bio-atmo carbon exchange
TRANSCRIPT
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 1/26
1
13 December, 2000
NASA Carbon Cycle Science NRA
Washington, D.C.
To Whom It May Concern:
Enclosed is a proposal to the NASA Research Announcement 00-OES-08,
Quantifying North American sources and sinks of CO2
via synthesis of in situ data,
remote sensing and inverse modeling
Principal Investigators: Kenneth J. Davis, The Pennsylvania State University
A. Scott Denning, Colorado State University
Collaborator: Peter S. Bakwin, NOAA Climate Monitoring and Diagnostics Lab
which should be directed to the Carbon Cycle Science announcement. Thank you.
Sincerely,
Kenneth J. Davis
Associate Professor
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 2/26
2
Quantifying North American sources and sinks of CO2 via synthesis of
in situ data, remote sensing and inverse modeling
Principal Investigators: Kenneth J. Davis, The Pennsylvania State University
A. Scott Denning, Colorado State University
Collaborator: Peter S. Bakwin, NOAA Climate Monitoring and Diagnostics Lab
1. AbstractNorthern hemisphere terrestrial ecosystems appear to absorb a significant portion of the
CO2 emitted by human activity. This terrestrial sink is highly variable from year to year,the mechanisms governing it are not understood, and its potential response to future
climate change is unknown. Land-surface models can predict terrestrial fluxes over large
scales, but large-scale flux observations needed for validation are lacking. Tower fluxmeasurements provide invaluable insight into the processes controlling CO2 exchange but
the observed fluxes cannot be simply extrapolated to regional or continental scales.
Atmospheric CO2 mixing ratio measurements have the potential to bridge the gap inspatial scales between tower-based eddy flux measurements and regional to continental
scale land surface modeling. Current inverse studies have focused primary on annually
averaged CO2 fluxes integrated over latitude bands. Resolving smaller spatial andtemporal scales is limited by a lack of mixing ratio data, especially over the continents.
Our first objective is to establish a network of high-precision, high-accuracy CO2 mixingratio measurements at ten AmeriFlux towers. The measurements will be sub-sampled so
that the data are representative of the continental boundary layer. The result will be a
North American network of CO2 data suitable for regional to continental scale studies of
CO2 sources and sinks over synoptic, seasonal and annual time scales.
Our second objective is to incorporate these data into annual inversion studies to
determine the North American net annual flux of CO2. These new data should reduce theuncertainty in the annual net flux of CO2 over North American by a factor of two.
The final objective of this study is to examine regional and continental CO2 fluxes onsyntopic and seasonal time scales. MODIS vegetation indices and a seasonal inversion
will be used to compute the net North American flux of CO2 at seasonal time scales.
Synoptic and seasonal variability in mixing ratio patterns, combined with MODISvegetation indices, climate data, and AmeriFlux data, will serve to diagnose the driving
forces behind interannual variability in seasonal CO2 fluxes. The result will be an
improved understanding of the mechanisms governing continental-scale fluxes of CO2,including potential responses to climate change.
This proposal addresses this NRA’s call to couple in situ network data, in particular CO2
flux and mixing ratio observations, with remote sensing data to improve regional andcontinental carbon budgets and emission estimates. It also addresses in part the call to
evaluate carbon cycle models across regional to continental scales and develop scaling
methodologies and approaches for coupling local processes to regional and global fluxes.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 3/26
3
2. Technical plan
2.1 Introduction
Human modification of the atmospheric concentration of CO2 is indicative of an era
where the course of human development has global environmental impacts. The draftingof the Kyoto protocol on CO2 emissions represents a first step towards a global response
to the threat of climate change created by rising concentrations of greenhouse gases suchas CO2. Emissions limits carry an economic cost that must be balanced by the potential
costs of climate change in the absence of regulation. Our ability to regulate global CO2
concentrations is dependent on our understanding of natural sources and sinks of CO2, thegross magnitude of which greatly exceeds anthropogenic emissions (IPCC, 1996). At the
present a large fraction of the CO2 emitted by human activity, that is, fossil fuel burning
and deforestation, does not remain in the atmosphere. The terrestrial biosphere is thought
to absorb roughly 1/3 of this CO2, while the oceans absorb another 1/3 of these emissions(IPCC, 1996). Inversion modeling studies suggest that northern hemisphere ecosystems
are a large portion of the terrestrial sink (Tans et al, 1990; Denning et al, 1995).Interannual variability in terrestrial uptake of CO2 is quite large (Conway et al, 1994) butthe processes controlling this sink are not well understood. Isotopic inversion studies
support the hypothesis that northern terrestrial ecosystems play a large role in this sink
(Ciais et al, 1995; Francey et al, 1995). Interest in CO2 emissions regulations as well asbasic science, therefore, motivates the need for methods to determine the net fluxes of
CO2 between the surface and the earth’s atmosphere for geopolitical regions such as
continents and nations, and an understanding of the processes governing those net fluxes.
Current methods of studying earth-atmosphere CO2 fluxes can be divided into three broad
categories. One approach uses temporal and spatial patterns in atmospheric
concentrations of CO2
to infer surface-atmosphere fluxes. Inversion models fall into thiscategory. The atmosphere circulates rapidly around the globe, hence this method is
naturally suited to studying fluxes integrated over large regions. Local fluxes and
processes governing fluxes, however, are difficult to discern. A second approach is toobserve fluxes directly. The most prominent method at the present is eddy covariance;
continuous measurements have been initiated at more than one hundred towers globally.
Tower-based eddy covariance is excellent for examining the processes governing the netfluxes over the relatively local region of the flux footprint. Extrapolating these results
over large regions, however, is problematic. Finally observations of the earth’s biomass
can be used to infer biosphere-atmosphere carbon exchange. Of particular interest to this
study is remote sensing of the earth’s surface, which yields a quantitative description of the biosphere at fine spatial resolution while simultaneously providing global coverage.
Direct observations of the flux of CO2 via remote sensing, however, are not possible at
this time. Remote observations of the land surface must be interpreted by numericalmodels to estimate CO2 fluxes. No one of these approaches is sufficient to gain a
comprehensive understanding of the atmospheric carbon cycle.
We propose an effort whose ultimate goal is to link these methods, and to bridge the gaps
between global inversions and local eddy flux measurements, between net global fluxes
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 4/26
4
and detailed process measurements. In particular, we propose to deploy a trial network of
terrestrial mixing ratio measurements across North America. We will analyze theseobservations, using atmospheric inversion methods bounded by remote sensing of the
earth’s surface, to both “regionalize” eddy covariance flux measurements, and to test the
predictions of land surface biogeochemical models. We expect that the results will
include a reduction in the uncertainty in the North American annual net flux of CO 2 by atleast a factor of two, and a substantially improved understanding of the processes
governing continental-scale net fluxes.
Inversion modeling of atmospheric CO2 mixing ratios has generally been limited to
showing globally-averaged or zonally-average fluxes. Fan et al (1998) attempted acontinental study with an admittedly sparse data set and found the provocative result that
North America emissions of CO2 from fossil fuel burning were balanced by net uptake by
the terrestrial biosphere from 1988 to 1992. This work has been met with substantial
skepticism (e.g. Holland et al, 1999) and has been contradicted by other inversion studies(e.g. Rayner et al, 1999). Nevertheless, Fan et al (1998) raise with great clarity the fact
that CO2 mixing ratio data over the continents is extremely sparse, thus limiting ourability to resolve continental-scale terrestrial uptake in inversion models of the globalCO2 budget. The conclusions of Fan et al (1998) were based essentially on the mean
annual observed CO2 mixing ratio difference between North Pacific and North Atlantic
stations of about 0.2 ppm. This difference is both very small, and the data are far fromthe signal of interest (continental uptake of CO2). The mean annual N. Pacific to N.
American difference is much larger (1-2 ppm, see Table 1). Tall towers (Bakwin et al,
1998) and aircraft (Tans and Bakwin, 1997) have been advocated as methods for
acquiring data over the continents, and these data are being acquired at a limited numberof locations in North America (currently one tall tower, and one aircraft profiling site).
An alternative is the construction of “virtual tall towers,” as has been demonstrated by
Potosnak et al (1999) and Davis et al (1998a). Implementation of such a network, andutilization of the data in concert with existing tall towers is a major focus of this proposal.
The goal is to dramatically increase the density of continental lower tropospheric CO2
mixing ratio measurements suitable for inversion studies.
