This page has been archived. Only use for reference purposes.

2016 AmeriFlux PI Meeting—Poster Abstracts

AgendaParticipants List • For Poster Presenters

Poster Abstracts

#1 Climate and age influence on drought-induced cavitation vulnerability of eastern deciduous tree species

*Michael C. Benson, A. Christopher Oishi, John-Christophe Domec, and Kimberly A. Novick
Contact:, Indiana University

Drought induced water stress, which may cause runaway cavitation of xylem elements, puts many forests at risk for severe mortality. This risk can be estimated through cavitation vulnerability curves that describe the decline of hydraulic conductance (K) as xylem water tension increases. These sigmoidal curves provide water stress thresholds (P50) representing the tipping point between hydraulic resilience and severe cavitation, which can be compared with in-situ xylem water tensions data to determine hydraulic safety margins. Species’ cavitation vulnerability can vary across age and climate, as xylem elements change during development of new growth rings and in response to variable seasonal moisture. However, studies considering both age and climate effects on vulnerability are rare. We hypothesize that climate and age interact to determine phenotypic response of xylem anatomy, and predict that mature species growing in very mesic sites will be associated with the most vulnerable hydraulic systems. Cavitation vulnerability, xylem anatomy, and mid-day leaf water potential (Ψmd) measurements of two key eastern deciduous U.S. tree species, Q. alba (QUAL) and L. tulipifera (LITU), are compared across two chronosequences (5-, 15- and 85-year-old stands) in a wet Western North Carolina (NC) biome and a relatively warmer and drier Southern Indiana (IN) biome.

#2 Flux observations of an insect-induced mortality event

J. William Munger *, Steven C. Wofsy, David A. Orwig, and Chris Williams
Contact:, Harvard University  

Forests in the northeastern United States include large areas dominated by mosaics of oak/maple and hemlock stands, however, long-term survival of hemlock in this region is threatened by Hemlock Wooly Adelgid (HWA), an invasive insect that is fatal to eastern hemlock. At the Harvard Forest in central Massachusetts, two long-term eddy flux towers are measuring carbon exchange in a >100 year old hardwood stand since 1992 (EMS- Ha1) and in a 100-200 year old hemlock stand (Ha2) since 2004. The flux measurements are complemented by vegetation dynamics plots. Carbon exchange at the two sites has distinctly different seasonality. The hardwood site has a shorter carbon uptake period, but higher peak fluxes, while the hemlock stand has a long carbon uptake period extending from spring thaw until early winter freeze. Adelgids have reached the hemlock stand near Ha2 and have been widely distributed in the canopy since spring of 2012. The hemlock canopy in that stand is thinning and net carbon uptake and evapotranspiration have been decreasing since 2012. Adelgids have also been observed in scattered stands near the Ha1 tower, but as of 2015 the trees are still healthy. Because hemlocks stands have different seasonality and provide a distinct soil and sub- canopy light environment, their mortality and replacement by hardwood species will have significant impacts on forest dynamics, carbon balance, and hydrology. Ongoing flux and vegetation plot observations are documenting the changes in forest structure and function as this disturbance progresses.

#3 The seasonal soil respiration-temperature hysteresis relation regulated by photosynthesis and soil moisture

Quan Zhang*, Kimberly A Novick, Richard Phillips, Stefano Manzoni, Andrew C Oishi, Edoardo Daly, Russell L Scott and Rodrigo Vargas
Contact:, Indiana University Bloomington

Soil respiration, as a major component of global carbon cycle, responds to temperature in an exponential manner across most ecosystems, but pronounced seasonal respiration-temperature hysteresis relations have been commonly reported in field measurements, with the underlying mechanisms remaining unclear. Our poor understanding to such hysteresis limit the ability to accurately simulate soil respiration. Here, we applied both numerical analysis and field measurements across various ecosystems, to identify the underlying mechanism and key drives of such seasonal hysteresis. We used numerical model runs to show that both photosynthesis and soil moisture contribute to the formation of such hysteresis, supported by field experiments. A time lag between seasonal photosynthesis and soil temperature series plays an important role in decoupling soil respiration from temperature, providing one reason of the hysteresis. Specifically, the total soil respiration-temperature hysteresis is in clockwise direction when seasonal temperature lags behind photosynthesis, and in counterclockwise direction when photosynthesis lags behind soil temperature. Cross sites analysis shows that the time lag of respiration and temperature correlates linearly with the lag of photosynthesis and temperature (regression function: y=1.11x-2.35, R2=0.86, p<0.05). But when photosynthetic input to belowground part is eliminated at a girdling experiment, hysteresis vanished in the soil respiration-temperature relation, because the time lag of respiration and temperature is suppressed. In water limited ecosystem, soil moisture becomes a dominant factor of soil respiration so that the time lag between soil moisture and temperature provides another reason of the seasonal hysteresis.

#4 Eddy covariance measurements of evaporation over two Lake Erie sites: Does evaporation mostly occur in winter?

Changliang Shao1*, Jiquan Chen1, Housen Chu2, Carol A. Stepien3, Thomas B. Bridgeman3, Kevin P. Czajkowski3, Richard H. Becker3, Zutao Ouyang1
Contact:, Michigan State University

Evaporation (E) is a critical component of the water balance over lakes. This study analyzed the lake surface E process and its drivers by the eddy covariance systems measured continuously datasets at two Lake Erie sites during September 2011– May 2016. The monthly E varied from 5-120 mm. The maximum E month was in August, September and October. The annual E was 635±42 mm at CB, and 604±32 at LI, respectively, whereas the winter period (October-March) E was 189±61 mm at CB, and 178±25 at LI, which meant winter period shared around one third of the annual E. The lake removed 0.62 m water through Lake E annually, equal to 81% of the annual rainfall. We found that there is ~2 months delay between the energy input and the E. The E tended to be smaller in off shore water than shallow water near the shore. These findings suggest that predictive models may need to consider the time lag between E and Rn in order to accurately simulate the dynamics of E over aquatic ecosystems.

