derived from the MODIS NDVI,
VOLUME 8
but their LAIs between plantation and harvest are predicted in response to hydrometeorological conditions in
the same way as grasses are.
In predicting leaf green-up, development, and senescence, the winter deciduous phenology scheme considers the impact of 10-day average air temperature, accumulated growing degree-days (AGDD), soil temperature, and photoperiod. The base temperatures for
AGDD are 0°C for trees and ⫺5°C for grass as grass
can survive under colder temperature than trees. Once
the criteria for leaf green up or senescence are met, it is
assumed that the full leaf display in the beginning of the
growing season or complete leaf offset at the end of the
growing season takes 15 days. The drought deciduousness is predicted based on the whole plant water stress
factor, which depends on soil water potential in different soil layers and the plant rooting profile. It ranges
from zero at the permanent wilting point to one at saturation. The drought-deciduous phenology scheme predicts leaf shedding and growing based on the 10-day
running mean of plant water stress. Further details
about the phenology scheme can be found in Kim and
Wang (2005).
In the land model, changes in LAI influence land
surface properties such as albedo, surface roughness,
and stomata resistance. In particular, stomata resistance, which is coupled with photosynthesis and transpiration, is important in determining the amount of
soil moisture transpired to the overlying atmosphere
(Bounoua et al. 2000). CLM uses a stomata resistanceleaf photosynthesis model similar to Collatz et al. (1991,
1992). The inverse of stomata resistance (i.e., stomata
conductance) is linearly related to the leaf photosynthesis, which is limited by temperature and soil wetness
and estimated with PFT-specific parameters. Further
details about CLM3 can be found in Dai et al. (2003)
and Oleson et al. (2004).
b. Methodology
Primary simulations using the coupled CAM3–CLM3
model include an initial integration and a large number
of ensemble simulations with different initial soil moisture conditions and different vegetation treatments.
Driven with the climatological SST, the initial integration is carried out for 12 yr. Data from the first 2 yr are
discarded as the model spinup, and the last 10 yr of
data, although it may be short, are used to derive the
model climatology of soil moisture on the first day of
each month. This soil moisture climatology is used to
initialize subsequent experimental ensemble simulations, integrated from the first day of a given month
until the end of the year.
Each ensemble includes five members, which are dif-
JUNE 2007
KIM AND WANG
537
FIG. 2. Percentage of the grid cell occupied by and maximum LAI of (a) needleleaf trees,
(b) broadleaf trees, (c) shrubs, (d) grasses, and (e) crops on 0.5° ⫻ 0.5° map.
ferent from each other only in the initial soil moisture
condition. For an ensemble without initial soil moisture
anomalies, for example, its five members are initialized
with 100%, 99%, 98%, 97%, and 96% of the soil mois-
ture climatology. For an ensemble with 80% dry (or
wet) anomalies of soil moisture climatology, its five
members are initialized with 20%, 19%, 18%, 17%, and
16% (or 180%, 179%, 178%, 177%, and 176%) of the
538
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
TABLE 1. Lists of simulations.
Name of
ensemble
Control
SM
FIG. 3. Map of the study domain. A lined box presents North
America (“NA”), the domain of initial soil moisture anomalies in
the numerical experiments. The shaded area defines the Mississippi River basin on a 2° ⫻ 2.5° resolution (Bosilovich and Chern
2006).
C_Apr
C_May
C_Jun
C_Jul
C_Aug
SM_D80_Apr
SM_D80_May
SM_D80_Jun
SM_D80_Jul
SM_D80_Aug
SM_W80_Apr
SM_W80_May
SM_W80_Jun
SM_W80_Jul
SM_W80_Aug
SM_D30_Jun
Start
date
1 April
1 May
1 June
1 July
1 August
1 April
1 May
1 June
1 July
1 August
1 April
1 May
1 June
1 July
1 August
1 June
SM_W30_Jun
soil moisture climatology. In the experimental simulations, an increase or decrease of soil moisture equivalent of 80% and 30% of its climatology is applied. However, to distinguish signal from noise, our result analysis
in section 3 will mostly focus on an extremely large
magnitude (i.e., 80%) of soil moisture anomalies, although results from ensembles with a smaller magnitude (i.e., 30%) of soil moisture anomalies are also presented for comparison purpose. Note that in CAM3–
CLM3 over much of North America, more than 80%
increase of climatology is required to reach the field
capacity; about 20% decrease of climatology is needed
to reach the wilting point (Fig. 4 of Kim and Wang
2007). Further, the Illinois State Water Survey observed that soil moisture ranges from about 90% below
to about 50% above its mean value at the most variable
station, and ranges from about 40% below to about
30% above at the least variable station. This indicates
that 80% increase and decrease of soil moisture climatology may be beyond the natural variability in some
places, although the observed soil moisture is not directly comparable to the model soil moisture due to
their discrepancies in the spatial and temporal resolutions (section 4b of Kim and Wang 2007).
Initial soil moisture anomalies are applied across
much of North America (the lined box in Fig. 3)
throughout the whole soil depth in the model (⬃3.4 m).