Long-term ecosystem-atmosphere CO2 exchange observations using the eddy-covariance
flux measurement technique have become widespread following the landmark works of Baldocchi (1988), Wofsy et al (1993) and Goulden et al (1996). This measurement,
when based on a tower and run continuously, provides direct observations of the net
ecosystem-atmosphere exchange (NEE) of CO2 (e.g. Wofsy et al, 1993) as well as other
trace gases, energy and momentum. These direct observations provide a powerfulcomplement to inversion studies. Results to date show net annual fluxes of CO2 over
northern hemisphere terrestrial ecosystems that may be comparable to those necessary, if
extrapolated to global scales, to explain the terrestrial CO2 sink (Valentini et al, 2000;Hollinger et al, in preparation) and perhaps its interannual variability (Goulden et al,
1996). A network of long-term flux stations now spans many of the earth’s continents
and ecosystems (http://www-eosdis.ornl.gov/FLUXNET/index.html) with some notableexceptions. This global flux measurement network, merged with high-accuracy, high-
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 5/26
5
precision CO2 mixing ratio measurements and the methodology of Davis et al (1998a),
provides the basis for the “virtual tall tower” network proposed here.
The surface area represented by eddy-covariance flux measurements, however, is quite
small compared to continental scales, at most a square kilometer. It is impossible,
therefore, to “blanket the earth” with flux towers to quantify the global terrestrial carbonbudget, and extrapolating the NEE of CO2 observed via flux towers to continental scales
is problematic due to the heterogeneity of the earth’s surface. We cannot expect the
magnitude of NEE of CO2 observed at any one tower to be valid over large spatial scales.
The synoptic, seasonal and interannual variability in NEE of CO2 observed at flux
towers however, responds to large-scale climate forcing, is often representative of largeregions. The effective surface “footprint” of the time rate of change of the mixing ratio
of CO2 in the atmosphere is much larger than the turbulent footprint of a flux
measurement. Thus there is the potential for synergy between local flux measurements
made via eddy covariance, and co-located, high-precision CO2 mixing ratiomeasurements. We propose regionalizing the findings of eddy flux towers via a synthesis
of tower-based flux and mixing ratio data, and large-scale observations of vegetationindices and climate.
Land-surface biogeochemical models (e.g. Running et al, 1999; Sellers et al, 1997), can
predict NEE of CO2 over regional to global scales given proper forcing data, particularlyvegetation indices and meteorological conditions. These models, driven by global land-
surface and meteorological data such as can be provided by MODIS and operational
weather models, have the potential of providing global CO2 flux calculations driven by a
process-based understanding of the earth system. Such models have the potential forglobal coverage, mechanistic interpretation of observed variability in the global CO2
cycle, and prediction of future atmospheric CO2 budgets as climate and land use change.
One limitation in development of these models is validation. An intercomparisonbetween MODIS- and climate-data driven land-surface models and AmeriFlux eddy
covariance observations has been initiated. This exercise will test the participating
models at individual tower sites. Regional- to continental-scale validation data is lacking.This proposal will provide data that will become a critical element in the validation of
land surface models operating for synoptic to interannual time scales and regional to
continental spatial scales.
We propose, in addition to creating a North American CO2 mixing ratio measurement
network, to analyze these data using inverse models and simple boundary layer budget
methods. We will conduct global inversion studies of annually averaged mixing ratiodata, focusing on North America. The addition of continental mixing ratio data should
reduce the uncertainty in the net flux from North America by a factor of two or more. A
great deal of information is also contained, however, in the synoptic and seasonal CO2
mixing ratio patterns (see Figures 1 and 2). We seek to extend inversion modeling to
seasonal and synoptic scales. The goal is to reduce the spatial and temporal regimes
covered by inversion studies to scales where comparisons with land surface modeling andtower flux measurements can be readily made, and where the processes governing
variability in the fluxes are resolved in time and space.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 6/26
6
Seasonal inversions will be conducted at global scales. A significant portion of theinterannual variability observed at tower flux sites is caused by year-to-year differences
in the length of growing season, snow cover, and other seasonal scale biological and
climatic phenomena (e.g. Goulden et al, 1996). Seasonal inversions would provide an
independent method of quantifying interannual variations in fluxes integrated overcontinental scales. Remote sensing of the seasonality of the land surface, particularly
vegetation indices, is critical to conducting seasonal inversion studies. Net fluxes of CO2
will be scaled to vegetation indices as they vary in both space and time.
The development of synoptic-scale inversion studies, which would involve assimilationof CO2 mixing ratio data and MODIS-derived vegetation indices into a coupled land-
surface model and mesoscale atmospheric model, is the topic of a separate proposal.
Once such tools are developed, the data gathered under this proposal can be analyzed to
determine regional fluxes over time scales of a few days. For this study we proposeobservational analyses of the spatial and temporal coherence of flux, mixing ratio,
vegetation and climate data on syntopic and seasonal time scales, and Eulerian andLagrangian budget studies of well-defined airmasses as they move across North Americato determine regional fluxes over time scales of days.
2.2 Objectives and their scientific relevance
2.2.1 Objectives
1) Establish a trial North American network of high accuracy, high precision CO2
mixing ratio measurements based at approximately 10 AmeriFlux towers. Proposed
sites are listed in Table 2. Absolute accuracy and inter-site precision in CO2 mixing
ratios of approximately 0.2 ppmv will be achieved.
2) Utilize data from these sites in addition to one South American and one African site
to compute boundary layer CO2 mixing ratios needed to study regional or continentalfluxes of CO2 on synoptic, seasonal and annual time scales. Weekly and monthly
mean mixing ratios will be computed. The precision and accuracy (0.5 ppmv or
better) will be sufficient to resolve annually averaged continental-scale gradients inCO2 mixing ratios (order 2 ppmv), and will easily exceed the accuracy and precision
needed to resolve seasonal and synoptic scale gradients.
3) Utilize the continental CO2 mixing ratio data in a global inversion. The inversionstudy will estimate annual net flux of CO2 from North America. Conduct pseudodata
experiments to study optimal placement of continental CO2 mixing ratio measurement
sites.
4) Utilize the continental CO2 mixing ratio data in combination with MODIS vegetation
indices to conduct a global inversion study that resolves the annual net flux of CO 2
from North America on seasonal time scales.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 7/26
7
5) Explore the spatial and temporal coherence, from local flux sites to continental scales,
of CO2 fluxes, boundary layer mixing ratios, MODIS vegetation indices and climatedata. Include in these analyses data from the high-precision CO2 towers as well as
flux data from other AmeriFlux sites, NOAA CO2 flask data, and airborne or balloon
campaigns that can provide high accuracy CO2 profiles within and above the
continental boundary layer. Focus on synoptic, seasonal and interannual time scales.Diagnose the mechanisms driving observed variability in regional to continental scale
CO2 fluxes and mixing ratios.
6) Attempt to compute regional net CO2 fluxes over synoptic time scales for well-
defined meteorological conditions using Lagrangian and Eulerian boundary layerbudget methods. Compare these values to local flux observations.
2.2.2 Scientific relevance of the objectives
This work will significantly increase the data available for studies of the net annual
exchange of CO2 between the North American continent and the atmosphere viatraditional inversion methods (e.g. Tans et al, 1990). This is especially significant giventhe evidence of a large, highly variable northern terrestrial sink of CO2 (e.g. Conway et
al, 1994), recent studies debating the location of this sink (Fan et al, 1998; Rayner et al,
1999), and the lack of continental CO2 mixing ratio data available to constrain suchfindings.
Methods and data for assessing terrestrial CO2 fluxes at intermediate spatial and temporal
scales are lacking. Seasonal forcing has been shown to be responsible for interannualvariations in the net annual flux of CO2 at the flux tower level (Goulden et al, 1996), but
the mechanisms that cause interannual variability in the atmospheric CO2 budget at
continental and global scales are not well understood. Land-surface models can predictearth-atmosphere CO2 fluxes over intermediate temporal and spatial scales, but data to
validate the output of such models is lacking. Inverse models have previously been
limited to annual time scales and most often to zonally integrated fluxes.
Eddy covariance flux measurements, global atmospheric inversion studies, and remote
sensing of terrestrial biomass are potent, independent methods for studying earth-atmosphere exchange of CO2. Disparate spatial scales, however, limit the synergy among
these methods. Meaningful comparisons are difficult. Linking these independent
approaches by examining the coherence among flux, mixing ratio, satellite and climate
data will strengthen our understanding of the mechanisms governing the globalatmospheric CO2 budget. A mechanistic understanding of the atmospheric CO2 budget is
critical to understanding future climate and our ability to respond to climate change.