#5 Assessing the climate impacts of land cover and land management using an eddy flux tower cluster in New England

Andrew Ouimette*, Lucie Lepine, Scott Ollinger, Elizabeth Burakowski
Contact:, University of New Hampshire

Land use, land management, and land cover change influence both local and global climate through changes in net greenhouse gas exchange, albedo, evapotranspiration, and aerodynamic roughness. We used a cluster of eddy flux towers located within 7.5 km of one another in southern New Hampshire across four land cover types—mixed forest, cornfield, hayfield, and impervious surface (parking lot)—to investigate the impact of land use on land-climate interactions in New England. Preliminary results from 2013-2016 indicate that the forested site had the lowest annual and growing season albedo (i.e. it absorbed more incoming radiation than the other three land cover types) but maintained the lowest surface temperature. The low surface temperature appears to be maintained due to the comparatively high aerodynamic roughness and exchange of sensible heat compared to the other three land cover types. Contrary to expectations, the agricultural sites had similar latent heat fluxes to the forested site but comparatively lower sensible heat fluxes. The forested site acted as a moderate C sink during the 3-year period we examined. While eddy covariance measurements indicate that the agricultural sites were also moderate C sinks during this period, when we accounted for C added and removed during management, the agricultural sites were near carbon neutral. The cluster of flux tower sites here offers the possibility to more directly assess the impact of land use and management on land-climate interactions. Preliminary results suggest tradeoffs between biophysical (albedo) and greenhouse gas forcings that have relevance to global climate, and also highlight the importance of aerodynamic roughness for influencing local surface climate.

#6 Linking forest carbon mass increment and CO2 flux: 20 years of eddy covariance data from the Howland Forest, Maine

Aaron Teets*, Shawn Fraver, David Hollinger
Contact:, University of Maine

Forests play a pivotal role in the terrestrial carbon cycle by sequestering and storing carbon in woody biomass. By converting atmospheric CO2 to long-lived biomass, forests can mitigate the effects of elevated atmospheric CO2 by offsetting emissions from burning fossil fuels. Technological advances have allowed researchers to measure stand-level CO2 exchange (flux) using relatively non-invasive techniques, namely eddy flux towers. However, there are important knowledge gaps concerning the relationship between CO2 flux and tree growth. These knowledge gaps exist in part because most eddy flux towers have not been operating long enough to determine how these factors co-vary through time. The purpose of this study is to clarify the relationship between CO2 flux measurements and stand-level aboveground carbon mass increment. The study benefits from the second longest running eddy flux tower in the US (Howland Experimental Forest, Maine; flux data back to 1996), which provides a data series long enough for robust analyses of these relationships. We re-mapped and re-inventoried a three hectare spruce–hemlock forest plot at Howland Forest (established in 1989) to determine mortality and ingrowth. We cored a subset of the mapped trees (10%) to estimate annual carbon mass increment for each tree. We compared total carbon mass increment from tree growth to annual carbon flux from the eddy flux measurements to understand year-to-year relationships. Our results concerning this link will strengthen our understanding of forest CO2 exchange and the terrestrial carbon cycle.

#7 Past, Present, and Future Research at the Rosemount, Minnesota

John Baker*, Tim Griffis*, Jeff Wood, Peter Ganzlin
Contact:, University of Minnesota/USDA-ARS

AmeriFlux Site Research at the Rosemount, Minnesota site over the past 14 years has focused on the carbon, water, and reactive nitrogen budgets of agricultural ecosystems. Estimates of Net Biome Productivity (NBP) for our conventional cropping system (corn-soybean rotation) indicate a substantial loss of carbon (773 g C m-2) integrated over the past 11 years (2005 to 2015). Alternative cropping systems (involving cover crops and reduced tillage) have been more carbon neutral, attributable in part to less carbon removal in grain (lower yields). The conventional and alternative cropping systems have been compared using the radiative forcing (RF) framework. The inclusion of cover crops into conventional corn-soybeans gives rise to a local albedo forcing of -3 W m-2 at the top of the atmosphere, which could be of regional to global significance if the practice is widely adopted. The carbon benefits, do however, appear to be more muted because these alternative systems are still intensively managed. Beginning in 2007 our research was extended to include observations from the KCMP tall tower at Rosemount. Regional scale analyses of NBP, gross primary productivity (GPP), nitrous oxide, and water vapor have been ongoing. Using Solar Induced Fluorescence (SIF) observations from NASA’s Orbiting Carbon Observatory (OCO-2) satellite and tall tower GPP, we have shown that GPP scales linearly with SIF from instantaneous to monthly time scales. This finding has important implications for reducing the uncertainty in GPP at local to regional scales. Finally, using Bayesian inverse analyses over the period 2010 to 2015, we have shown that nitrous oxide emissions from the US Corn Belt varied from 315 to 555 Gg N2O-N y-1 and were substantially larger than bottom-up estimates derived from emission inventories. Direct measurement of nitrous oxide emissions from streams within the region demonstrated that low order streams (tile drains, ditches, and rivers associated with agriculture) have a strong influence on the regional budget and help reconcile the differences between bottom-up and top-down approaches. Over the next three years our research at the Rosemount AmeriFlux site will evaluate:

  1. How restoring croplands to prairie can influence the local and regional carbon and water budgets;
  2. How beef and dairy farming influence the region’s methane budget;
  3. How ammonia emissions and deposition impact the region’s reactive nitrogen budget and indirect emissions of nitrous oxide.