While Kim and Wang (2007) examined the impact of
spatial coverage and depth of soil moisture anomalies
on subsequent precipitation in details, this study focuses on vegetation feedback by applying soil moisture
anomalies over a same spatial coverage and soil depth.
Our results analysis will focus on the Mississippi River
basin (shaded in Fig. 3) where precipitation is most
SM_Veg
SM_Veg_D80_Apr
SM_Veg_D80_May
SM_Veg_D80_Jun
SM_Veg_D80_Jul
SM_Veg_D80_Aug
SM_Veg_W80_Apr
SM_Veg_W80_May
SM_Veg_W80 _Jun
SM_Veg_W80_Jul
SM_Veg_W80_Aug
SM_Veg_D30_Jun
SM_Veg_W30_Jun
1 April
1 May
1 June
1 July
1 August
1 April
1 May
1 June
1 July
1 August
1 June
Initial soil moisture
(% of climatology)
100, 99, 98, 97,
and 96
20, 19, 18, 17,
and 16
180, 179, 178, 177,
and 176
70, 69, 68, 67,
and 66
130, 129, 128, 127,
and 126
20, 19, 18, 17,
and 16
180, 179, 178, 177,
and 176
70, 69, 68, 67,
and 66
130, 129, 128, 127,
and 126
sensitive to initial soil moisture anomalies (Kim and
Wang 2007). This region also includes most of the
North American areas of strong coupling between soil
moisture and precipitation in CAM3–CLM3 (Koster et
al. 2004; Wang et al. 2007). In addition, the dominant
vegetation in this region includes grasses and crops
(Fig. 2a), both of which respond to soil water stress.
Therefore, vegetation–soil moisture–precipitation coupling is expected to be strong in this region.
Three different types of ensembles are designed: the
Control, SM Anomaly, and SM_Veg Anomaly. The
Control ensemble is initialized with the soil moisture
climatology, and the SM Anomaly and SM_Veg
Anomaly ensembles are initialized with certain soil
moisture anomalies imposed to the soil moisture climatology. Table 1 lists all ensemble simulations carried
out in this study. Vegetation seasonality in the Control
and SM_Veg Anomaly ensembles is predicted by the
predictive phenology scheme, and is prescribed in each
of the SM Anomaly simulation using model output
JUNE 2007
KIM AND WANG
from the corresponding Control simulation. Therefore,
climate differences between the SM Anomaly ensemble and the Control ensemble are attributed to the impact of soil moisture initialization through soil moisture–precipitation interactions; climate differences between the SM_Veg Anomaly ensemble and the Control
ensemble are attributed to the impact of soil moisture
initialization and vegetation feedbacks; and climate differences between the SM_Veg Anomaly ensemble and
the SM Anomaly ensemble represent the impact of vegetation feedback. The focus in this study is on the role
of vegetation in modifying the impact of initial soil
moisture anomalies.
Apart from the 12-yr initial simulation with climatological SST, a 20-yr simulation driven with interannually varying SST from 1979 to 1998 is available from our
previous study (Kim and Wang 2007). Based on this
20-yr integration, the t statistics are estimated to evaluate the statistical significance of simulated climate differences between two different types of ensembles
(e.g., difference between SM_Veg and SM ensembles)
in section 3. For each grid cell, monthly output from
this 20-yr simulation is used to derive the 90% confidence interval in the significance tests of monthly results over the 2D spatial domain. Daily output is used to
derive the 90% confidence interval in the significance
tests of the daily and 10-day running averaged results
over the Mississippi River basin. Here the simulation
with interannually varying SST is used to get a more
realistic estimate of the interannual variability of climate over our study domain. The simulation with climatological SST underestimates the interannual variability of climate over land, which if used would cause
the statistical significance to be spuriously overestimated.
3. Result analysis
Our previous study (Kim and Wang 2007) showed
that characteristics of soil moisture anomalies, including their timing and direction, influence the resulting
precipitation response. Since vegetation is limited by
different factors (soil moisture, temperature, and/or
photoperiod) during different seasons, the timing of
soil moisture anomalies will influence the vegetation
response. Moreover, the processes and mechanisms giving rise to soil moisture–precipitation feedback are
similar to those underlying the vegetation–precipitation
feedback, leading to the expectation that the impact of
vegetation anomalies on precipitation depends on the
timing and magnitude of such soil moisture anomalies
as well. Together these point to the potential dependence of the soil moisture–vegetation–precipitation
539
feedback on the characteristics of soil moisture anomalies. In this study, we first examine how the impact of
vegetation feedback on the response of precipitation to
soil moisture initialization differs between dry and wet
anomalies, and how it varies with the timing of soil
moisture anomalies (see Table 1 for the list of simulations). We will then analyze the results in greater detail
to develop some process-based understanding.
Vegetation responds to changes induced by initial
soil moisture anomalies in the SM_Veg Anomaly, but
such response is absent in the SM Anomaly. In Fig. 4,
initial wet/dry anomalies in the soil and subsequent
rainfall anomalies lead to an increase/decrease in LAI.
