ࡱ> hjg I{bjbj "jl,,,,,Xxxxxxxxxbdddddd$ xxxxx xx]   xxxb xb * FFxl Pc3,* FFs0F{ 4{F Water, Energy, and Carbon Exchange in Forest Systems Presiding; B Law, Oregon State University; P Thornton, National Center for Atmospheric Research; D Baldocchi, University of California, Berkeley Annotated reviews by H. Loescher, J. Styles Oregon State University The contributions to this session demonstrated a wide variety of approaches to understanding plant-atmosphere exchange, from ground-based measurements and modeling to satellite data and global GCMs. Interaction between biologists and atmosphericists, and between measurements and modeling, seems to be increasing and leading to new insights and better understanding. The importance of disturbance history (fires, land-use change) is receiving much attention. Eddy covariance measurements have now been made over a long enough period that corrections, interpolations and interpretations are now becoming standard and reliable, and estimates of annual budgets can be made with some confidence. Regional scale modeling approaches are starting to incorporate eddy fluxes and local scale information. Projects are underway to incorporate cross-scale information and multiple constraints within a single modeling framework at even larger scales. As many of the papers in this session demonstrate, model parameterization is an evolving process, as new important drivers are revealed and old ones are better understood. Introduction- Goulden-Ecological controls on land-atmosphere exchange-They outlined our current understanding of key processes governing ecosystem productivity, i.e., the usual suspects (LAI, microclimate, whole plant physiology, litter stocks and quality, plant functional type, allocation, phenology, etc.). However, in examining the role of fire across a black spruce chronosequence in Canada, ecosystem respiration (Re) in a 3-year stand was not higher than that of older stands, and canopy photosynthesis was as high, if not higher than older stands. It is common understanding that after fire, increases in detrital pools would increase Re, but the length of time of increase varies widely depending on recovery rates. In the young boreal forest, standing dead trees were too dry to decompose, and uptake by ruderals compensated for the loss of photosynthesis by the former tree canopy. In attempt to understand processes that control interannual variability in a freshwater wetland in southern California, Goulden et al. found net uptake differed ( 6 t C ha-1 y-1) between years that had similar microclimate and flooding regimes. Herbivory and shading due to the build-up of detritus was found to reduce productivity. In addition to the usual suspects governing productivity, other factors that may affect productivity include the vertical distribution of litter, plant life history, and herbivory. Xu-Seasonal variations in carbon, water and energy fluxes in an oak/savanna and in photosynthetic capacity of oak leaf in California-reported on preliminary findings from the Vaira and Tonza ranches in California. ~98% of the precipitation fell during the wet season (Oct April). In the dry summer (after DOY 140), soil moisture at surface layer dropped to less than 5% and grass senescenced as a result. During this time, carbon uptake reduced to zero, but Re continued at a very slow rate. Re increased after rain events (large efflux), and was interpreted as soil water enhanced microbial activity. Q10 relationship between Re and soil temperature were only significant over a short time intervals, and was found to be soil-moisture dependent. During the growing season, the ratio of daily GPP to PAR was higher on cloudy days (indirect light) than on sunny days (direct insolation) because of hyperbolic relationship between Fc and PAR, no direction linkage was made to increases in Re. Based on two-years results, the grassland was a weak annual carbon sink. Knohl-Large carbon uptake by an unmanaged 250-year-old deciduous forest in Central Germany-the overarching question whether forests at a late-successional stage are carbon sinks was addressed in the oldest forest in the EuroFlux network. This forest has a horizontally and vertically stratified canopy, and