2.3 Methods
2.3.1 High-accuracy, high-precision CO2 mixing ratio measurements
Bakwin et al (1995, 1998) have developed a field-tested, automated system for measuring
atmospheric CO2 mixing ratios at high absolute accuracy using commercially available
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 8/26
8
instrumentation (LI-COR 6251) and high-quality CO2 mixing ratio standards. This
methodology was deployed on the WITN TV tower (600m tall) in North Carolina andfunctioned for 7 years, and has operated on the WLEF Public Television tower (450m
tall) in northern Wisconsin (see http://cheas.umn.edu) since 1994. It has been adapted to
the 30m Willow Creek flux tower, also in northern Wisconsin, in support of a
Department of Energy project directed by Bolstad and Davis. While all AmeriFlux sitesmeasure CO2 mixing ratios for calculating eddy-covariance fluxes and the rate of change
in storage, these measurements typically have limited absolute accuracy. Themethodology developed by Bakwin has been proven capable of approximately 0.2 ppmv
precision and absolute accuracy. We propose to deploy similar measurements on five
flux towers within the AmeriFlux network. (see Table 2).
Two surface-layer tower flux sites in North America already possess above-canopy CO2
mixing ratio measurements of this quality and one additional site is currently being
instrumented (see Table 2). One tall tower site now exists, and a second is beinginstrumented. The first goal of this proposal, therefore, is to foster the development of a
test network of ten sites across North America with flux and high-accuracy mixing ratiomeasurements operating simultaneously and continuously. Principal investigators fromeach of the flux towers listed in Table 2 has agreed to share data and host additional
instrumentation (where necessary) to be part of this test network. Initially only 5 of the
sites listed which currently lack high-quality, high-precision CO2 measurements will bechosen for the study. The sites span North America and include many of its ecosystems.
Intercalibration will be assured by using four standard gases at each site, all of them
traceable to Climate Monitoring and Diagnostics Lab (NOAA) primary standards. Sitesalready conducting high-accuracy CO2 mixing ratio measurements will continue their
current protocols, or have them enhanced slightly if needed for uniformity of methods
across the sites. Flask sampling was considered, but the difficulty in matching flasks intime to the continuous LI-COR data stream and the degree of confidence in the LI-COR
based methodology suggests that flask sampling is not necessary. Flask sampling could
be adopted with relatively little additional effort if it was proven to be desirable.
2.3.2 Data sub-sampling and micrometeorological corrections to obtain boundary
layer mixing ratios.
CO2 mixing ratios in the surface layer above terrestrial vegetation can vary between
roughly 350 ppmv to 450 ppmv, depending upon atmospheric stability (e.g. Bakwin et al,
1998). These deviations are very large compared to the average interhemispheric andcontinental scale gradients that exist in the marine boundary layer over time scales of
weeks to months. The annually averaged interhemispheric gradient, for instance, is about
2 ppm (Conway et al, 1994). The annually averaged difference in mixing ratio betweenthe marine boundary layer over the Pacific and the continental boundary layer over North
America is also about 2ppm (Table 1). Surface layer data, therefore, must be treated
carefully when studying large-scale patterns in atmospheric CO2 mixing ratios.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 9/26
9
Data from the continental boundary layer, however, typically a 1-2 km deep layer during
the day, is representative of mixing ratios over a much larger region, and the diurnalcycle is only several ppmv during mid-summer (Bakwin et al, 1998). During midday
when convection is vigorous and mid summer when surface fluxes are the largest, the
offset between surface layer and mid-CBL mixing ratios is only 1-2 ppmv, as uptake of
CO2 due to photosynthesis depletes the surface layer relative to the mid-CBL. Eddycovariance flux measurements are the most reliable at this time of day when mixing is
dominated by quasi-stationary, large-scale turbulence and the rate of change of storage issmall (Yi et al, 1999). At this time of day, CO2 surface layer mixing ratio and flux data
can be used to study synoptic and seasonal scale fluctuations in CO2 with little or no
correction. Table 1 shows the monthly mean CO2 mixing ratio at WLEF at 11m (surfacelayer) and 396m for this sub-sampling method. WLEF is a located in a small clearing
within a heavily forested region of northern Wisconsin. Fluxes are measured at three
levels above ground, 30m, 122m and 396m, and mixing ratios are measured at six levels,
11, 30, 76, 122, 244 and 396m (see Berger et al, 2001). The upper level on the tower,therefore, directly samples the atmospheric boundary layer while the lower levels are
similar to measurements one would find on a typical AmeriFlux tower.
Note first from this table, that when surface layer data is sub-sampled for afternoon
conditions, the difference between the mean monthly mixing ratio at 11m and 396m is no
larger than about 2 ppmv. This difference is a maximum at times when fluxes are large(mid-summer), or when the depth of mixing (Yi et al, 2001) is shallow (winter).
Compare this difference to Figure 1, which shows the seasonal course of the mean
mixing ratio of CO2 at the WLEF tower in Wisconsin, and compares this to the annual
cycle at Mauna Loa for 1997. At the scale of this figure, 11m afternoon data is nearlyindistinguishable from 396m data when it is plotted on the same graph. It is evident from
Table 1 and Figure 1 that the difference between monthly-mean surface layer mixing
ratios and monthly mean boundary layer mixing ratios, when sampled during well-mixed
conditions, is much smaller than both 1) the amplitude of the seasonal cycle in the
continental boundary layer; and 2) the difference between monthly mean marine and
continental boundary layer mixing ratios. Thus the study of seasonal patterns could beaccomplished with surface layer flux and mixing ratio data solely via sub-sampling for
early afternoon conditions. Note that this sub-sampling is necessary. Figure 1 shows that
the surface layer (11 m) monthly mean mixing ratio for all 24 hours of the day is muchdifferent from the boundary layer mean (represented by the 11 m afternoon data, and the
values for 396 m listed in Table 1).
Second, Figure 2 shows synoptic scale patterns of CO2 mixing ratios from September of 1997 at WLEF. A continuous time series of 396m data is plotted, in addition to early
afternoon 11m data. It is further evident from this plot that for synoptic scale variations
in CO2 mixing ratios, surface layer mixing ratios sampled under well-mixed conditions
are indistinguishable from boundary layer mixing ratios. Thus studies of earth-
atmosphere fluxes that utilize mixing ratio measurements, such as Lagrangian budgets as
air flows across North America, can be conducted purely with sub-sampled, surface layermixing ratio data. It is also evident from Figure 2 that a huge wealth of information,
largely unexamined to date, is present in these synoptic scale patterns.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 10/26
10
Finally, we turn to annually averaged continent-to-ocean, and interhemisphere mixingratio gradients, the subject of traditional inversion studies. Table 1 shows that the
magnitude of the monthly mean difference between 11m and 396m afternoon data (0 to 2
ppmv, depending on the month) is significant compared to these gradients (both
approximately 2 ppmv). Table 1 also shows, however, that the annually averaged, earlyafternoon WLEF surface layer mixing ratio is only 0.4 ppmv lower than the early
afternoon boundary layer mean, and only 0.8 ppmv lower than the 24-hour boundarylayer mean. This is the result of winter and summer biases cancelling in the annual
average. It is quite likely, therefore, that useful results could be derived from sub-
sampled surface layer data with no need to apply a micrometeorological correction forthe bias between surface layer and boundary layer data. We maintain, however, that the
methodology for this correction exists and that it is preferable to apply this correction to
the sub-sampled continental surface layer data in attempt to eliminate this bias. This
correction will benefit both seasonal- and annual-scale inversion studies.
Extrapolation to mid-boundary layer mixing ratios will be achieved using mixed-layersimilarity theory (Wyngaard and Brost, 1984). This theory states that for a well-mixedboundary where solar heating of the earth’s surface drives vigorous convection, the mean
vertical mixing ratio gradient is governed via the following expression:
i
C
z
it
i
C
ib
zw
F
z zg
zw
F
z zg
z
C i
**
0÷ ø ö
çè æ
−÷ ø ö
çè æ
−=∂
∂
where C is the scalar mixing ratio (e.g. CO2), F0C and Fzi
C are the surface and entrainment
fluxes of the scalar, zi is the depth of the convective boundary layer, w* is the convective
velocity scale (a function of the surface buoyancy flux and z i), z is altitude above ground
(or, for a forest, above the displacement height) and gb and gt are dimensionless gradientfunctions that depend on normalized altitude within the convective layer.