Vegetation responds to initial wet soil moisture anomalies relatively slowly in ensembles starting from mid- or
late spring such as 1 April and 1 May and much faster
in ensembles starting after 1 June. This may result from
the cold temperature stress on vegetation during spring
and the high sensitivity of rainfall to wet soil moisture
anomalies applied in the beginning of June, July, and
August as evident in Fig. 5 (see section 3b for details).
In the case of dry anomalies, regardless of when soil
moisture anomalies are applied, the impact of vegetation is small, and there seems to be some oscillation
between positive feedback (vegetation feedback reinforcing the impact of initial soil moisture) and negative
feedback (vegetation feedback suppressing the impact
of initial soil moisture). That is, compared with the 90%
confidence interval of precipitation differences, the
magnitude of precipitation anomalies in the SM_Veg
ensemble is sometimes larger than that in the SM ensemble, and sometimes smaller. However, overall, the
difference between the two ensembles is small following dry soil moisture anomalies (Figs. 4 and 5). This
lack of strong response to vegetation feedback may be
attributed to dry biases of the model, as detailed in
section 3a. The dry bias in the model causes such a
severe water stress in vegetation that vegetation has
little room to further decrease in response to dry soil
moisture anomalies.
As shown in Fig. 5, in the case of wet anomalies in
May through July, vegetation feedback reinforces the
impact of initial soil moisture on precipitation; that is, a
positive feedback occurs. However, in case of wet
anomalies in April, negative feedback is dominant; that
is, vegetation damps the impact of initial soil moisture
(Fig. 5). Changes in precipitation due to vegetation
feedback are considerable relative to those due to soil
moisture feedback especially during June, July, and
August (Fig. 6). While Fig. 6 presents spatial and temporal averages, the following analysis examines spatial
details about the relative contribution of soil moisture
feedback, vegetation feedback, as well as the detailed
540
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
FIG. 4. Daily LAI anomalies as a response to an 80% increase (gray) and an 80% decrease
(black) of soil moisture climatology applied on (a) 1 April, (b) 1 May, (c) 1 June, (d) 1 July,
and (e) 1 August (SM_Veg-Control or SM_Veg-SM). Each line presents the ensemble mean
of five members averaged over the Mississippi River basin. The shaded area presents the 90%
confidence interval for the daily average of LAI.
pathways of soil moisture–vegetation–precipitation interactions. And we use ensembles starting on 1 June as
an example for the summer months and ensembles
starting on 1 April as an example for spring.
a. Summer
How vegetation feedback modifies the response of
precipitation to summer soil moisture anomalies is investigated with the ensembles starting on 1 June. From
Fig. 7b, first we observe that initial wet soil moisture
anomalies over North America increase LAI particularly over the Mississippi River basin. This is a region
where precipitation is sensitive to initial soil moisture
conditions, causing persistence of anomalies in water
availability. These persistent anomalies of water availability (in precipitation and/or soil moisture) eventually
lead to the response of vegetation since vegetation response is a fairly slow process. Over places where precipitation is not responsive, initial soil moisture anomalies will not cause persistent water availability anomalies, thus no lasting response from vegetation is found.
During summer, water availability is the only factor
limiting the LAI (Fig. 8). Wet soil moisture anomalies
in the SM_Veg Anomaly ensembles therefore cause
LAI to increase over the Mississippi River basin. Second, the increase in LAI lasts throughout the growing
season (longer than four months), as a result of rainfall increase (see Fig. 9) in response to initial wet soil
JUNE 2007
KIM AND WANG
541
FIG. 5. Ten-day running averages of precipitation anomalies as a response to an 80%
increase (gray) and an 80% decrease (black) of soil moisture climatology applied on (a) 1
April, (b) 1 May, (c) 1 June, (d) 1 July, and (e) 1 August. Solid/dash lines represent the
differences between the SM_Veg/SM Anomaly ensemble and the Control ensemble. Each line
presents the ensemble mean of five members averaged over the Mississippi River basin. The
shaded area presents the 90% confidence interval for the 10-day average of precipitation.
moisture anomalies through the positive soil moisture–
vegetation–precipitation feedback. Relative to the SM
Anomaly, changes in LAI and the resulting changes in
precipitation following the initial soil moisture anomalies in the SM_Veg Anomaly further influence soil
moisture as shown in Fig. 7c. On the one hand, the
increase in LAI due to initial wet soil moisture anomalies leads to increase in water consumption by vegetation through transpiration and reduces soil water replenishment through interception loss (not shown),
which tends to reduce soil moisture. On the other hand,
the increase of LAI enhances evapotranspiration,
which favors more precipitation and therefore tends to
increase soil moisture. Whether soil is wetter or drier
in the SM_Veg Anomaly (compared with the SM
Anomaly) depends on the competition between the two
mechanisms. There is no definitive winner, even though
the direct drying impact seems to dominate over vast
areas of vegetation increase (Fig. 7c).