#8 Comparison of carbon flux estimates using 10 years of eddy covariance data and plot-level biometric measurements from the Bartlett Experimental Forest, New Hampshire

Andrew Ouimette*, Scott Ollinger, Andrew Richardson, Trevor Keenan, David Hollinger
Contact:, University of New Hampshire

In the northeastern United States, forest regrowth following 19th and 20th century agricultural abandonment represents an important carbon (C) sink. The ability of these secondary forests to continue sequestering C as they mature is unclear. Given uncertainties in eddy covariance estimates and forest inventory measurements, efforts to derive C budgets using multiple approaches are needed. Thus far, the number of forested sites reporting enough information to compare top down and bottom up approaches of C flux estimates has been limited and often lack a complete assessment of the uncertainties associated with each approach. Here we compare estimates of ecosystem C fluxes from a 100-125 year old forest at the Bartlett Experimental Forest, NH using three approaches:

  1. eddy covariance
  2. biometric estimates of net primary production (NPP) and heterotrophic respiration
  3. a carbon inventory approach for 10 years (2004-2014) of plot-level data.

We include estimates of uncertainty for all three approaches. Ten year mean estimates of net ecosystem production (NEP) compared well among all three methods (ranging from 133-141 g C/m2/yr), and indicate that this aging deciduous stand is still acting as a moderate C sink. The largest uncertainty in eddy covariance C flux estimates was associated with the selection of an appropriate ustar filter, while uncertainties in biometric approaches were dominated by belowground C fluxes (fine root and mycorrhizal production). Comparisons of interannual variations in biometric and eddy covariance C fluxes highlight a lack of correlation between wood growth and NEP or GPP, suggesting that source limitation (C supply) may not be controlling rates of wood production. The application of all 3 approaches helped place constraints on difficult to measure C fluxes and provided a comprehensive C budget for future model-data comparisons.

#9 Ecological and environmental controls on multi-decadal carbon cycling processes in the University of Michigan Biological Station forest

Ellen Stuart-Haentjens*, Christopher Gough, Brady Hardiman, Christoph Vogel, Gil Bohrer, Tim Morin, Peter Curtis Contact:, Virginia Commonwealth University

Recent carbon (C) cycling studies demonstrate that forests may continue to accumulate C past early and middle succession; however, the mechanisms behind sustained rates of C storage in aging forests have yet to be identified. Multi-decadal continuous data from Ameriflux towers provide emerging opportunities to examine ecological and environmental controls over such long-term biogeochemical processes. At the University of Michigan Biological Station (UMB) forest, we found that annual net ecosystem production (NEP) remained stable over the 15-yr measurement record, even as leaf area index (LAI) declined and dominant canopy species shifted from early to later succession, owing to sustained overall canopy physiological capacity. Canopy photosynthesis, expressed as GPP, remained relatively unchanged as early successional aspen photosynthetic capacity (Amax) decreased from 20.9 to 18.4 mmol m-2 s-1 and was offset by increases in red oak and red maple Amax from 13.6 to 15.0 and 9.2 to 10.5, respectively. Shifts in leaf-level physiology correspond with changing canopy species’ dominance and increased stand structural complexity as the forest reaches the century mark. For example, while aspens tend to concentrate leaf area at the top of the canopy, the foliage of maples is distributed more evenly at different heights within the canopy. We hypothesize that as this forest continues to transition from early to middle succession, NEP will continue to be sustained due to the combined offsetting changes in leaf-level physiology and increasing structural complexity. Additionally, we are exploring how other variables affect long-term carbon cycling processes at UMB, including climate, atmospheric CO2, and canopy phenology.

#10 Does wetland ‘restoration’ restore ecosystem function? Greenhouse gas fluxes and carbon stocks of restored and old-growth forested wetlands

Ellen Stuart-Haentjens*, Christopher Gough, Scott Neubauer, and Beth Lawrence
Contact:, Virginia Commonwealth University

Wetlands store large amounts of carbon (C) in both biomass and soils, playing a crucial role in offsetting greenhouse gas (GHG) emissions; however, they also account for roughly 30% of the global yearly CH4 emissions. Anthropogenic disturbance has led to the decline of natural wetlands throughout the United States, with a corresponding increase in created and restored wetlands. Studies characterizing biogeochemical processes in restored wetlands, particularly those that are both tidal and freshwater, are lacking but essential for informing science-based carbon management. To fill this knowledge gap, we recently installed an eddy-covariance flux tower measuring CO2, CH4, H2O, and energy exchange in a restored, tidal, freshwater wetland located in Charles City, Virginia at the VCU Rice Rivers Center ( Additionally, we will pair chamber flux and soil measurements in adjacent established and restored wetlands to evaluate the extent to which restoration activities reestablish C cycling functions.