The difference between surface layer and mid-boundary layer mixing ratios are computedby integrating the flux-gradient relationship across this vertical interval,
,0*0*
00 z z
zg zw
F z
z zg
zw
F
z z
C z
zC C
zABL
zi
t
i
C
z zABL
zi
b
i
C
ii
ABL i∂÷
ø ö
çè æ
−∂÷ ø ö
çè æ
−=÷ ø ö
çè æ
−÷ ø ö
çè æ
=∆ ò ò
where z0 is the altitude of the surface layer measurement, and z ABL is an altitude in the
well-mixed atmospheric boundary layer. Note that the gradient varies linearly with themagnitude of the surface flux (so that in winter, if fluxes are very small, essentially no
correction is required), and that the difference in mixing ratio is proportional to theintegral of the gradient functions. Note also that for the lower half of the boundary layer,
the top-down gradient function is quite small (Moeng and Wyngaard, 1989).
Moeng and Wyngaard (1984; 1989) simulated these gradient functions using large eddysimulations. Davis (1992) conducted limited observational comparisons, but found it
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 11/26
11
difficult determine the functions with precision using aircraft data due to limitedsampling. Davis et al (1998a), using CO2 flux and mixing ratio data from the WLEF tall
tower for one month, calculated the bottom-up gradient function shown in Figure 3. Data
is limited to convective days. The error bars are standard deviations, and result from the
variability in the data used to solve the flux-gradient equation, that is, the CO2 surface
flux and CBL mixing ratio data. Also included in Figure 3 are recent computations of thebottom-up gradient function from a nested forest-boundary layer large eddy simulation
(Patton et al, 2000, 2001), the first nested model of its kind. A leaf area density similar toa closed deciduous forest canopy was used for the forested simulation, and a no-canopy
case was run as a comparison. Turbulent eddies were resolved down to sub canopy
scales in the surface layer. This is a substantial improvement for applying this methodover a forest canopy, as will be the case for many of the proposed sites.
Table 3 shows the computed differences in mixing ratio between the surface layer and themid-boundary layer for various values of surface flux, typical daytime values for the
boundary layer depth and convective velocity scale, and typical flux tower altitudes.
Also included in Table 3 are the estimated random uncertainty in the monthly meanmixing ratio correction calculation, where the uncertainty is estimated by propagating theuncertainty in the observed bottom-up gradient function. One can see that the random
uncertainty, including the turbulent fluctuations in fluxes and mixing ratios, all captured
in the WLEF observations, is small (order 0.15 ppmv). It is evident from Figure 3 thatthe majority of the vertical gradient occurs very close to the surface, so that choosing a
precise upper boundary for this correction is not critical, and a small increase in altitude
above the canopy results in a significant decrease in the vertical gradient. The mostsubstantial source of uncertainty is the choice of gradient function since the WLEF data
(forested site) show a larger gradient function than the LES results over a canopy. The
non-canopy gradient functions match Monin-Obukhov similarity theory close to thesurface. M-O similarity has been extensively fitted to observational data.
We propose as a first step applying the WLEF-observed gradient functions over forested
sites, and the LES results over non-forested sites. Since all the necessary data to performthe corrections are being collected, revisions can be made if future studies refine these
functions further. Also note that the maximum reasonable error that could be made (e.g.
choosing the observed gradient functions when the LES canopy functions should beapplied) is about 50% of the total change in CO2 mixing ratio from surface layer to mid-
boundary layer. Table 1 shows that on a monthly mean basis, this amounts to a
maximum potential systematic bias of roughly 0.5 to 1.0 ppmv between the computedand actual mid-CBL mixing ratio. Furthermore, since the surface layer to boundary layer
gradient changes sign seasonally, this bias should partly cancel on an annually averaged
basis. Since with no correction at all, WLEF data shows a 0.4 ppmv annual bias between
surface layer and mixed layer data, we expect that with a correction applied it isreasonable to expect an annually averaged maximum systematic bias of about 0.2 ppmv
between the corrected surface layer mixing ratios and mid-boundary layer afternoon
values. Note that an additional issue we will address is the relatively small but persistentbias between early afternoon boundary layer mixing ratios and the 24-hour average
boundary layer mixing ratios (see Table 1).
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 12/26
12
Surface layer profile measurements at specific sites will also be utilized where available
(e.g. Hollinger, tethered balloon measurements at Howland Forest) to further refine the
uncertainty in the near-surface gradient function for particular sites, as will continued
observations at WLEF and the Texas tower currently being instrumented. Further LES
studies are not covered under this proposal, but would be utilized if available.
In practice the depth of the convective layer and entrainment flux of CO2 (mixingdownwards from above the mixed layer) will not be known at most sites. If the
integration from surface to mid-CBL mixing ratio is limited to 0.3 to 0.5 z/zi (typically
300 to 800m) the influence of entrainment is small. At midday the CBL depth can beestimated with sufficient accuracy by convective cloud base, or thermodynamic CBL
growth models. Surface buoyancy and CO2 flux, and the surface layer CO2 mixing ratio
will be obtained directly from the instrumented flux towers.
Independent validation of these monthly mid-CBL mixing ratios will be achieved via
several methods. Further calculations of the gradient functions using the observationsfrom the 450m WLEF tower will be conducted. As a methodological test, the nearbyWillow Creek and Lost Creek flux tower data (see http://cheas.umn.edu) will be used to
compute mid-CBL CO2 data that can be compared to WLEF mid-CBL data. Data from
the Texas tall tower (Bakwin, personal communication) that will be operational in 2001will be applied to this question. Comparisons will also be made when aircraft or balloon
data is acquired at tower sites. Direct comparisons will take in future years as part of the
NOAA/CMDL airborne sampling program, and during upcoming balloon or poweredparachute measurements by Birks, Balsley and Davis. It should be noted, however, that
it is difficult to validate a continuously measured monthly mean calculated with
campaign-style in situ data because of the contrast in sampling interval. The most useful
validation points are the WLEF tall tower and the Texas tower, once operational.
A final potential source of error concerning the application of our micrometeorological
correction scheme is the degree to which surface layer data is biased by landscapeheterogeneity. A worst-case scenario would be a flux tower in an ecosystem
characterized by large CO2 fluxes, just downwind from a large, barren region (no net
flux). While the surface layer flux data would show the influence of the local ecosystem,the surface-layer to mixed layer mixing ratio difference would be better described by the
net fluxes from the barren region.
It is critical to point out that three different “influence areas” or “footprints” are important
here. Smallest is the flux footprint of a surface layer flux measurements (e.g. Horst and
Weil, 1992). The maximum upwind extent of the footprint of a surface layer eddy flux
measurement is about 1 km under well-mixed conditions. The next largest relevantfootprint is the area of the earth’s surface that governs the difference in mixing ratio
between the surface layer and the mid-boundary layer. This footprint is roughly the size
of the footprint of a flux measurement at the geometric mid-point of the two altitudes(Horst and Weil, 1999). This altitude, a few hundred meters, leads to a maximum
upwind extent of this flux footprint of 5 to 10 km under convective conditions (Weil and
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 13/26
13
Horst, 1992). Thus if the landscape in a 1km around a flux tower is similar to thelandscape at 5 to 10km radius, the flux and mixing ratio difference footprints are similar,
and it is valid to apply the surface layer eddy flux measurements to correct the surface
layer to mixed layer mixing ratio bias. Heterogeneity will cause an error that will scale
roughly linearly with the degree of heterogeneity in the surface flux.
The largest footprint is that of the region influencing the time rate of change of the mean
boundary layer CO2 mixing ratio. This “influence area” is roughly equal to the meanwind speed times the length of time over which the mixing ratio is monitored, hence can
easily extend hundreds of kilometers upwind of a site. These large scales are precisely
why we have chosen to study CO2 mixing ratios – to provide insight into regional fluxes.This “mixing ratio influence area” does not need to be homogeneous for the
micrometeorological methods to be valid. Heterogeneity in surface fluxes beyond the
mid-boundary layer flux footprint does not influence the CBL vertical gradient.
Table 2 describes qualitatively the degree to which the proposed sites are heterogeneous
on these two scales. As with the canopy gradient functions, corrections could be appliedif the heterogeneity of sites becomes important. Our approach at this point is to 1) limitour work to the most homogeneous sites available while retaining a good spatial
distribution across North America, and 2) emphasize that the importance of this
correction is nearly negligible for seasonal and synoptic-scale mixing ratio gradients, andthat even in the annually averaged case the ocean-continent and interhemispheric
gradients are roughly 5 times larger (see Table 1) than the annually-averaged surface to
mid-boundary layer gradient.
Finally, the influence of local plumes of anthropogenic CO2 must be considered. The
micrometeorological method proposed "corrects" the surface-layer CO2 data for the
influence of local surface fluxes, but does not eliminate the influence of advectedanthropogenic CO2. If the anthropogenic CO2 is regionally well mixed, this is not a
problem since CO2 modeling studies typically include anthropogenic sources at coarse
resolution. If the sampling site is often influenced by strong local sources, however, thecomputed mid-CBL CO2 mixing ratios will not be representative of a regional value.