The drying effect of vegetation on soil moisture complicates the response of precipitation to initial soil moisture anomalies, competing with the positive impact of
LAI increase on precipitation (Fig. 1). Between the
two, the impact of increased vegetation on precipitation
seems to dominate the impact of vegetation-induced
soil drying, leading to increase in precipitation as shown
in Fig. 9. The statistically significant increases in precipitation suggest that vegetation feedbacks reinforce
542
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
FIG. 6. Changes in precipitation due to soil moisture feedback (white; SM-Control), vegetation feedback (light gray; SM_Veg-SM), and both soil moisture and vegetation feedback
(dark gray; SM_Veg-Control), averaged over the Mississippi River basin throughout the
second and third months (as indicated inside parentheses) following the 80% wet soil moisture
anomalies applied on 1 April, 1 May, 1 June, 1 July, and 1 August. The error bar presents a
standard deviation among five ensemble members.
the impact of initial wet soil moisture anomalies on
subsequent precipitation in this example. Furthermore,
comparison between Figs. 9b and 9c suggests that increases in precipitation induced by vegetation feedback
(Fig. 9c) are comparable in magnitude with those by
soil moisture feedback (Fig. 9b) especially over the
Mississippi River basin in July and August.
Albedo decreases as soil moisture increases in the
SM relative to the Control as expected (Fig. 10b). Generally, albedo is expected to decrease with the increases
of LAI. However, our results show increases of albedo
(Fig. 10c) as LAI increases (Fig. 7b). This can happen
when vegetation is brighter than the ground surface
(Bounoua et al. 2000). In this specific case, over the
Mississippi River basin, vegetation with relatively high
albedo (i.e., grasses and crops) exists on the dark (prescribed in the model) and wet soil background (due to
wet soil moisture anomalies). Therefore, such increases
in albedo, together with the increased cloudiness that
accompanies the precipitation increase, reduce the total
net shortwave radiation (Fig. 11b). However, the increased cloudiness results in more downward longwave
radiation, and enhanced evapotranspiration cools down
the ground surface, leading to less upward longwave
radiation. These imply an increase in net longwave radiation at the land surface (Fig. 11c). The increase of
longwave radiation outcompetes the shortwave impact
of albedo and clouds, resulting in an increase of net
radiation (Fig. 11a). A similar effect was found by Pal
and Eltahir (2003) who showed an increase in net radiation as a result of a soil moisture increase, with the
longwave radiation impact dominant over the shortwave radiation impact. In addition, the LAI increase
leads to a low Bowen ratio in the SM_Veg Anomaly
ensembles, favoring the increase of latent heat at the
expense of sensible heat (Figs. 11d and 11e).
The large magnitude of soil moisture anomalies (i.e.,
80% increase or decrease of the soil moisture climatology) may be beyond the range of natural variability
(see section 2b). We therefore add another set of ensemble experiments with a smaller magnitude of soil
moisture anomalies. Increases in LAI due to a 30%
increase of initial soil moisture in Fig. 12a are as large
as that in Fig. 7b, indicating that even a 30% increase of
soil moisture climatology is enough for vegetation to
reach its full leaf display. Unless cold stress exists, vegetation reaches its full leaf display once the whole plant
water stress, ranging from zero at the permanent wilting
point to one at saturation, is above a certain threshold
[Wth ⫽ 0.4 in Eq. (6) of Kim and Wang (2005)]. Increases in LAI lead to increases in evapotranspiration,
and therefore decreases in soil moisture (negative feedback from vegetation to soil moisture), which may
eventually lead to a decrease in precipitation (negative
feedback from vegetation to precipitation); the increased evapotranspiration, however, favors precipitation (positive feedback from vegetation to precipitation), which tends to increase soil moisture (positive
feedback from vegetation to soil moisture). Comparison between Fig. 12b and Fig. 7c suggests that the negative feedback from vegetation to soil moisture is dominant in both the 30% and 80% anomaly cases, but it is
more so with the 30% wet anomalies. Between the two
cases, the direct drying impact of vegetation does not
differ much; the wetting impact through precipitation
increases with the magnitude of initial wet soil moisture
anomalies. This is because the extra soil moisture
anomalies beyond a certain threshold (i.e., the thresh-
JUNE 2007
KIM AND WANG
543
FIG. 7. (a) LAI in the Control ensemble (or the SM Anomaly ensemble), and (b) LAI differences and (c) soil water differences
between the SM_Veg Anomaly ensemble and the SM Anomaly ensemble in June, July, August, and September. Only differences
exceeding the 90% confidence level are shaded. The Anomaly ensembles are initialized with an 80% increase of soil moisture
climatology on 1 June. The numbers in the left bottom of each panel indicate averages over the Mississippi River basin.
old whole plant water stress) do not enhance vegetation
growth, but do enhance the wetting impact through
precipitation.