high amounts of CWD. Source areas from above- and below-canopy can be different due to shear within the canopy under high u*. Under nighttime conditions, low u* and density stratified flow, drainage can be detected through directional analyses. Additional variability in the Re Q10 response can be accounted for by leaf-fall, bud-break, and drought. Uptake by the understory can be seen early in the growing season, before bud-break. Interannual variability in NEP is, in part, due to changes in the onset of the growing season, i.e. 15 d. Preliminary results from Keeling plots show that the isotopic signature can vary across temporal scales, hour to day to season. VPD effects were also seen in the isotopic signature, but were lagged by ~ 4d. Isotope sampling strategies should incorporate these findings. The Hainich forest was found to be a net sink of ~500 gC m-2yr-1. Lindfors-Disturbance effects on the seasonal dynamics of carbon uptake in boreal forest ecosystems-Fire is a major factor that alters the timing and magnitude of NEP in boreal ecosystems. Lindfors et al. presented eddy covariance measurements from a 3-year old grass and shrub site, a 15-year aspen stand, and an ~80-year black spruce in Alaska. Spring soil warming was delayed and reduced with increasing disturbance age. Snow melt was delayed in the mature black spruce forest. The onset of photosynthesis and ecosystem uptake at the 15-year burn was delayed by ~3 weeks compared to the 80 year burn. The leaf out and senescence of the aspen at the 15-year burn lead to abrupt changes in direction of carbon exchange. The ~80-year black spruce forest had an earlier onset of uptake and extended growing season than the 15-year aspen stand which led to increased seasonal photosynthetic uptake. The cooler soil temperatures at the 80-year spruce stand led to decreased respiration compared with the 15 year aspen stand. The 3-year stand was a small source of CO2 throughout the growing season, both the ~80-year and 15-year sites were net sinks. Morgenstern-Effects of climate variability on estimates of annual NEP for a coniferous west coast forest during El Nio/La Nia cycle- Annual NEP from a 53-y-old Douglas-fir stand on the Canadian west coast varied considerably during 1998-2001, which included the El Nio/La Nia cycle of 1998/1999. EC flux data are subject to large biases due to the choice of gap filling methods, energy balance correction, and the criteria for exclusion of low turbulent fluxes. Morgenstern et al. applied a criteria of u* > 0.4 m s-1 and gap-filled with Q10 response for nighttime, and gap-filled using Michaelis-Menton response to PPFD for daytime data. They estimate the random error to be ~ 30 g C m-2 y-1, whereas uncertainties due to systematic error can be as large as 600-800 g C m-2 y-1. Hence, they can reliably identify the environmental factors responsible for the interannual variation in net uptake. The interannual variation was due largely to differences in Re between the four years, the rate and onset of summer productivity. During El Nio year of 1998, high soil temperatures led to high annual Re, while annual gross ecosystem photosynthesis (GEP) was similar to 2000 and 2001. During the La Nia year 1999, cool and cloudy conditions reduced Re and resulted in an annual NEP similar to 2000, when both Re and GEP were high. Finally, in 2001, Re decreased due to cool temperatures, and reduced cloudiness allowed annual GEP to remain high, leading to the high carbon uptake for that year. Daytime photoinhibition can contribute as much as 15% of error. A very convincing approach in addressing the uncertainties first to insure the observed interannul variation is a real trend. Take home message, differences in Re due to the response to ENSO dictates the interannual variability. The annual balance is most sensitive to conditions in May and June because this is when temperature is rapidly increasing. Curtis-Ecosystem respiration in a northern mixed hardwood forest; source dynamics and interannual variation-Carbon losses from forests by autotrophic and heterotrophic respiration are large and may be a primary determinant of regional differences in annual NEP. Respired CO2 from different sources (e.g., microbes, roots, wood, and leaves) respond differently to changes in temperature and moisture. At the UMBS in mid-northern Michigan, Curtis et al. measured Re throughout the year and across three years (1999-2001) from individual source components using chambers and the eddy covariance technique. LAI and bud-break were similar for each year. They partitioned into the major contributing fluxes. Soil respiration varied from a high in 1999 to a low in 2000 and paralleled annual differences in soil temperature. The residuals from this flux were found to be a small function of soil H2O, and most significant in the springtime. Bole respiration was highly variable and species dependent. Soil respiration was ~5 times more sensitive to changes in temperature than bole respiration. Leaf respiration was also variable and species dependent. Scaling to the ecosystem-level was based on LAI dynamics, assumed no within canopy variation and no daytime leaf respiration. Results of scaling to ecosystem gave soil respiration as 60-70% of total, with bole and leaf respiration being ~equal. Largest errors were associated with scaling chamber measurement to the ecosystem. Sum of all respiration components exceeded EC estimated flux by 50%, but follows similar annual trends. Liu-isolating vegetation and soil contributions to energy and carbon fluxes in Siberian forest tundra- EC measurements of carbon and energy fluxes were made from two similar stands of a larch-tundra ecosystem in Siberia. They cleared the vegetation cover and surface soil at the experimental site to ~15 cm depth to assess the influence of the surface vegetation on energy and carbon exchange. They assumed that the contribution of the surface vegetation and soil to ecosystem respiration was equal to night-time flux at the control site minus that at the cleared site. Before the surface vegetation removal, the Rn, H and lE fluxes at the two sites were similar, mid-day Bowen ratios were ~ 2.4, and both sites were carbon sinks at a rate of ca. -0.4 g C m-2 day-1. After removal of the surface vegetation, the albedo decreased from 0.14 to 0.09, roughness length decreased from 0.072 m to 0.015 m. H decreased, lE however, was unchanged due to evaporation from the bare soil that compensated for the lost canopy. The two sites had different rates of carbon uptake -0.79 g C m-2 day-1 for the control site and -0.43 g C m-2 day-1 for the clearing site. CO2 flux from the experimental site was fairly uniform on a daily basis and, at times, was a slight source of carbon suggesting that much of the diurnal and short-term variability in NEP originated from aboveground vegetation and the surface layer. Styles - Improving regional and continental CO2 flux estimates with surface concentration measurements-CO2 concentrations measured just above the vegetation at 17 forested flux tower sites were used in an atmospheric boundary layer budget to infer CO2 flux and compare with eddy covariance measurements. Concentration draw-down was calculated at each site at monthly resolution as average daily minimum minus 24-hour average, with the latter weighted by boundary layer height so that daytime concentrations contribute more. Comparing the concentration term in the boundary layer budget with the flux measured by eddy covariance, scaled with daylength, air density and boundary layer height gave a very strong correlation at both monthly and annual scales. An even stronger correlation (r2=0.98) was found when each month was averaged across all sites, giving an average seasonal spread. This suggests that concentrations measured at the ground (above the vegetation at ~30m) are meaningful and can be interpreted to infer fluxes. Concentrations have a larger footprint than flux measurements, and hence can provide estimates of larger scale fluxes. They may also potentially be used to infer mid-boundary layer concentration for use in regional and global inverse modeling either by use of a transfer function with an aerodynamic resistance and surface flux estimate, or results from the WLEF tall tower suggest there may be a quantitative relationship between surface concentration draw-down and within minus above boundary layer concentration. Helliker-Measurements of