Potosnak et al (1999) used CO measurements to correct for this effect at Harvard Forest.
We will attempt to avoid this problem by focusing on sites removed from strong localanthropogenic sources of CO2, since adding CO measurements to all sites involves a
large additional cost and work effort
2.3.3 Inverse model study – annual time scale
We will use the CSU GCM to produce global gridded fields of CO2 concentration that
include realistic diurnal and vertical variations over land. This is possible because themodel is coupled to SiB2, which calculates photosynthesis and respiration at the GCM
time step and because the model uses a unique prognostic formulation for the depth of
trace gas mixing in the turbulent PBL (Denning et al, 1996a,b). The transport model hasbeen compared to both global trace gas observations and other models, and found to
perform well (Denning et al, 1999). The virtual tall tower network proposed here will not
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 14/26
14
sample concentrations as monthly means (typically used for inverse modeling), but ratherwill provide data for mid-day, mid-CBL conditions. Generating analogous pseudodata in
the GCM requires parameterization of both CBL turbulence and diurnally-varying fluxes,
and could not be done in a typical inverse model. Note that sampling midday only
observations of the boundary layer at WLEF shows that the monthly mean difference
between midday-only and full day mixing ratios is at most -2 ppmv, and only -0.4 ppmvon an annually averaged basis. Reasonable “scenarios” of the global carbon budget will
be tested, including fossil fuel emissions, air-sea gas exchange, tropical deforestation, andvarious distributions of net terrestrial sinks corresponding to CO2 fertilization, nitrogen
deposition, and recovery from previous disturbance.
We will sample the global pseudodata at the flask stations and under meteorologically
appropriate conditions (mid-day, mid-CBL) above all the AmeriFlux towers. These
virtual samples will be perturbed with noise to represent observational error in the realworld, and these pseudo-observations will be used to estimate regional net carbon
exchange by the Bayesian synthesis inversion method (Rayner et al, 1999). We anticipate
that using the full complement of AmeriFlux sites as virtual tall towers will allow us torecover regional fluxes at a much finer spatial scale over North America than with theNOAA flask data alone. Nevertheless, we realize that such complete data coverage is
unlikely to be obtained in the near future, so we will perform the pseudodata inversion
with multiple subsets of the sites including the subset actually being instrumented.Because we will know the carbon budget of each region (having modeled it to create the
pseudodata), we will be able to quantify the improvement in the fluxes recovered by the
inversion for each virtual tall tower added to the calculation. We will use thesepseudodata inversions to recommend priorities for the deployment of the measurement
equipment in the future implementation of the observing system.
When the mid-CBL CO2
data become available from the selected AmerFlux sites in year2 of this project, we will perform an inversion of the actual data by a similar method. The
WLEF tower, Harvard Forest, and NOBS sites are already collecting the relevant data.
These sites will be used in the inversion beginning with 1995 data. The new sites will beadded to the calculation as they come online. This is expected to allow a much more
reliable estimation of the carbon balance of North America, and also to allow
subcontinental resolution of net CO2 fluxes for the first time. Figure 4 shows an estimateof the expected improvement in the carbon balance of North America assuming that all
AmeriFlux towers were to have high-precision CO2 measurements 24 hours/day. The
improvement is substantial for only modest precision in the mid-CBL mixing ratioobservations. This estimate does not yet take into account that only a subset of sites will
have the additional CO2 data, or that only midday data will be used.
It should be emphasized that a global inversion based on a model that does not include atime-varying boundary layer depth and diurnal variations in net surface fluxes of CO2
might have difficulty using the continental data proposed here, since the continental
boundary layer mixing ratio is not representative of the entire troposphere. Denning et al(1995) suggests that this is also true of the marine boundary layer data to a more limited
extent. Work is underway testing methods to estimate the difference between boundary
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 15/26
15
layer and lower tropospheric mixing ratios from surface data (Davis et al, 1998a). Apreliminary estimate of the 1998 above-CBL mixing ratio at WLEF is shown in Figure 1.
It is possible that the data collected under this study could eventually be used for these
calculations, but the method is still experimental and not a focus of this proposal.
2.3.4 Inverse model study – seasonal time scales
A wealth of information is contained in the seasonal patterns of NEE of CO2, CO2 mixingratio, vegetation cover and climate that are being collected currently, and which would be
enhanced by continental CO2 mixing ratio data under this proposal. An example is
shown in Figure 1. Note, for example, that early in August, while NEE measurementsfrom WLEF still indicate uptake of CO2 by the local ecosystem, the boundary layer CO2
mixing ratio has begun to increase. The local ecosystem does not reach the point where
the daily integral of NEE changes sign, until mid-September, about one month later. Asimilar phase lag occurs in the spring. The phase lag must be due to transport, and these
data provide a strong constraint for forwards or inverse models. We propose to take
advantage of this information by applying the same global model (CSU GCM coupled toSiB2) to estimate net CO2 exchange by the Bayesian synthesis method, but in this casefocusing on seasonal time scales. MODIS vegetation indices will constrain the inversion
by prescribing seasonally varying spatial patterns of greenness. Seasonal inversion of
continental and perhaps regional fluxes for North America will represent major progresstowards bringing the inversion models down to the scales where more direct
interpretation of mechanisms driving net fluxes, and comparison with other flux
measurement methods are both possible.
2.3.5 Spatial and temporal coherence of flux and mixing ratio data
Co-locating CO2 flux and mixing ratio data over North America will broaden our ability
to interpret each data set. For example, 1998 was an early spring, and very dry late
summer in northern Wisconsin. Interannual patterns of CO2 mixing ratio from the 396mlevel of the WLEF tower clearly show an early drop in mid-CBL mixing ratio
corresponding to early onset of photosynthesis, and the early transition to net respiration
in the late summer as water stress decreased canopy photosynthesis. A comparison of May 1998 and August 1998 WLEF flux data to previous years with more typical climatic
conditions shows the early onset both photosynthesis and senescence are evident in the
flux data, mirroring the WLEF mixing ratio data. This comparison indicates thatinterannual variability in NEE of CO2 at WLEF is representative of a large enough regionto alter the regional atmospheric CO2 mixing ratio. Flux data show the mechanism
driving the mixing ratio anomaly, and the mixing ratio data lend credence to the spatial
representativeness of the flux anomaly. Other North American sites (e.g. Black et al,2000) also observed anomalous fluxes in the spring of 1998. The spatial extent of these
patterns can be further quantified by satellite vegetation indices and climate observations.
Preliminary results such as these illustrate the potential synergy between co-located
boundary layer mixing ratio data, the measurements of NEE of CO2 already being
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 16/26
16
collected at the AmeriFlux sites, climate data and MODIS vegetation indices. Thisproject would examine this coherence over North America, using it as a well-
instrumented test-bed. If a large-scale drought occurred, for instance, the impact would
be felt across many AmeriFlux towers. One could not readily quantify the large-scale net
ecosystem exchange of CO2 during the drought by direct extrapolation of the tower flux
data. The impact of the drought on atmospheric CO2, however, could be determined byexamining the temporal and spatial distributions of mid-CBL CO2 mixing ratios. The
mixing ratio data provides a spatial integral, while the flux data provides anunderstanding of the mechanisms driving the atmospheric mixing ratio. The focus of our
analyses would be on the timing and magnitude of basic seasonal events such as green-up
and senescence, the transition from net photosynthesis to net respiration, and interannualvariability forced by climate. These diagnostic and primarily observational investigations
would complement the seasonal inversion studies.
We further seek to explore the linkages among flux, mixing ratio, vegetation cover and
meteorological conditions down to the regional and synoptic scale. Figure 2 shows an
example, only mixing ratio and meteorological data, from the WLEF tower, followingDavis et al (2000). The period selected was September of 1997, when the seasonal phaselag noted above is evident. At this point NEE of CO2 is changing sign, and boundary
layer CO2, when smoothed to show the seasonal cycle is increasing monotonically. The
hourly data, however, show an abundance of synoptic-scale detail. It is clear the CO2
mixing ratio patterns are highly correlated with weather patterns. The magnitude of this
variability exceeds the offset between surface layer and boundary layer mixing ratios by a
factor of ten or more. The monthly increase in CO2 mixing ratio is dominated by brief events that correspond roughly with frontal passages, followed by relatively quiescent
periods where, early in the month, mixing ratios slowly drop, and later in the month, stay
relatively constant. We cannot explain these patterns at this time, but suspect vertical
mixing during frontal passages and slow net ecosystem uptake with little advection orvertical mixing in between fronts (Davis et al, 2000). Combining these results with flux
and mixing ratio measurements distributed across the continent, and interpreting the
network data in the context of the seasonal variations in vegetation cover will clarify thesources of these patterns, and move us towards diagnosing the synoptic scale events that
create the observed seasonal patterns of CO2 fluxes and mixing ratios.