#11 Contrasting hydrologic sensitivity of auto- and heterotrophic respiration in a subtropical forested wetland

Guofang Miao, Asko Noormets*, Jean-Christophe Domec, Montserrat Fuentes, Carl C. Trettin, Ge Sun, Steve McNulty, John King
Contact:, North Carolina State University

Wetlands cycle and store a disproportionately large fraction of global organic carbon relative to their areal coverage, and this ecosystem service of carbon sink is highly vulnerable to environmental change and land use pressure. Qualitatively, it is well recognized that the strength of the wetland carbon sink is controlled by the hydroperiod, which affects the balance between carbon input through assimilation and loss through respiration. However, these processes are poorly constrained quantitatively, making it difficult to describe their response to climate and land use forcings. In the currents study we partition ecosystem respiration to its main components (soil, coarse woody debris, and aboveground biomass) and describe their response to water table fluctuations as well as temperature. Overall, both the total RE (1890-2093 g C m2 yr-1) and the contribution from different components (51-53% from RS, 10-14% from RCWD and 36-38% from RAGP) was conservative during the three study years. By using mass balance approach based on eddy covariance and chamber flux measurements, we show that the heterotrophic components of the respiration budget (soil CO2 efflux, RS, and coarse woody debris respiration, RCWD) were effectively turned off by water level above ground surface, whereas the autotrophic aboveground respiration (RAGP) was apparently insensitive to hydrologic stage. Furthermore, we hypothesize that even root respiration continued unaltered by the hydrologic stage, and that the CO2 produced exited the system as RAGP, as the resistance with the transpiration flow is lower than that of physical diffusion through surface water. Thus, the response of RAGP to environmental drivers is likely confounded by the counteracting effects of temperature and variable shunting of root-derived CO2 through the roots and transpiration stream, as opposed to diffusion into the soil air space and RS. We will explore the implications of two alternative numeric models for describing such partitioning of respiration in this forested wetland.

#12 High carbon uptake in a southeastern South America (Argentina) salt marsh

Natalia E. Tonti*; María I. Gassmann and Claudio F. Pérez
Contact:, University of Buenos Aires

Recent studies show the importance of coastal marshes as carbon sinks because of its high primary productivity and low rate of decay. As they are located in preferential areas for urban settling usually are also endangered areas. Although there are several studies in the Northern Hemisphere, the capacity to store carbon in South American marshes is still unknown. Therefore the aim of this study is to analyze the CO2 fluxes (NEE) and to characterize the net ecosystem production of Mar Chiquita salt marsh located in southeastern Buenos Aires province, Argentina. NEE was measured with eddy covariance technique at 6m height over a 0.7m height Spartina densiflora canopy from February 2014 to March 2015. Simultaneously, biomass and vegetation cover was measured to relate vegetation changes with NEE. Five months of measurement loss, could not be filled with any available gap filling technique, but the use of a 7 order Chebyshev polynomial fit allowed to estimate the lost NEE monthly data. Preliminary results showed that unlike other environments, Mar Chiquita salt marsh is a sink for CO2 throughout the year, with a net ecosystem production of approximately -16.15 ton of CO2/ha year which is, 2 to 8 fold higher than mean annual carbon sequestration measured in USA restored salt marshes (Artigas et al. 2015) or estimated mean values for salt marshes all over the world (Murray et al. 2011). Promising results encourage future research in these South American environments.

#13 Delta Water Management Research Unit Drivers of CO2 and CH4 exchange in mid-South US rice (Oryza sativa) agriculture

Bryant Fong*, Benjamin Runkle, Michele L. Reba
Contact:, USDA-ARS

Rice (Oryza sativa) is grown under flooded soil conditions resulting in greater water use than many other agricultural crops. Nearly 75% of US rice is grown in the Mid-South region of the United States, but limited research on water and greenhouse gas (GHG) flux has been completed at the field scale in this region. Eddy Covariance (EC) measurements of water, carbon dioxide (CO2) and methane (CH4) fluxes allow for integrated measurements of landscape and atmosphere interactions. Measurements were made in continuously flooded drill seeded, production sized rice fields during the 2012 and 2014 production seasons (May to August). The net CO2 flux or Net Ecosystem Exchange (NEE) was found largely to be driven by net solar radiation and air temperature. Constant submergence (water table above soil) of rice and higher temperatures increased CH4 fluxes. The mean daily NEE during vegetative and reproductive growth was 40 g m-2 d-1 and average ET 5.8 mm d-1, two times greater compared to ET after removal of floodwaters in early August. After the field was dried, ET was reduced due to decreased water availability. NEE reduced because of compounding effect of plant maturity and reduced water availability. Entire season GHG flux characterizations in agriculture are important to identify possible GHG reductions in sustainable practices such as alternate wetting and drying (AWD) and multiple inlet irrigation (MI).

#14 Contrasting ecosystem CO2 fluxes of inland and coastal wetlands: A meta-analysis of eddy covariance data

Jingfeng Xiao*, Weizhi Lu, Fang Liu, Yue Zhang, Guang’An Liu, Guanghui Lin
Contact:, University of New Hampshire