In contrast to the dominant negative feedback from
vegetation to soil moisture, positive feedback from vegetation to precipitation is dominant in both the 30%
and 80% anomaly cases, and it is more so with the 80%
wet anomalies (Fig. 12c versus Fig. 9c). The fact that the
impact of vegetation on precipitation is smaller in the
30% anomaly case than in the 80% anomaly case is
consistent with the stronger negative feedback from
vegetation to soil moisture in the 30% anomaly case
544
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
b. Spring
FIG. 8. Seasonal change of phenology factor [D in Eq. (1)],
Ddrought for drought-deciduousness, and Dwinter for cold-deciduousness, averaged over the Mississppi River basin. The shaded
area presents one standard deviation of monthly D. These are
derived from the 20-yr CAM integration driven with interannually
varying SST from 1979 to 1998. Note that D is always set to be one
for evergreen PFTs.
(Fig. 12b versus Fig. 7c). Note that the sensitivity of
vegetation to soil moisture anomalies depends on a tunable parameter, the threshold whole plant water stress
[Wth ⫽ 0.4 in Eq. (6) of Kim and Wang (2005)]. If this
parameter increases (e.g., from the current value 0.4 to
0.6), the difference in the strength of vegetation feedback between the 30% and 80% anomaly cases will be
smaller.
Initial dry soil moisture anomalies cause LAI to decrease, as expected, but this reduced LAI does not
seem to significantly reduce precipitation (not shown).
As a result, such LAI decrease does not last long, and
is much smaller in magnitude than the LAI increase in
the wet case (Fig. 7a). This insensitivity is likely related
to a dry bias in the coupled model CAM3–CLM3 over
the Mississippi River basin (Bonan and Levis 2006;
Hack et al. 2006). Kim and Wang (2007) also compared
the precipitation and soil moisture between the CAM3–
CLM3 and the North American Regional Reanalysis
(NARR) data, showing a dry bias of the model. For
example, over this region, the Global Precipitation Climatology Project (GPCP) precipitation during June–
August (JJA) is about 2–4 mm day⫺1, about 1–2 mm
day⫺1 higher than the model climatology (http://www.
ccsm.ucar.edu/models/atm-cam/sims/cam3.0). The dry
bias in the Control ensembles leads to severe water
stress in vegetation to such an extent that there is not
much room for further LAI decrease in the SM_Veg
relative to the SM (and the Control). Also, changes in
albedo due to dry soil moisture anomalies are very
minimal (not shown).
The impact of vegetation feedback during spring is
examined using the SM_Veg Anomaly ensembles starting on 1 April as an example. Without considering vegetation feedback, Kim and Wang (2007) found that the
impact of spring soil moisture anomalies on precipitation is not evident until early summer although the impact of anomalies on the large-scale circulation leads to
slight changes in precipitation during spring. This is because the convective rainfall that responds to land surface condition changes does not become the dominant
type of rain over North America until May or June. A
similar delay in vegetation response exists (Fig. 13a),
but for different reasons. The dominant land cover
(grass and crops) in the Mississippi River basin responds to both cold stress and water stress (see Fig. 8).
During spring, vegetation growth is still limited by low
temperature. Vegetation in April, therefore, cannot
take advantage of the increased soil moisture. Instead,
the increase in LAI becomes obvious in May and
reaches its peak in June. Evapotranspiration during
April, however, is enhanced as a result of the wet soil
(based on the comparison between SM and Control;
not shown), but the response of precipitation does not
occur until May or early June. Therefore, soil moisture
is on its way back to normal in April and May in the
SM ensembles, while the enhanced vegetation in the
SM_Veg speeds up this process and may even lead to
dry anomalies in the soil. As a result, precipitation may
decrease, and vegetation feedback may weaken the impact of initial soil moisture, leading to a negative feedback.
Differences in soil water and precipitation between
the SM_Veg Anomaly and the SM Anomaly (Figs. 13b
and 13c) are insignificant during the first two months
(i.e., April and May) as a result of little change in LAI.
In June and July, shaded (statistically significant based
on a t test) negative anomalies suggest that vegetation
feedback tends to weaken the impact of wet spring soil
moisture anomalies, and therefore weaken the summer
precipitation anomalies. Similar negative feedback by
vegetation is also found in simulations initialized with
dry soil moisture anomalies (not shown).
4. Conclusions and discussion
We carried out ensemble simulations using the
coupled CAM–CLM model to examine how vegetation
feedback modifies the impact of initial soil moisture
anomalies on subsequent precipitation over North
America. Vegetation feedback may reinforce or suppress the soil moisture–induced persistence of seasonal
JUNE 2007
KIM AND WANG
545
FIG. 9. (a) Precipitation in the Control ensemble, (b) precipitation differences between the SM Anomaly ensemble and the Control
ensemble, and (c) precipitation differences between the SM_Veg Anomaly and the SM Anomaly ensemble in June, July, August, and
September. Only differences exceeding the 90% confidence level are shaded. The Anomaly ensembles are initialized with an 80%
increase of soil moisture climatology on 1 June. The numbers in the left bottom of each panel indicate averages over the Mississippi
River basin.