2.3.6 Regional boundary layer budget studies
If in fact we are correct in describing the slow CO2 draw-down period in Figure 2 (e.g.days 249 to 256) as a one-dimensional situation where there is little horizontal advection
or vertical mixing from above, then the change in mixing ratio over this time is caused
entirely by the daily integrated NEE of CO2. The decrease in CO2 over time at WLEF
would represent the integrated influence of an influence area extending hundreds of kilometers upwind (Uliasz et al, 2000). One could attempt to compute a large area flux in
this way. Given a network of mixing ratio measurements across the continent, one could
track an airmass, perhaps a high pressure outbreak spreading from Canada across theeastern U.S., or Gulf air being advected northwards across the Great Plains.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 17/26
17
If there is little vertical venting of the boundary layer on a daily average basis (areasonable first assumption in the absence of frontal passages or significant convective
cloud activity), then for midsummer conditions one might expect a draw-down of
boundary layer CO2 mixing ratio of the order of 2.5 ppmv per day, assuming a net daily
uptake of 2 gC m-2
d-1
, a value typical for mid-summer at WLEF. The draw-down over
this 7 day period is about 15 ppmv, showing surprisingly good, probably somewhatfortuitous agreement with the rough estimate of 2.5 ppm d-1
. By the end of the month,
net uptake at WLEF has stopped and the mixing ratios do not drop in the intervalsbetween rapid increases in CO2 mixing ratios.
Our goal will be to explore the limits of simple Eulerian and Lagrangian budget analyses,and to what degree these analyses are consistent with eddy flux data. These flux
measurements would be particularly suited to comparisons with land surface models,. In
the future, assimilation of the mixing ratio data into a mesoscale atmospheric modelrunning a land-surface biogeochemical model would supercede this simple budget
analysis, but these analytical tools do not exist at this time.
2.4 Expected results, their significance, and their relevance to this NRA
One primary result of this study will be the establishment of a trial flask-quality CO2
mixing ratio network across North America. This data will provide a missing link criticalto coupling CO2 flux and mixing ratio observations with remote sensing data to improve
regional and continental carbon budgets and emission estimates.
A second result will be a factor of two or greater reduction in our uncertainty in the mean
annual North American carbon sink, and the emergence of some ability to attribute the
net annual flux to particular regions of the continent. Further, seasonal-scale inverse
studies will be used to determine seasonal patterns in the North American carbon sink.The development of sub-continental, seasonal resolution in inverse modeling would
represent a major step forward in developing scaling methodologies and approaches for
coupling local processes to regional and global fluxes. The fluxes derived via the inversestudies will be suitable for evaluation of alternative carbon cycle models across regional
to continental scales. This progress is made possible by recent model developments, the
availability of satellite observations of terrestrial vegetation, and the new continental CO2
mixing ratio data.
Final results will be the documentation of the spatial coherence of flux anomalies across
North America and diagnosis of the processes leading to these flux anomalies. Ourunderstanding of the response of the terrestrial carbon budget to climate variability at
continental scales will be improved, as will our ability to predict continental-scale
responses to future climate change. This result meets the call to couple in situ network data, in particular CO2 flux and mixing ratio observations, with remote sensing data to
improve regional and continental carbon budgets and emission estimates.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 18/26
18
3. References
Bakwin, P. S., C. Zhao, III, W. Ussler, P. P. Tans, and E. Quesnell, 1995. Measurements of carbon dioxide
on a very tall tower, Tellus, 47b, 535-549.
Bakwin, P. S., P. P. Tans, D. F. Hurst, and C. Zhao, 1998. Measurements of carbon dioxide on very tall
towers: Results of the NOAA/CMDL program. Tellus, 50b, 401-415.
Baldocchi, D. D., B. B. Hicks, and T. P. Meyers, 1998. Measuring biosphere-atmosphere exchanges of biologically related gases with micrometeological methods. Ecology, 69, 1331-1340.
Berger, B.W., K.J. Davis, P.S. Bakwin, C. Yi and C. Zhao, Tall-tower observations of carbon dioxide fluxes
over a northern forest: Flux measurement methodology. In press, J. Atmos. Oceanic Tech.
Black, T.A., W.J. Chen, A.G. Barr, M.A. Arain, Z. Chen, Z. Nesic, E.H. Hogg, H.H. Neumann and P.C.
Yang, 2000. Increased carbon sequestration by a boreal deciduous forest in years with a warm spring ,
Geophys. Res. Lett . 27, 1271-1274.
Ciais, P., P. Tans, M. Trolier, J. White, and R. Francey, 1995. A large northern hemisphere terrestrial sink
indicated by the13
C/ 12
C ratio of atmospheric CO2. Science, 269, 1098-1102.
Conway, T. J., P. Tans, L. S. Waterman, K. W. Thoning, D. R. Kitzis, K. A. Masarie, and N. Zhang, 1994.
Evidence for interannual variability of the carbon cycle from the National Oceanic and Atmospheric
Administration/Climate Monitoring and Diagnostics Laboratory Global Air Sampling Network. J.
Geophys. Res., 99, 22831-22855.
Davis, K. J., Surface Fluxes of Trace Gases Derived from Convective-Layer Profiles, 1992. Ph.D. dissertation,University of Colorado, NCAR Cooperative Thesis 139.
Davis, K.J., C. Yi, P.S. Bakwin, W.M. Angevine, C. Zhao, J.G. Isebrands and R. Teclaw, 1998a. Profiles,
fluxes and entrainment of carbon dioxide observed from a very tall tower. Proceedings of the 4th
International Symposium on Tropospheric Profiling: Needs and Technologies, September, Snowmass,
Colorado, 73-75.
Davis, K.J., B.C. Cook, W.B. Sea, C. Yi, P.V. Bolstad, J. Martin, P.S. Bakwin, C. Zhao, J.G. Isebrands, R.
Teclaw and V. Gutschick, 1998b. The Chequamegon Ecosystem-Atmosphere Study: Overview and
preliminary results. Proceedings of the 23rd
Conference on Agricultural and Forest Meteorology,
Albuquerque, NM, American Meteorological Society, 51-53.
Davis, K. J., C. Yi, B. W. Berger, R. J. Kubesh, and P. S. Bakwin, 2000. Scalar budgets in the continental
boundary layer. Proceedings of the 14th
Symposium on Boundary Layer and Turbulence, Aspen, CO,
American Meteorological Society, 100-103.
Denning, A. S., I. Y. Fung, and D. Randall, 1995. Latitudinal gradient of atmospheric CO2 due to seasonal
exchange with land biota. Nature, 376, 240-243.Denning, A. S., J. G. Collatz, C. Zhang, D. A. Randall, J. A. Berry, P. J. Sellers, G. D. Colello, and D. A.
Dazlich, 1996a. Simulations of terrestrial carbon metabolism and atmospheric CO2 in a general
circulation model. Part 1: Surface carbon fluxes. Tellus, 48B, 521-542.
Denning, A. S., D. A. Randall, G. J. Collatz, and P. J. Sellers, 1996b. Simulations of terrestrial carbon
metabolism and atmospheric CO2 in a general circulation model. Part 2: Spatial and temporal variations
of atmospheric CO2. Tellus, 48B, 543-567.
Denning, A. S., P.-L. Vidale, L. Prihodko, N. P. Hanan, K. J. Davis, and P. S. Bakwin, 1998. Simulations
and Observations of Forest-Atmosphere Interactions Across Spatial Scales at the WLEF-TV Tower in
Wisconsin, presented at Fall 1998 AGU Meeting, San Francisco.
Denning, A. S., M. Holzer, K. R. Gurney, M. Heimann, R. M. Law, P. J. Rayner, I. Y. Fung, S.-M. Fan, S.
Taguchi, P. Friedlingstein, Y. Balkanski, J. Taylor, M. Maiss, and I. Levin, 1999. Three-dimensional
transport and concentration of SF6: A model intercomparison study (TransCom 2). Tellus, 51B, 266-297.
Fan, S., M. Gloor, J. Mahlman, S. Pacala, J. Sarmiento, T. Takahashi, P. Tans, 1998. A large terrestrialcarbon sink in North America implied by atmospheric and oceanic carbon dioxide data and models.