Wetlands play an important role in regulating the atmospheric carbon dioxide (CO2) concentrations and thus affecting the climate. However, there is still lack of quantitative evaluation of such a role across different wetland types, especially at the global scale. Here we conducted a meta-analysis to compare ecosystem CO2 fluxes among various types of wetlands using a global database compiled from the literature. This database consists of 143 site-years of eddy covariance data from 22 inland wetland and 21 coastal wetland sites across the globe. Coastal wetlands had higher annual gross primary productivity (GPP), ecosystem respiration (Re), and net ecosystem productivity (NEP) than inland wetlands. Coastal wetlands provided large CO2 sinks, while inland wetlands provided small CO2 sinks or were nearly CO2 neutral. The annual CO2 sink strength was 93.15 and 208.37 g C m-2 for inland and coastal wetlands, respectively. Annual CO2 fluxes were mainly regulated by mean annual temperature (MAT) and mean annual precipitation (MAP). For coastal and inland wetlands combined, MAT and MAP explained 71%, 54%, and 57% of the variations in GPP, Re, and NEP, respectively. The CO2 fluxes of wetlands were also regulated by LAI. The CO2 fluxes also varied with water table depth (WTD), but the effects of WTD were not statistically significant. NEP was jointly determined by GPP and Re for both inland and coastal wetlands. However, the NEP/Re and NEP/GPP ratios exhibited little variability for inland wetlands and decreased for coastal wetlands with increasing latitude.. The contrasting of CO2 fluxes between inland and coastal wetlands globally can improve our understanding of the roles of wetlands in the global C cycle. Our results also have implications for informing wetland management and climate change policy making, e.g., the efforts being made by international organizations and enterprises to restore coastal wetlands for enhancing blue CO2 sinks.

#15 Supersite for eco-hydrometeorological observation and black carbon monitoring at Poker Flat Research Range, Alaska

R. Suzuki*, Y. Kim, H. Kobayashi, S. Nagai, K. Saito, K. Sugiura, Y. Kanaya, F. Taketani, G. Iwahana, R.C. Busey
Contact:, Japan Agency for Marine-Earth Science and Technology

A supersite for eco-hydrometeorological observation and aerosol monitoring that was established by the collaboration study between Japan Agency for Marine-Earth Science and Technology (JAMSTEC) and International Arctic Research Center (IARC), University of Alaska Fairbanks (UAF) is currently in operation at Poker Flat Research Range located about 35km northeast from Fairbanks, Alaska. A 17m scaffold tower, equipped by sensors for general meteorological measurements and water/energy/CO2 fluxes in the atmospheric boundary layer, was constructed in the black spruce forest in 2010. Snow, precipitation, and soil temperature/moisture were continuously monitored. A distributed temperature sensing system with fiber optics monitors spatially continuous surface and ground temperatures. The floor-level carbon dynamics are monitored with the automated open/close chamber system that has 16 chambers. This supersite also plays a role in acquiring the ground-truth for satellite remote sensing. The spectral reflectance of forest canopy/floor and the forest landscape is being monitored by the spectral-radiometers and the automatic digital fisheye lens camera, respectively. Those data are utilized for the study of biogeochemical modeling. Recently, aerosol monitoring system was installed to measure the concentration of the atmospheric black carbon which is mainly emmited from wild fire and anthropogenic combustion and subsequently reduces the albedo of snow surface. General meteorological data and flux data are archived and opened from the website of AmeriFlux as the site “US-Prr.”

#16 Implications of the Kok effect for interpreting eddy-covariance observations

Trevor F. Keenan*, Markus Reichstein, Mirco Migliavacca, Dennis Baldocchi, Dario Papale, Margaret Torn
Contact:, LBNL

Eddy-covariance techniques have been widely used to estimate ecosystem photosynthesis and respiration, both at the site and global scale. Current approaches to estimating both either assume that measured night time respiration increases during the day following a night-time-data estimated temperature response, or assume that the light response of photosynthesis during the day is well approximated by a fitted light response curve. Until recently, observations did not exist with which to test these approaches. Recent evidence suggests, however, that the most commonly used partitioning approaches might be subject to a systematic bias, due to the inhibition of leaf respiration by light (commonly known as the Kok effect). In this study we examine the implications of the Kok effect for partitioning eddy-covariance observations of net carbon exchange into estimates of gross primary productivity and ecosystem respiration, and explore the likely impact on global estimates of photosynthesis and respiration.

#17 Cross-Site Evaluation of Methods for the Estimation of Aerodynamic Roughness Parameters from Flux-Tower Data

Housen Chu*, Dennis D Baldocchi, Cristina Poindexter, Michael Abraha
Contact:, University of California, Berkeley

FLUXNET is a vast network of more than 800 eddy covariance flux sites dispersing across all continents and most of the world’s ecoregions. The global network provides valuable datasets of the direct and in situ measurements of fluxes and ancillary variables that are used across different disciplines and applications. Aerodynamic roughness parameters (i.e., roughness length, zero plane displacement height, aerodynamic canopy height) are one of the potential data products that are crucial for the applications of land surface models and flux footprint models, but have not yet been routinely generated in FLUXNET. This study aims to compare several available methods for the estimation of aerodynamic roughness parameters from single-level eddy covariance measurements and evaluate their feasibility and robustness for the application of FLUXNET datasets. Two groups of methods are adopted based respectively on surface-layer theory (i.e., logarithmic wind profile) and flux variance similarity (i.e., turbulent characteristics). Each approach is implemented by using a series of different estimation techniques, such as least-square regression, numerical iteration, and Markov chain Monte Carlo method. We run the tests across a broad range of ecosystem types ranging from tall-, short-canopy, to open water sites, from closed-canopy/homogeneous to open-canopy/heterogeneous sites, and also from evergreen to deciduous sites. Our findings show no single method dominates in terms of model performance and robustness across all sites. An ensemble average of all plausible estimates or only estimates from site-specific/pre-selected methods may be the most feasible approach. Alternatively, a semi-empirical approach based on the assumptions of presumably known relationships among aerodynamic roughness parameters may provide a robust and sufficiently accurate estimate.