regional-scale isotopic discrimination and CO2 flux for the north-central US- They constructed monthly and annual estimates of regional-scale 13C isotope discrimination and CO2 flux using monthly flask measurements from the WLEF tower and 3 different proxies for free troposphere measurements (Niwot ridge, marine boundary layer, and COBRA flights). Daily drawdowns of CO2 to the landscape were readily visible on daily scale, as were influences from possible synoptic events and weather fronts. Comparisons between these ABL flux estimates and those derived from EC were loosely associated. Thus, the ABL approach potentially can obtain regional NEE integrated over longer (monthly) timescales, has the capacity to integrate over complex terrain, and shows promise for isotopic information. Such that the resultant flux weighted by isotope signatures can constrain contributions to the CO2 flux by natural and anthropogenic sources. Problems associated with this approach include, difficultly in determining source areas and assumes no advection. Interpretation of ABL estimated fluxes should be at the top of the CBL, not the forest or specific ecosystem in question. Method relies on linking transport of CO2 to that of H2O, assuming eddy diffusivity for both is the same. H2O flux can be measured or modeled relatively easily using the Penman-Monteith equation or other gradient/ energy balance methods, then CO2 flux can be estimated by the flux-gradient method presented here. Hemming-Results from the stable isotope sampling network; CARBOEUROFLUX- This presentation addresses the ability to partition carbon sources through integrated stable isotopic measurements across sites within the CARBOEUROFLUX network. Monthly flask samples of nocturnal canopy air and biomass were collected during the growing season from 13 sites in 2001 and 18 sites in 2002. Ecosystem discrimination (D13Ce) was estimated as the difference between the d13C signatures of background atmospheric and ecosystem respired CO2 (Re). The d18O and d13C of leaf, stem and soil organic samples were measured, as was the d18O of waters extracted from these samples. More intensive sampling campaigns, including branch-bags, and leaf, trunk and soil incubation chambers, were also conducted at specific sites to examine, more closely, the partitioning of CO2 and d13C sources. Significant variations in the d13C of Re and D13Ce seasonally and across sites were observed. The general seasonal trend toward increased enrichment (by ~2) throughout July and August followed the trends of increased daytime temperatures and decreased daytime relative humidities during these months. Also during July and August significant relationships were noted between D13Ce and the longitudinal and latitudinal distributions of sites. At the end of the growing season this signal was depleted significantly (by ~80 ). In contrast, soil and leaf d13C compositions were relatively stable throughout the season, suggesting that the d13C of photosynthate may be varying dramatically and inducing a significant influence on D13Ce. The most enriched compositions were observed during the mid-season and in the warmer, drier, more continental locations. In attempts to estimate continental-scale terrestrial C fluxes, it has been shown that a large uncertainty (up to 0.5GtC yr-1) is attributable to spatial and temporal variations in D13Ce caused by environmental variability. Again, we see results that confirm seasonal and spatial changes in D13Ce that are clearly influenced by environmental variations, indicating that it is important and possible (at least partially) to account for these changes when estimating C fluxes from/to terrestrial ecosystems. The results also suggest that more information regarding sources and functional processes are available from this technique, and this should be considered when developing sampling strategies and scaling-up processes from plant to ecosystem levels. Ollinger-Modeling forest productivity in a complex forest landscape using high spectral resolution remote sensing and extensive field data- Because of higher degrees of uncertainty with eddy-covariance in complex terrain, and difficulty in validating satellite modeling products (MODIS), this