Science, 282, 442-446.
Francey, R. J., P. P. Tans, C. E. Allison, I. G. Enting, J. W. C. White, and M. Trolier, 1995. Changes in
oceanic and terrestrial carbon uptake since 1982. Nature, 373, 326-330.
Goulden, M. L., J. W. Munger, S. M. Fan, B. C. Daube, and S. C. Wofsy, 1996. Exchange of carbon
dioxide by a deciduous forest: Response to interannual climate variability. Science, 271, 1576-1578.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 19/26
19
Holland, E. A., S. Brown, C.S. Potter, and S.A. Klooster, 1999. Comments on S. Fan, M. Gloor, J.
Mahlman, S. Pacala, J. Sarmiento, T. Takahashi, P. Tans, North American carbon sink. Science, 283,
1815-1815.
Hollinger, D., P. Bakwin, D. Baldocchi, B. Berger, K. Davis, B. Law, X. Lee, T. Meyers, W. Munger, S.
Verma, S. Wofsy, C. Yi, et al. AmeriFlux: Initial results from a network of long-term, terrestrial CO2
flux measurement sites. In preparation.
Horst, T. W., and J. C. Weil, 1992. Footprint estimation for scalar flux measurements in the atmospheric
surface layer. Boundary-Layer Meteorology, 59, 279-296.
Horst, T. W., 1999. The footprint for estimation of atmosphere-surface exchange fluxes by profile
techniques. Boundary-Layer Meteorology, 90, 171-188.
Intergovernmental Panel on Climate Change (IPCC), 1996. Climate change 1995: The science of climate
change, edited by J.T. Houghton et al., Cambridge University Press, New York.
Moeng, C. -H., and J. C. Wyngaard, 1989. Evaluation of turbulent and dissipation closures in second-order
modeling. J. Atmos. Sci., 45, 2311-2330.
Patton, E.G., K.J. Davis, M.C. Barth, C.-H. Moeng and P.P. Sullivan, 2001. Decaying scalars emitted by a
forest canopy: A numerical study. Submitted to Boundary-Layer Meteorology.
Patton, E.G., P.P. Sullivan and K.J. Davis, 2000. On the influence of a forest canopy on top-down and bottom-
up diffusion in the planetary boundary layer. Proceedings of the 14th
Symposium on Boundary Layer and
Turbulence, 7-11 August, American Meteorological Society, Aspen, Colorado, 545-548.
Potosnak, M. J., S. C. Wofsy, A. S. Denning, T. J. Conway, J. W. Munger, and D. H. Barnes, 1999.
Influence of biotic exchange and combustion sources on atmospheric CO2 concentrations in NewEngland from observations at a forest flux tower, J. Geophs. Res., 104, 9561-9569.
Prihodko, L., A. S. Denning, N. P. Hanan, G. J. Collatz, P. S. Bakwin, and K. Davis, 1998. Simulation and
sensitivity analysis of carbon and energy fluxes at the WLEF-TV Tower site in Wisconsin, presented at
Fall 1998 AGU Meeting, San Francisco.
Rayner, P.J., I.G. Enting, R.J. Francey, and R.L. Langenfelds, 1999. Reconstructin the recent carbon cycle
from atmospheric CO2, δ13
C and O2 /N2 observations. Tellus, 51B, 213-232.Running S.W., Gower, S.T., Bakwin P.S., Hibbard K.A., Baldocchi D.D., Turner D.P., 1999. A global
terrestrial monitoring network integrating tower fluxes, flask sampling, ecosystem modeling and EOS
satellite data. Remote Sensing of Environ., 70,108-127.
Sellers et al, 1997. BOREAS in 1997: Experiment overview, scientific results, and future directions, J.
Geophys. Res., 102 , 28731-28769.
Tans, P. P., P. S. Bakwin, W. D. Guenther, 1997. A feasible global carbon observing system: a plan to
decipher today’s carbon cycle based on observations. Global Change Biology 2, 309-318.Tans, P. P., I. Y. Fung, and T. Takahashi, 1990. Observational constraints on the global atmospheric CO2
budget, Science, 247, 1431-1438.
Uliasz, M., 2000. A modeling approach to evaluate aircraft sampling strategies for estimation of terrestrial
CO2 fluxes. Proceedings of the 14th
Symposium on Boundary Layer and Turbulence, Aspen, CO,
American Meteorological Society, 630-632.
Valentini, R., et al., 2000. Respiration as the main determinant of carbon balance in European Forests.
Nature, 404, 861-865.
Weil, J. C., and T. W. Horst, Footprint estimates for atmospheric flux measurements in the convective
boundary layer. Precipitation Scavenging and Atmosphere-Surface Exchange, vol. 2, S. E. Schwartz and
W. G. N. Slinn, coords, Hemisphere Publishing, 1992.
Wofsy, S. C., M. L. Goulden, J. W. Munger, S. M. Fan, P. S. Bakwin, B. C. Daube, S. L. Bassow, and F. A.
Bazzaz, 1993. Net exchange of CO2 in a mid-latitude forest, Science, 260, 1314-1317.
Wyngaard, J. C., and R. A. Brost, 1984. Top-down and bottom-up diffusion in the convective boundarylayer, J. Atmos. Sci., 41, 102-112.
Yi, C., K.J. Davis, P.S. Bakwin, B.W. Berger, and L. C. Marr, 2000. The influence of advection on
measurements of the net ecosystem-atmosphere exchange of CO2 observed from a very tall tower. J.
Geophys. Res., 105, 9991-9999.
Yi, C., K.J. Davis, P.S. Bakwin, and B.W. Berger, 2001. Long-term observations of the evolution of the
planetary boundary layer. In press, J. Atmos. Sci.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 20/26
20
4. Management plan
4.1 Personnel
4.1.1 Penn State effort
The Penn State group is led by Kenneth Davis. He has extensive expertise in flux
measurement methods and boundary layer meteorology, and is a close collaborator withPeter Bakwin of NOAA on the tall tower research that is the basis of the
micrometeorological methods being proposed here. He will work with Research
Assistant Bruce Cook, who will be primarily responsible for creating and installing thehigh-precision, high-accuracy CO2 measurements following the designs of Dr. Bakwin.
Dr. Davis will also supervise a graduate researcher and ½ time postdoctoral fellow whose
tasks will be the collection of flux data from cooperating sites, calculation of midboundary layer mixing ratios from surface layer data, and posting of that data on our
group web site for public access. The original mixing ratio data (not sub-sampled for
CBL mixing ratios) will also be posted. The student, postdoc and Davis will share thework of analyzing the coherence among flux, mixing ratio, vegetation and climate data,and will collaborate on the inversion models with the Colorado State group.
4.1.2 Colorado State effort
Scott Denning leads CSU’s Biocycle Group (http://biocycle.colostate.edu ). He worked
extensively on innovative inverse modeling techniques for the study of the global carboncycle. He will be responsible for overall intellectual leadership in the development and
implementation of the models, mentoring the graduate student assistant, and coordination
and planning of the research activities. A part time research scientist (Marek Uliasz) and
scientific and systems programmer (John Kleist) will support the project’s computationaland data management needs. He will procure and set up the computer system and
software requested, analyze data for the study areas, and administer the computer systems
required to conduct the research. One Graduate Research Assistant will be fullysupported by this project to aid in modeling studies and analysis.
4.2 Work schedule
Year 1: July, 2001 - June, 2002.
1) High-accuracy CO2 instrumentation and measurements: Add high-accuracy CO2
mixing ratio instrumentation including flask sampling to at least three additional fluxtower sites. Measurements initiated in the winter or spring of 2002.
2) Mid-CBL, midday mixing ratio calculations: Acquire flux and mixing ratio data from
sites where surface layer mixing ratio and flux data are already available. Work with datacovering approximately 1995 through 2000. Calculate mid-CBL mixing ratios from
these sites for the period from 1995 through 2000. Results are finalized and made
available in June of 2002. Draft a paper comparing these mid-CBL mixing ratios withtall tower and flask data for the same period. Present preliminary results at a scientific
conference.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 21/26
21
3) Evaluation of micrometeorological methodology: Test midday, mid-CBL mixing ratioderivations using WLEF direct observations. Use existing studies of flux-gradient
relationships in the surface layer of forest canopies, to suggest modifications to the
WLEF-based functions for other canopy architectures. Collaborate with AmeriFlux
investigators (e.g. Hollinger tethered-balloon data) to evaluate surface layer relationships
at specific sites.4) Applications of mid-CBL CO2 mixing ratio data: Analyze optimal locations for
additional high-accuracy CO2 measurements. Consider the benefit of mixed layer dataalone vs. mixed layer and lower free troposphere data. Submit a paper describing the
results of this sensitivity study.