climate anomalies through water, energy, and momentum exchanges, depending on timing and direction of
soil moisture anomalies. During summer months, wet
soil moisture anomalies increased LAI, leading to increased precipitation via increased evapotranspiration
and surface heating. That is, vegetation feedback reinforces the impact of initial soil moisture on precipitation. Dry soil moisture anomalies in the summer
months, however, did not show significant impact on
subsequent vegetation and precipitation, which may be
546
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
FIG. 10. (a) Albedo in the Control ensemble, (b) albedo differences between the SM Anomaly ensemble and the Control ensemble,
and (c) albedo differences between the SM_Veg Anomaly and the SM Anomaly ensembles in June and July. Only differences exceeding
the 90% confidence level are shaded. The Anomaly ensembles are initialized with an 80% increase of soil moisture climatology on 1
June.
attributed to the dry bias in the coupled CAM–CLM
model. For wet soil moisture anomalies in spring, vegetation showed delayed response and the vegetation
feedback is negative—during the summer following
spring wet soil moisture anomalies, vegetation feedback tends to suppress the impact of soil moisture on
precipitation.
Vegetation feedback in the coupled soil moisture–
vegetation–precipitation system has been discussed in
recent studies based on observational data analysis
(Notaro et al. 2006; Wang et al. 2006). Note that these
studies are different from ours since they do not specifically examine initial soil moisture anomalies.
Rather, they directly relate vegetation anomalies to
subsequent precipitation without considering how or
why anomalies in vegetation take place. Using remotely
sensed FPAR for vegetation data, Notaro et al. (2006)
estimated vegetation feedback parameter for precipitation in the United States for every season. They showed
that the impact of vegetation on precipitation is spatially inhomogeneous—positive over the corn and soybean belt and negative over the winter wheat belt, while
our present study shows the feedback can be positive
or negative depending on season. Further, Wang et al.
(2006) analyzed the NDVI data over the North American Grasslands during the growing season using
Granger causality test and found that above-average
NDVI leads to lower rainfall during the growing season. Their EOF analysis in the frequency domain further showed that interaction between vegetation and
precipitation tends to suppress each other at short time
scales (less than two months), enhance each other at
long time scales (interannual time scales), and oscillate
at intermediate time scale (four to eight months). Their
finding of negative feedback at the short time scales is
consistent with our results with spring soil moisture
anomalies, while other GCM studies generally disagree
on negative feedback (see the reviews in Notaro et al.
2006). The oscillatory vegetation feedback was detected in our simulated LAI (Fig. 5) as well, although
the magnitude is rather small and the time scale is
shorter than what Wang et al. (2006) found.
The coupling between soil moisture and precipitation
is strong under moderate soil moisture conditions, and
weaker under dry and wet soil moisture conditions in
general (Koster et al. 2006). In other words, model sensitivity depends to a certain degree on the model’s
mean climate. Given the dry bias of CAM3, the response of precipitation to soil moisture feedback and
vegetation feedback, therefore, may change if the
model bias is reduced (Koster et al. 2006).
The impact of vegetation feedback is studied under
soil moisture anomalies that are applied throughout the
whole soil depth in the model (⬃3.4 m). Given that
JUNE 2007
KIM AND WANG
547
FIG. 11. Differences in (a) net radiation, (b) net shortwave radiation, (c) net longwave radiation, (d) latent heat, and (e) sensible heat
between the SM_Veg Anomaly ensemble and the SM Anomaly ensemble, averaged through the JJA season. Only differences exceeding the 90% confidence level are shaded. The Anomaly ensembles are initialized with an 80% increase of soil moisture climatology on
1 June.
different types of vegetation have different rooting
depth, therefore respond selectively to soil moisture
anomalies at different depth, theoretically the magnitude of the impact of vegetation feedback may vary
with the depth of initial soil moisture or with the dominant vegetation type. However, over the Mississippi
River basin, which is the part of our model domain
where precipitation and vegetation are most responsive
to initial soil moisture anomalies, the land cover is
dominated by grass and crops. Their root system is
fairly shallow, mostly residing in the top 1 m of the soil.
As shown in Kim and Wang (2007), reducing the depth
of soil moisture anomalies to about 0.83 m (the top
seven layers in the model) does not significantly influence the precipitation response. These two together imply that reducing the depth of soil moisture anomalies
to 0.83 m will not significantly influence the strength of
vegetation feedback. Over places where trees dominate
(e.g., the U.S. East), the hydrological regime is probably too wet to support a strong coupling between soil
moisture and precipitation; therefore, the applied soil
moisture anomalies will not persist, causing the lack of
persistent response in vegetation.
Our model simulates the seasonal variation of LAI in
response to natural hydrometeorological conditions.
The impact of other dynamic processes operating at the
seasonal time scale such as fire and irrigation were not
considered. Fire tends to take place more frequently
under a drought condition, which if considered would
reinforce the vegetation response to the hydrological
anomalies, and therefore reinforce the significance of
vegetation feedback. Irrigation, which can be important
for Midwest croplands, can wipe out a dry anomaly
applied to the system and will also reduce the difference between wet and normal conditions as farmers are
likely to irrigate more during normal years than during
wet years. Irrigation therefore would reduce the response of precipitation to natural soil moisture anomalies and reduce the significance of vegetation feedback.