#18 A Bayesian analysis of eddy covariance uncertainty

John Frank*, William J. Massman, Brent Ewers
Contact:, U.S. Forest Service  

The uncertainty in eddy covariance measurements is often characterized from the variance between co-located or nearby systems, differences between consecutive days or other time periods, and deviations from modeled predictions. We present a different approach where uncertainties in each measurement step can be combined in a Bayesian framework to estimate the uncertainty of an eddy covariance flux. We reanalyze data from field experiments featuring multiple Campbell Scientific CSAT3 sonic anemometers to resolve the three-dimensional shadowing correction by optimizing differences between anemometers mounted simultaneously vertical and horizontal. The resulting posterior correction increased the standard deviation of vertical wind velocity by ~ 10% and when applied to various sites across North America increased the turbulent components of the surface energy balance (sensible and latent heat) by an average of 8-12% with 95% credible intervals averaging between 6-14%. We present an intercomparison of seven fast-response hygrometers of both open and closed path designs that were produced during three different eras: > 20 years including the Resonance Lyman-alpha and the Campbell Scientific KH2O, ~ 15 years comprising the LI-COR LI7000 and LI7500, and < 10 years containing the LI-COR LI7200 and Campbell Scientific EC155 and EC150. Preliminary results show that when calibrated against a dew point generator, the closed path analyzers were more accurate in measuring the slow-response changes in ambient humidity than the open path instruments; this performance was improved by in situ calibration against a chilled-mirror hygrometer. These types of analyses can be summarized and combined via the Bayesian cut function to estimate eddy covariance uncertainty.

#19 Probing photosynthesis using solar induced chlorophyll fluorescence in drought-prone deciduous forests

Jeffrey Wood*, Lianhong Gu, Jeff Riggs
Contact:, University of Missouri

Solar induced fluorescence (SIF) shows great promise for probing photosynthesis. We have established a research program directed towards examining SIF and photosynthesis from leaf to regional scales. Here, we present analyses of gross primary production (GPP) and satellite-based SIF from 2015 and describe a new system deployed to measure canopy-scale SIF. We synthesized SIF from NASA’s OCO-2 satellite with ecosystem flux and physiological measurements in the Missouri Ozarks. Community predawn leaf water potential showed mild to severe water stress from mid-July through the end of the growing season. Water stress dynamics were reflected in the GPP and SIF time series, which showed remarkable coherency. There was strong linear GPP-SIF scaling (R2≥0.96) at instantaneous and daily timescales; with no statistical difference in coefficients. These analyses suggest that variable fluorescence at the photosystem level due to changes in photoprotection do not affect ecosystem GPP-SIF relationships at scales relevant to remote sensing GPP mapping applications. There is, however, much to learn regarding variations in SIF at the leaf to canopy scales to permit the extraction of mechanistic information on photosynthesis. We therefore, deployed a system centered around a spectrometer (QEPro, Ocean Optics Inc.) to measure SIF in the O2-A band. The fiber optic is affixed to a smart-motor that alternates between sky and canopy views. Data acquisition and control is accomplished using a CR1000 datalogger, with the program dynamically optimizing the integration time to account for changes in the radiation environment to take advantage of the highest detector sensitivity offered by the spectrometer. Controlling the QEPro with a datalogger also allows key supporting meteorological variables to be measured concurrent with spectral samples, permitting a more thorough interpretation of the SIF signal. We will also be making measurements of leaf-level gas exchange and fluorescence parameters. This suite of measurements, coupled with satellite-based observations will provide unprecedented insight into photosynthesis across a broad range of spatiotemporal scales.

#20 Fast air-temperature fluctuations affect open-path CO2 flux measurements

Ivan Bogoev* Contact:, Campbell Scientific, Inc.  

Uncertainties in net ecosystem CO2 exchange (NEE) measured by the open-path (OP) eddy-covariance method have limited inter-site comparability and have hindered the ability to characterize the variation of carbon dynamics across diverse ecosystems. In an effort to reduce the uncertainties in OP flux measurements, a recently developed CO2 eddy-flux system (IRGASON) integrates the sensing paths of the infrared gas analyzer and the sonic anemometer, eliminating the need for spectral corrections associated with sensor separation. In addition, the spatial co-location of the gas analyzer and the sonic anemometer allows direct measurement of sensible and latent heat flux in the gas-sensing path, compensating for CO2 dilution effects related to temperature and humidity fluctuations. This new sensor has been tested over a wide variety of environmental conditions and flux regimes. Several northern flux sites using IRGASON open-path gas analyzers have observed small apparent CO2 uptake during the late-winter season. It has been determined that the apparent CO2 uptake is caused by inadequate correction for spectroscopic effects associated with high-frequency temperature fluctuations and is proportional to the magnitude of the sensible heat flux. This study demonstrates that sonically derived fast-response air temperature in the optical sensing path of the OP gas analyzer can be used as a scaling parameter in the calibration model to correct for the temperature-induced spectroscopic effects. The approach is further evaluated in field comparisons between open- and closed-path flux systems under a wide range of heat-flux regimes and varying meteorological parameters. In all cases, the use of fast air temperature considerably improved the agreement between the two systems and reduced the overall uncertainty in the NEE estimates.

#21 A Low Power Configuration for a Tunable Diode Laser Trace Gas Analyzer to Measure N2O & CO2 Fluxes using Flux Gradient Method

Ben Conrad* and Steve Sargent Contact:, Campbell Scientific Inc.