study addressed the ability to remotely determine leaf N concentrations (from AVIRIS, and HYPERION), and linked them with Pnet to estimate ANPP in New Hampshire's White Mountain National Forest. Model predictions were compared with field-measured productivity data from forest inventory plots, as well as with predictions using directly-measured canopy N. They show that Pnet predictions using remotely-sensed canopy N were improved over estimates using aggregated cover-type data. Comparison between ANPP and Foliar N estimates explained ~ 80 and 60 % of the variation using AVIRIS and HYPERION data, respectively. Loss of explanatory power using HYPERION data was due to its lower prediction accuracy for canopy nitrogen. ANPP estimates derived from AVIRIS data explained ~ 84% of the variation when compared to the field-collected data. Canopy N was also found to differ with elevation in deciduous species, but not for coniferous species. Prediction error was mostly related to error in the remote sensing coverage, indicating that model accuracy was largely a function of input data quality. Afternoon Session Thornton - Controls on the seasonal cycle of NEE: Inferring temperature dependence of photosynthesis in evergreen needle-leaf forests using eddy covariance data Earlier work showed model simulations of coupled energy, H2O, C, and N cycles across a climate gradient in evergreen forests were able to capture variability in leaf area and latent heat among sites, but modeled seasonal cycle of NEE compared poorly to eddy covariance estimates. Model bias in NEE seasonal cycle was not responsive to variation in either base rates or temperature dependence of heterotrophic respiration, due to feedbacks between soil organic matter (SOM) decomposition, soil N mineralization, and gross primary production (GPP). Seasonal cycle bias was also insensitive to variation in base rates for maintenance respiration (MR), as GPP varied with MR in these N-limited systems. Seasonal cycle bias showed limited sensitivity to the relative temperature dependencies of Michaelis-Menton coefficients for Rubisco-mediated carboxylation and oxygenation, and to the temperature dependence of Rubisco enzyme activity. Seasonal cycle bias was found to be most sensitive to the modeled seasonal timing of deployment of nitrogen previously retranslocated from senescing leaves, and to the proportion of new growth allocated to woody (low nitrogen) biomass. A demand-based algorithm for deployment of previously retranslocated nitrogen was found to produce significant improvements in the modeled seasonal cycle of NEE, compared to a simpler constant-deployment algorithm. This change resulted in a significant improvement in both seasonal and annual predictions of NEE for the three youngest stands, eliminating the previous double peak in seasonal carbon uptake and increasing the annual total uptake at two sites where it was previously underestimated. These model changes had minimal effects on the seasonal cycle and no discernible effects on the annual NEE predictions at the mid-aged and old sites. Predictions of LAI and latent heat across sites were the same or slightly better, compared to the original model formulation. Relationships between monthly temperature and NEE were significantly improved with the new formulation. Across all sites the R2 for predicted vs. observed monthly NEE improved from 0.30 to 0.63. Osborne- A New Model for Scaling from Leaf Lifespan to Conifer Forest Structure and Function-This study attempts to determine behavior of ancient forests during warm intervals in the Mesozoic and early Tertiary periods based on known relationships between the lifespan, physiology and biochemistry of leaves to that in fossil records. For example, leaf life span has been shown to be a function of foliar N (R2 = 0.7), and ring markedness, in turn, a function of leaf life span (R2 = 0.9). Osborne et al. used these relationships to develop the University of Sheffield Conifer Model (USCM) that simulates conifer carbon, nitrogen, and water fluxes using leaf lifespan, climate and soils as input data. Their model results compare well with the behavior of present day forests across a wide climatic gradient, and so, can be