5) Data management: Anonymous web/ftp site for mid-CBL mixing ratio data and relatedwork is established.
Year 2: July, 2002 - June, 20031) High-accuracy CO2 instrumentation and measurements: Add high-accuracy CO2
mixing ratio instrumentation to at least two more flux tower sites. Leave final site
selection open to results of the model sensitivity study. Operations begin in the winter orspring of 2003. Complete one year of high-accuracy mixing ratio measurements at thethree initial flux tower sites. Prepare a report on data quality.
2) Mid-CBL mixing ratio calculations: Calculate mid-CBL mixing ratios from ChEAS
sites, Harvard Forest, Northern Old Black Spruce, and the ARM-CART site for the year2000. Acquire flux data from AmeriFlux investigators for the period during which high-
accuracy CO2 mixing ratio data has been acquired at these sites (roughly calendar year
2002). Calculate mid-CBL CO2 mixing ratios for these sites.3) Methodology evaluation: Paper on the accuracy of mid-CBL mixing ratio calculations
is submitted.
4) Applications of mid-CBL CO2 mixing ratio data: Initiate a trial inversion study based
on the mid-CBL CO2
mixing ratio data derived for the period from 1995 through 2000.Draft paper on continental-scale inversion modeling using mid-CBL mixing ratios.
Evaluation of spatial and temporal coherence of fluxes and mixing ratios begins.
5) Data Management: 1995 through 2000 mid-CBL mixing ratio data from HarvardForest, NOBS, ChEAS sites and ARM-CART are made available, in addition to 2002
data from the three newly instrumented AmeriFlux sites, the Texas tall tower, and the
Santarem, Brazil and Kruger Park, South Africa sites.
Year 3: July, 2003 - June, 2004
1) High-accuracy CO2 instrumentation and measurements: Maintained at all sites.2) Mid-CBL mixing ratio calculations: Continue for all instrumented sites.
4) Applications of mid-CBL CO2 mixing ratio data: Initiate seasonal scale inversion
using all available sites. Draft paper on continental, seasonal-scale inversion using mid-
CBL mixing ratios. Submit paper on observational analyses of the spatial and temporalcoherence of fluxes, mixing ratios, vegetation indices and climate data.
5) Data Management: Processed data continues to be archived on-line.
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 22/26
21
5. Tables and Figures
Table 1: 1997 monthly mean CO2 mixing ratios from the WLEF tower.
Month
CO2 (ppm)
at 11m,early pm only
CO2 (ppm)
at 396m,early pm only
∆CO2 (ppm)
11m – 396m,early pm only
CO2 (ppm)
at 396m,entire day
∆CO2 (ppm)
396m pm – 396mentire day,
1 371.4 370.3 1.1 369.7 0.6
2 371.4 371.2 0.2 371.1 0.1
3 371.4 371.0 0.4 371.0 0.0
4 370.4 370.4 0.0 370.4 0.0
5 368.1 368.2 -0.1 368.3 -0.1
6 355.5 357.3 -1.8 359.4 -2.1
7 348.0 350.2 -2.2 351.1 -0.9
8 346.1 348.1 -2.0 349.3 -1.2
9 354.9 356.2 -1.3 358.0 -1.8
10 365.8 365.6 0.2 366.0 -0.411 370.3 369.9 0.3 369.6 0.3
12 371.5 370.6 0.9 370.2 0.4
Annual
mean 363.7 364.1 -0.4 364.5 -0.4
Annual mean at Mauna Loa, 1997, was 366.7 ppm.
Table 2: (See following page)
Table 3: Calculations of the difference between surface layer and mid-atmospheric
boundary layer CO2 mixing ratios for various bottom-up gradient functions (see Figure 3)
and typical convective layer meterological conditions (zi = 1500m, w* = 1.5 m s-1). Theupper level of each integral is (zABL-d)/zi = 0.5. Displacement height, d, is roughly 2/3
the height of a closed forest canopy and smaller for an open canopy.
∆CO2 (ppm)
no canopy gb
∆CO2 (ppm)
canopy gb
∆CO2 (ppm)
observed gb
CO2 surfaceflux* (ppm m s-1)
z0 = 10m z0-d = 12m z0-d = 30m
Random error(ppmv)
observed gb
-0.1 0.50 0.11 0.25 0.03
-0.2 1.00 0.21 0.50 0.06
-0.3 1.50 0.32 0.75 0.09-0.4 2.00 0.43 0.99 0.12
-0.5 2.50 0.54 1.24 0.15
-0.6 3.00 0.64 1.49 0.17
*1 ppm m s
-1~ 35 µmol m
-2s
-1
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 23/26
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 24/26
21
Figure 1: Seasonal patterns of CO2 mixing ratios and fluxes at WLEF, 1997.
Figure 2: Synoptic scale variability in CO2 mixing ratios at WLEF, September, 1997.
320
330
340
350
360
370
380
244 248 252 256 260 264 268 272
Day of Year
C O 2 ( p p m )
5
10
15
20
25
30
35
40
T e m p e r a t u r e ( o C )
CO2 (30m) afternoon data CO2 (396m) Temperature
0
5
10
15
20
25
30
35
40
244 248 252 256 260 264 268 272
Day of Year
W i n d S p e e d ( m
s - 1 ) ,
W i n d D i r e c t i o n ( d e g / 1 0 )
91
92
93
94
95
96
97
98
99
Wind Speed Wind direction Pressure
340
350
360
370
380
1 2 3 4 5 6 7 8 9 10 11 12Month
C O 2 ( p p m )
-200
-150
-100
-50
0
50
100
C u m .
N E E ( g c m - 2 )
afternoon at 11m whole day at 11m Mauna Loa
CO2 above PBL Cum. NEE
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 25/26
21
Figure 3: Bottom-up gradient functions derived from observations at WLEF and fromlarge eddy simulations.
Figure 4: Pseudodata simulation of the uncertainty in the net annual flux of CO2 fromNorth America as determined using an inverse model that utilizes boundary layer mixing
ratio observations at all AmeriFlux sites. The uncertainty in the net annual continental
CO2 flux is plotted as a function of the uncertainty in the mixing ratio observations.
0
20
40
60
80
100
120
140
160
180
200
0.0 0.1 0.1 0.2 0.2 0.3 0.3
(Z-d)/Zi
G r a d i e n t F u n c
t i o n g b
LES-no canopy
LES-canopy
Observed
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 1 2 3 4 5 6 7 8
Boundary Layer Concentration Uncertainty (ppm CO2)
F l u x U n c e r t a i n t y ( G t C y
r - 1 )
North America
Global
8/2/2019 Bio-Atmo Carbon Exchange
http://slidepdf.com/reader/full/bio-atmo-carbon-exchange 26/26
Extra text
The modeling tools for conducting inversion studies at regional, synoptic scales
(essentially CO2 data assimilation) are to be developed under a separate proposal. For
this study, back-trajectory analyses under atmospheric conditions where vertical mixing
is limited (e.g. advection of cold air driven by a high-pressure system) will be used to
estimate regional-scale fluxes on the time-scale of days.
Tower flux data and remote sensing of continental-scale vegetation data will illustrate themechanisms controlling these fluxes. If successful the virtual tall tower methodology
could be propagated among the 80+ long-term CO2 flux sites being established globally.
Tower flux networks such as AmeriFlux and Fluxnet are conducting direct albeit small-
scale observations of the net exchange of CO2 between various terrestrial ecosystems and
the atmosphere. The spatial scale described by fluxes from an eddy covariance fluxtower is on the order of a square kilometer.
Temporal changes and spatial gradients in atmospheric mixing ratios are driven bysurface fluxes integrated over air mass trajectories, where the spatial scale increases withthe temporal and spatial scales over which mixing ratio changes are monitored. Inverse
models use this principal to derive net earth-atmosphere fluxes of CO2.
The measurements will be made using an established method that has been shown
capable of 0.2 ppmv absolute accuracy with continuous, unattended operation. Seven
sites that have such measurements will also be included in this study.
Surface layer mixing ratio measurements are strongly influenced by local fluxes. The
difference between surface layer and boundary layer CO2 mixing ratios can be substantial
compared to observed interhemispheric and continental-scale gradients. The secondobjective of this study will be to sub-sample the mixing ratio data for well-mixed
conditions, and to apply a flux-gradient relationship to compute mid-boundary layer CO2
mixing ratios from the surface layer data.