Campbell Scientific has been manufacturing tunable diode laser trace gas analyzers (TGAs) since 1993. These instruments provide measurements with high accuracy and low noise all while remaining rugged and portable for continuous field operation. One limitation of the TGA for many applications has been the requirement of AC mains power to drive the pumps, heaters and electronics of the system. Here we provide a novel system configuration for the flux gradient method to measure N2O & CO2 with the TGA that drops the power requirement substantially allowing for a more practical solar power solution. We demonstrate that while operating in this low power configuration for flux gradient, the TGA still provides low noise/high accuracy measurements.

#22 Biosciences Advanced Tools for Flux Network Management and Cross-sharing of Flux Stations

L. Xu*, D. Johnson, M. Velgersdyk, I. Begashaw, D. Allyn, and G. Burba
Contact:, LI-COR

With more stations and networks, larger data flows from each station, and smaller operating budgets, modern on-line tools can help effectively and efficiently handle the entire process, including sharing data amongst collaborative groups. Such tools can maximize time dedicated to publications answering research questions, and minimize time and expenses spent on data acquisition, processing, quality control and station management. Cross-sharing the stations with external collaborators may help leverage available funding, and promote data analyses and publications. A new low-cost, advanced system, FluxSuite, utilizes a combination of hardware, software and web-services to address these specific demands. It automates key stages of flux workflow, minimizes day-to-day site management, and modernizes the handling of data flows:

  • The system can be easily incorporated into a new flux station, or as an upgrade to many presently operating flux stations, via weatherized remotely-accessible microcomputer, SmartFlux 2
  • Each next-generation station will measure all parameters needed for flux computations in a digital and PTP time-synchronized mode, accepting digital signals from a number of anemometers and data loggers
  • The field microcomputer will calculate final fully-processed flux rates in real time, including computation-intensive Fourier transforms, spectra, co-spectra, multiple rotations, stationarity, footprint, etc.
  • Final fluxes, radiation, weather and soil data will be merged into a single quality-control file
  • Multiple flux stations can be linked into an automated time-synchronized network
  • Flux network managers, or PIs, can see all stations in real-time, including fluxes, supporting data, automated reports, and email alerts
  • PIs can assign rights, allow or restrict access to stations and data: selected stations can be shared via rights-managed access internally or with external institutions
  • Researchers without stations could form “virtual networks” for specific projects by collaborating with PIs from different actual networks

This presentation provides detailed examples of FluxSuite currently utilized to manage two large flux networks in China (National Academy of Sciences and Agricultural Academy of Sciences), and smaller networks with stations in the USA, Germany, Ireland, Malaysia and other locations around the globe. Very latest 2016 developments and expanded functionality are also discussed.

#23 New CO2/H2O Flux Measurements Systems: Development and Field Tests

L. Xu*, I. Begashaw, G. Fratini, F. Griessbaum, J. Kathilankal, D. Franz, E. Joseph, E. Larmanou, S. Miller, D. Papale, S. Sabbatini, T. Sachs, R. Sakai, D. McDermitt, and G. Burba
Contact:, LI-COR Biosciences

New open-path and enclosed-path flux measurement RS systems are based on established LI-7500A and LI-7200 models, with the focus on improving stability of the analyzer performance in the presence of contamination, refining temperature control and compensation, and providing more accurate gas concentration measurements. In addition to analyzer redesign, both RS systems include automated on-site flux calculations using EddyPro by a new weatherized remotely-accessible microcomputer, SmartFlux 2, with digital inputs from multiple models of sonic anemometers and dataloggers. The ultimate goal of such development was to streamline and standardize processing of CO2 and H2O hourly fluxes and long-term carbon and water budgets. Field tests of both systems were conducted over six periods, each 5-14 months long, at 6 sites with diverse environments, setups, and types of contamination, using 26 gas analyzers. The open-path LI-7500RS system performed significantly better than the original LI-7500A model in terms of contamination-related drifts in mean concentrations. Improvements in CO2 drifts were strong, with RS models often drifting few-to-tens of times less than the original. Improvements in H2O contamination-related drifts were particularly significant, with modified models often drifting many tens of times less than the original. The enclosed LI-7200RS system performed substantially better than the original LI-7200 in terms of the drifts in H2O, sometimes drifting few-to-tens of times less than the original. Improvements in CO2 contamination-related drifts were modest, similar or just a bit better than the original. Results from field tests suggest that both RS systems can improve flux data coverage and reduce site maintenance:

  • Frequency of required cleaning and maintenance should noticeably decrease, especially for the open-path design
  • Amount of total data coverage over long periods of deployment and amount of highest quality data with smallest error bars should increase substantially
  • Digital data collection and automated traceable processing scheme should help minimize processing errors and lead to better intercomparisons between different groups and sites

#24 Methane Soil Flux Determination Using Cavity Ring-Down Spectroscopy for Mobile and High Resolution Concentration Measurements

David Kim-Hak*, Derek Fleck, Taku Ide, Liam Gannon
Contact:, Picarro Inc.