The phenology scheme used in this study predicts
LAI based on environmental stress factors and the annual maximum leaf area index. The latter is a spatially
varying, PFT-specific parameter, and represents the
idealized peak growing-season LAI that would occur in
absence of environmental stress, or the “potential
LAI.” In our study the annual maximum LAIs are derived from MODIS LAI data and stay constant regardless of what soil moisture anomalies are considered. In
reality, for deciduous woody plants and perennial grass,
this “potential LAI” depends largely on nonstructural
carbon reserves in perennial tissues at the beginning of
the growing season, which results from carbon dynamics of the previous year. The level of environmental
stresses during the current growing season do influence
the “potential LAI,” but to a lesser degree. Ideally, one
can combine a vegetation dynamics model (that func-
548
JOURNAL OF HYDROMETEOROLOGY
VOLUME 8
FIG. 12. (a) LAI differences, (b) soil water differences, and (c) precipitation differences between the SM_Veg Anomaly ensemble and
the SM Anomaly ensemble in June, July, August, and September. Only differences exceeding the 90% confidence level are shaded. The
Anomaly ensembles are initialized with a 30% increase of soil moisture climatology on 1 June. The numbers in the left bottom of each
panel indicate averages over the Mississippi River basin.
tions at the interannual time scale or longer) and a
phenology model (that functions at the seasonal time
scale) to get a more accurate estimate of the annual
maximum LAI. Specifiying this “potential LAI” based
on observations may overestimate or underestimate
LAI throughout the whole simulation period. However,
its impact on the seasonality of LAI and on the relative
comparison between, for example, SM and SM_Veg
ensembles in this study may be small. It is also less
problematic in this specific study as our focus is on the
general mechanism involved in soil moisture–vegetation–precipitation interactions. For studies that focus
JUNE 2007
KIM AND WANG
549
FIG. 13. (a) LAI differences, (b) soil water differences, and (c) precipitation differences between the SM_Veg Anomaly ensemble and
the SM Anomaly ensemble in April, May, June, and July. Only differences exceeding the 90% confidence level are shaded. The
Anomaly ensembles are initialized with an 80% increase of soil moisture climatology on 1 April. The numbers in the left bottom of each
panel indicate averages over the Mississippi River basin.
on the role of vegetation feedback in specific historical
climate events, such as the 1988 drought or 1993 flood
in the United States, it will be more important that the
“potential LAI” be estimated using a dynamic vegetation model driven with the climate forcing from the
year preceding the event of interest before a phenology
model is used to predict the seasonality of LAI. These
issues will be tackled in future research.
Acknowledgments. The authors thank Dr. Michael
G. Bosilovich at the NASA Global Modeling and Assimilation Office for helpful input and for providing the
550
JOURNAL OF HYDROMETEOROLOGY
mask data of the Mississippi River basin. The authors
also thank Dr. Samuel Levis at NCAR, Dr. Michael
Notaro at the University of Wisconsin, and two anonymous reviewers for their constructive comments on earlier versions of the manuscript. This work is supported
by the NOAA GEWEX Americas Prediction Project
program (NA03OAR4310080).
REFERENCES
Bonan, G. B., and S. Levis, 2006: Evaluating aspects of the community land and atmosphere models (CLM3 and CAM3) using a Dynamic Global Vegetation Model. J. Climate, 19,
2290–2301.
Bosilovich, M. G., and W. Y. Sun, 1999: Numerical simulations of
the 1993 Midwestern flood: Land–atmosphere interactions. J.
Climate, 12, 1490–1505.
——, and J.-D. Chern, 2006: Simulation of water sources and
precipitation recycling for the MacKenzie, Mississippi, and
Amazon River basins. J. Hydrometeor., 7, 312–329.
Bounoua, L., G. J. Collatz, S. O. Los, P. J. Sellers, D. A. Dazlich,
C. J. Tucker, and D. A. Randall, 2000: Sensitivity of climate
to changes in NDVI. J. Climate, 13, 2277–2292.
Buermann, W., D. Jiarui, X. Zeng, R. B. Myneni, and R. E. Dickinson, 2001: Evaluation of the utility of satellite-based vegetation leaf area index data for climate simulations. J. Climate, 14, 3536–3550.
Chase, T. N., R. A. Pielke, T. G. F. Kittel, R. Nemani, and S. W.
Running, 1996: Sensitivity of a general circulation model to
global changes in leaf area index. J. Geophys. Res., 101 (D3),
7393–7408.
Collatz, G. J., J. T. Ball, C. Grivet, and J. A. Berry, 1991: Physiological and environmental regulation of stomatal conductance, photosynthesis, and transpiration: A model that includes a laminar boundary layer. Agric. For. Meteor., 54, 107–
136.
——, M. Ribas-Carbo, and J. A. Berry, 1992: Coupled photosynthesis-stomatal conductance model for leaves of C4 plants.