used to infer the behavior- and potential feedbacks of conifer forests in the geological past. This approach seems restricted to conifer forests in the temperate and boreal regions, The model works well for both warm temperate forest ecosystems and has been tested in Taxodium and boreal Larch stands. One of its aims is to examine the costs and benefits of the deciduous verses evergreen leaf habits. Drewry- Large Eddy Simulation of Coupled Water and Carbon Exchange and Transport Through and Above Forest Canopies-Because the representativeness of eddy covariance data from a single tower is in question, they used the LES approach to determine the relative effects of LAI variability on source-sink strength. They assume there is both linear and non-linear dependence of foliar density on momentum, H, lE and C fluxes, and used 3 characteristic vegetation length scales. They find as vegetation becomes more concentrated into denser areas (more clumped foliage), the response of the vegetation-atmospheric exchange becomes more non-linear. This implies similar canopy turbulence effects on eddy covariance flux measurements. Katul- Two-Dimensional Airflow within Canopies on Hilly Terrain: Implications for Flux Monitoring, Inverse Models, and Data Assimilation-Because topography influences almost all aspects of forest-atmosphere carbon exchange, they examine how topography alters the forest-atmosphere CO2 exchange rate when compared to uniform flat terrain through the use of a first order closure model that accounts for the flow dynamics, radiative transfer, and the nonlinear eco-physiology within the canopy on idealized two-dimensional [cosine] ridge. Their results show that the horizontally averaged and vertically integrated uptake departs from flat terrain values by a factor comparable in magnitude to the mean slope. But as they traverse the hill, the surface-normal CO2 eddy flux departs from its flat terrain counterpart by a factor as large as 3. Flow separation can occur on the lee side of the hill causing internal below-canopy circulations and flows that can occur moving upslope. This results in difference between eddy flux and the integrated biological sources and sinks (Sc) supplied by horizontal and vertical advection terms are individually much larger than the Sc terms they wish to measure. Understanding the behavior of advected flows across the hill increase their ability to correct systematically for advection using standard techniques of data assimilation and inverse modeling. Still cannot answer the definitive question; where on a slope should a tower be placed? You can answer this for ONE WIND DIRECTION - as the wind direction shifts by 180 degrees what was the optimum position becomes the worst spot. The fact that separation occurs destroys the summetry of the problem (i.e. eventhough the hill has geometric symmetry, the NEE solution from the flow dynamics are not symmetrical....) Friend- Using Flux Measurements Over Forests to Improve Climate Predictions by the GISS GCM-They described a physiologically-based canopy photosynthesis and stomatal conductance model which is included in the Goddard Institute for Space Studies (GISS) GCM model. The model was parameterized and tested using H, lE, and C eddy covariance flux data across contrasting forest types. They derived a new canopy conductance (gc) relationship based on canopy photosynthesis, VPD and soil moisture. The response of gc to vpd played a major role in increasing the accuracy of predicted lE, NEE of C, surface temperatures, soil moisture and precipitation. The new parameterization of canopy conductance predicted higher canopy conductance than the old model, which led to large effect on predicted temperature response to doubling CO2 new model predicts much lower effect, with some areas decreasing in temperature. Annual GPP was predicted to vary by 76-84 Gt C between years, about 2/3 the value inferred from observations. This difference is due to remaining problems with precipitation in the pan-tropics. GPP was also predicted to increases by ~48% with the doubling of CO2 concentration in the atmosphere. Bachelet - Is the size of the Carbon Sink caused by "Woody Encroachment" of U.S. Grasslands overestimated? One DGVM calculates an answer - Recently Jackson et al. (2002) claimed that woody encroachment could cause losses of carbon and that U.S. carbon budgets may have overestimated its carbon sequestration potential. Researchers here examined the effects of woody encroachment on carbon storage using a dynamic vegetation model that includes a modified version the CENTURY biogeochemistry model. The model simulates both soil processes and the occurrence and effects of wild fires on natural vegetation. The model simulations showed a decrease in soil organic carbon with increased biomass as woody vegetation replaced grass vegetation. At high precipitation sites, the decrease in soil carbon associated with the conversion to woody vegetation was greater than at low precipitation sites, agreeing with Jackson et al. (2002). The negative relationship between annual precipitation and simulated changes in soil carbon mimicked that found by Jackson et al. (2002) but with a less steep slope. Moreover, even though there was an increase in aboveground carbon associated with the conversion to woody vegetation, the total biomass consumed by wildfires was 35 % lower in the shrublands. In conclusion, potential carbon storage in live plant biomass increases when shrublands invade the grasslands as trees can store more carbon and are less likely to be frequently consumed by wildfires. However, soil carbon sequestration decreases in the wetter sites. The study used a U.S. database (~200 sites) and found a similar but weaker relationship between precipitation and soil carbon changes than Jackson et al. (2002) who synthesized worldwide data but focused specifically on 6 desert sites with grazing histories in the SW U.S. 6A l z   wmn#$%*+pqsu1S^_\^tvҺҺ 0JH*aJ 0JH*aJ0JaJ0J56\]aJOJPJQJ^JH*H* 56\]5B*\ph0JaJ5\0J5\aJ 5\aJ0J5\aJB6  k l z $%RS##**1177R<==II{v  ^`bd  >!?!z!{!!!######C$D$$$%%X'Y'***u*v*,,--... .L/N/ 0K0M0Q0S0x0z0~00000111111111122·²H* 56\56B*H*\ph56B*\ph B*phH*0JOJQJaJ5B*\ph 56\]5\ 0JH*aJ 0JH*aJ0JaJE2447788H8J8i8j8.9/97;8;c;e;z<{<<<<<G=H====@@@f@h@l@@@AA AAAAAAAzC|CCCCCCCCDDFFFdGfGjG HHHHHH.K0K4K LLLNNNNbOcORRH*OJQJ5B*\ph 0JH*aJ 0JH*aJ0JaJ 56\]5\H*QINNNMTNT`Tf]g])b6b7bEfFfjjkk]m^mttcttH{I{`RRR`TiTkTUU!]"]f]g]n]o]p]]^B^K^__H_I_Yaaa7b=b>bbddeDfEfFfLfffagbgjj]m^mdmemmm,o.op pppDqFqrrþӾӾޭӾ 0JH*aJ0JOJQJaJH*0JOJQJaJ0J56\]aJ 0JH*aJ0JaJ0J5\aJ 0J5\H*0J56\]aJ0J5\aJ0J5B*\aJph0JaJOJPJQJ^J:rsst$t%tbtcttttI{56OJPJQJ\]^J 56\]5\5B*\ph 0JH*aJ0JaJ 1h/ =!"#$% i8@8 NormalCJ_HaJmH sH tH 2@2 Heading 1$@&5\<A@< Default Paragraph FontO ab<C< Body Text Indent `O tiO! auO1 snOA prj6  klz $%RS&&++11R677T>HAIAJAFFFOOTTTWW6\7\\\^^WcXcccjj00000000000000000000000000000000000000000000000000000v2RrI{CFGIJII{DHI{Ez ( e l r } dn%* S^^bfjjn <KBM. 9 x(z(@)F)..11223(37 7<<==JARABBCCIIeIsIJJJJJJK KKKBLPLQQ!T)TTT(U:UWW)Y2YU\]\^^.^8^A`C`````WbYbXc`c[hehhhjNTy/((,,77CCC DSSTT%]&]``ccj333333333333333 $2R X^jjloeschehC:\AmeriFlux\AGU-notes_4.docloeschehgC:\Documents and Settings\loescheh\Application Data\Microsoft\Word\AutoRecovery save of AGU-notes_4.asdloeschehC:\AmeriFlux\AGU-notes_4.docloeschehC:\AmeriFlux\AGU-notes_4.docloeschehC:\AmeriFlux\AGU-notes_4.docloeschehC:\AmeriFlux\AGU-notes_4.docloeschehC:\AmeriFlux\AGU-notes_4.docloeschehC:\AmeriFlux\AGU-notes_4.docloeschehC:\AmeriFlux\AGU-notes_4.docloeschehgC:\Documents and Settings\loescheh\Application Data\Microsoft\Word\AutoRecovery save of AGU-notes_4.asd+IAj@t()*'9':';r<r=r>r?r@>V>W>_>`>aj@,.`@@B@FHJL@d@np@UnknownGz Times New Roman5Symbol3& z ArialI& ??Arial Unicode MS"qh֊lFJqhW,bP)!20dk|^3qH7Goulden-Ecological controls on land-atmosphere exchangeloeschehloeschehOh+'0$ 8D ` l x8Goulden-Ecological controls on land-atmosphere exchangeoul loeschehcoloesoes Normal.dotl loeschehtl16sMicrosoft Word 9.0c@:@T@ RhW՜.+,D՜.+,x4 hp  Oregon State University,k2 8Goulden-Ecological controls on land-atmosphere exchange Title8_AdHocReviewCycleID_EmailSubject _AuthorEmail_AuthorEmailDisplayNameWT AGU notesilbev.law@oregonstate.eduLaw, Beverly Elizabeth  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKMNOPQRSTUVXYZ[\]^`abcdefiRoot Entry F@5k1TableL{WordDocument"SummaryInformation(WDocumentSummaryInformation8_CompObjjObjectPool@5@5  FMicrosoft Word Document MSWordDocWord.Document.89q