Aust. J. Plant Physiol., 19, 519–538.
Collins, W. D., and Coauthors, 2004: Description of the NCAR
Community Atmosphere Model (CAM3.0). NCAR Tech.
Note NCAR/TN-464⫹STR, Boulder, CO, 226 pp.
Dai, Y., and Coauthors, 2003: The common land model. Bull.
Amer. Meteor. Soc., 84, 1013–1023.
Dickinson, R. E., M. Shaikh, R. Bryant, and L. Graumlich, 1998:
Interactive canopies for a climate model. J. Climate, 11, 2823–
2836.
Dirmeyer, P. A., 1994: Vegetation stress as a feedback mechanism
in midlatitude drought. J. Climate, 7, 1463–1483.
Guillevic, P., R. D. Koster, M. J. Suarez, L. Bounoua, G. J. Collatz, S. O. Los, and S. P. P. Mahanama, 2002: Influence of the
interannual variability of vegetation on the surface energy
balance—A global sensitivity study. J. Hydrometeor., 3, 617–
629.
Hack, J. J., J. M. Caron, S. G. Yeager, K. W. Oleson, M. M. Holland, J. E. Truesdale, and P. J. Rasch, 2006: Simulation of the
global hydrological cycle in the CCSM Community Atmosphere Model version 3 (CAM3): Mean features. J. Climate,
19, 2199–2221.
Kim, Y., and G. L. Wang, 2005: Modeling seasonal vegetation
variation and its validation against Moderate Resolution Im-
VOLUME 8
aging Spectroradiometer (MODIS) observations over North
America. J. Geophys. Res., 110, D04106, doi:10.1029/
2004JD005436.
——, and G. Wang, 2007: Impact of initial soil moisture anomalies
on subsequent precipitation over North America in the
coupled land–atmosphere model CAM3–CLM3. J. Hydrometeor., 8, 513–533.
Koster, R. D., and Coauthors, 2004: Regions of strong coupling
between soil moisture and precipitation. Science, 305, 1138–
1140.
——, and Coauthors, 2006: GLACE: The Global Land–Atmosphere Coupling Experiment. Part I: Overview. J. Hydrometeor., 7, 590–610.
Lin, S. J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132, 2293–2307.
——, and R. B. Rood, 1996: Multidimensional flux-form semiLagrangian transport schemes. Mon. Wea. Rev., 124, 2046–
2070.
Lu, L., R. A. Pielke, G. E. Liston, W. J. Parton, D. Ojima, and M.
Hartman, 2001: Implementation of a two-way interactive atmospheric and ecological model and its application to the
central United States. J. Climate, 14, 900–919.
Notaro, M., Z. Liu, and J. W. Williams, 2006: Observed vegetation–climate feedbacks in the United States. J. Climate, 19,
763–786.
Oglesby, R. J., and D. J. Erickson III, 1989: Soil moisture and the
persistence of North American drought. J. Climate, 2, 1362–
1380.
Oleson, K. W., and Coauthors, 2004: Technical description of the
Community Land Model (CLM). NCAR Tech. Note NCAR/
TN-461⫹STR, 174 pp.
Pal, J. S., and E. A. B. Eltahir, 2001: Pathways relating soil moisture conditions to future summer rainfall within a model of
the land–atmosphere system. J. Climate, 14, 1227–1242.
——, and ——, 2003: A feedback mechanism between soil moisture distribution and storm tracks. Quart. J. Roy. Meteor.
Soc., 129, 2279–2297.
Pielke, R. A., R. Avissar, M. Raupach, A. J. Dolman, X. Zeng,
and A. S. Denning, 1998: Interactions between the atmosphere and terrestrial ecosystems: Influence on weather and
climate. Global Change Biol., 4, 461–475.
Shukla, J., and Y. Minz, 1982: Influence of the land surface evapotranspiration on the earth’s climate. Science, 215, 1077–1099.
Tian, Y., R. E. Dickinson, L. Zhou, R. Myneni, M. Friedl, C.
Schaaf, M. Carroll, and F. Gao, 2004: Land boundary conditions from MODIS data and consequences for the albedo of
a climate model. Geophys. Res. Lett., 31, L05504, doi:10.1029/
2003GL019104.
Tsvetsinskaya, E. A., L. O. Mearns, and W. E. Easterling, 2001:
Investigating the effect of seasonal plant growth and development in three-dimensional atmospheric simulations. Part I:
Simulation of surface fluxes over the growing season. J. Climate, 14, 692–709.
Wang, G. L., Y. Kim, and D. G. Wang, 2007: Quantifying the
strength of soil moisture–precipitation coupling and its sensitivity to surface water budget changes. J. Hydrometeor., 8,
551–570.
Wang, W., B. T. Anderson, N. Phillips, R. K. Kaufmann, C. Potter, and R. B. Myneni, 2006: Feedbacks of vegetation on summertime climate variability over the North American Grasslands. Part I: Statistical analysis. Earth Interactions, 10.
[Available online at http://EarthInteractions.org.]