Methane is a potent greenhouse gas that has a short-term global warming impact. It comes from a variety of natural and anthropogenic sources which include wetlands, landfills, oil/gas/coal extraction activities and natural gas distribution leaks. Locating and containing these emissions are critical to minimizing their environmental impacts and economically beneficial when retrieving large fugitive amounts. Generally, methane detection is conducted by making real-time atmospheric measurement and identifying large variations from the typical atmospheric concentration of 2 ppm. The measurement technique is well-developed and includes catalytic sensors and laser analyzers. On other hand, soil emission quantification (flux) is not a direct measurement as it is calculated based on rate of gas accumulation in a known chamber volume. In this case, high-performance analyzers are required to provide high temporal resolution and high-precision in situ measurement of the concentration. Here we present a highly-efficient method of measuring methane emission flux using the latest portable and lightweight Cavity Ring-Down Spectroscopy analyzer and a mobile accumulation chamber. The system enables rapid measurement (~1 minute) in a wide range of emission flux. We will present supplemental data acquired from a soil flux measurement campaign at a coal outcrop site in the Four Corner region

  • Picarro: David Kim-Hak and Derek Fleck
  • Koveva Ltd: Taku Ide and Liam Gannon

#25 NEON’s storage flux measurements of CO2 and H2O, processing and data products

Hongyan Luo*, Natchaya Pingintha-Durden, Andy Fox
Contact:, NEON

NEON’s eddy-covariance flux measurements and data products encompass two principal components: turbulent fluxes and storage fluxes. This paper will focus on the latter component. The storage fluxes of CO2 and H2O will be integrated from vertical profile measurements of CO2 and H2O concentration. Specifically, the CO2 and H2O concentrations from different levels are measured using a single infrared gas analyzer (IRGA) located in a climate controlled instrument hut. Instead of manually calibrating the IRGA in the field, an automatic field validation is performed on a daily basis. Once the data is retrieved, the streaming calculation of the storage flux poses several challenges, including:

  1. Application of daily field validation results to the reported observations;
  2. Allocation of the observations to the correct measurement level and their temporal aggregation;
  3. Spatial-temporal interpolation of measurements occurring at different times for the different levels;
  4. Determination of the storage flux footprint from various measurement levels that each source from a different area;
  5. Adaptive applicability of the algorithm across diverse ecosystems.

In addition to the CO2 and H2O measurements, the stable isotope measurements of 13C in CO2, δ18O and δ2H of H2O at the same levels as the storage flux profile provide an additional source of information: It presents the science community with the potential to partition NEE into ecosystem photosynthesis and respiration, as well as to partition ET into soil evaporation and ecosystem transpiration. In this talk we will present NEON’s gas profile system design and resultant data products. We welcome input on the identified issues and wish to discuss future opportunities.

#26 QA/QC and uncertainty budget of NEON’s eddy-covariance flux data products

Natchaya Pingintha-Durden*, David Durden, Cove Sturtevant and Stefan Metzger
Contact:, NEON

Eddy-covariance (EC) fluxes are filled with theoretical assumptions and require many additional quality assurance and quality control (QA/QC) tests in conjunction with quantification of uncertainty. Therefore, NEON’s EC flux data products must be subjected to a thorough data quality and uncertainty assessment that lead to both quality flags and quantitative uncertainty estimates. Here, we present NEON’s flux QA/QC and uncertainty quantification framework. In this framework, a wide range of qualitative and quantitative algorithmic processing routines are applied to flux data products including tests related to sensor errors, statistical tests, tests based on the degree of fulfillment of one or several assumptions of EC measurements, and top-down and bottom-up uncertainty calculations. By utilizing the general NEON’s data quality framework (Smith et al., 2014), the results from individual QA/QC tests are summarized in a way that is transparent and easily interpretable by the user. The final quality flag is produced individually for each flux data product, which represents a determination of the validity for further data analysis. Similarly, the final combined uncertainty will be calculated and supplied for each flux data product. A two-pronged approach consisting of a top-down and bottom-up approach is used to determine the combined uncertainty. For the bottom-up approach, quantifiable systematic and random uncertainties from the various uncertainty components are combined individually and propagated into combined standard random and systematic uncertainty. These uncertainties are then reported as a final, overall bottom-up uncertainty. The top-down approach uses the energy balance ratio to determine a measure of the overall systematic uncertainty of the measurement, which can be reported with the combined standard random uncertainty calculated in the same way as the bottom-up approach. Additionally, both the QA/QC and uncertainty budget functions will be available in the R package, eddy4R.

#27 From field notes to data portal—An operational QA/QC framework for tower networks

Cove Sturtevant*, Skyler Hackley, Timothy Meehan, Joshua Roberti, Greg Holling, and Santiago Bonarrigo
Contact:, NEON

As tower networks such as Ameriflux, ICOS, and NEON have grown in size and sophistication, tools have lagged for robust, efficient, scalable quality assurance and quality control (QA/QC). Even with the application of automated tests, QA/QC of tower data remains a largely manual process relying heavily on visual inspection of the data. In addition, field notes of observed measurement interference or visible problems are often recorded on paper without an explicit link to data flagging during processing. To complicate matters, there are often multiple personnel managing different sites or different steps in the data flow. As such, an increase in network size requires a proportional increase in personnel devoted to QA/QC, quickly stressing the human resources available. There is a need for a scalable QA/QC framework that combines the efficiency and standardization of automated tests with the power and flexibility of visual checks, and includes an efficient communication pathway from field personnel to data processors to end users. Here we present such a framework and an accompanying set of tools in development, including a mobile application template for recording tower maintenance and an R/shiny application for efficiently monitoring data quality. We seek to incorporate lessons learned from the Ameriflux community, gather feedback, and provide tools to aid continued advancements.

#28 Hierarchical data format for eddy-covariance data

David Durden*, Cove Sturtevant, Natchaya Pingintha-Durden, Hongyan Luo, Andy Fox, Greg Holling and Stefan Metzger
Contact:, NEON

Large data collecting networks have led to better understanding of environmental variation through an increase in available information. However, analyzing, curating, and archiving the observations with associated metadata for large datasets can be complicated. Tower networks, such as ICOS, Ameriflux, and NEON, illustrate the growing size of datasets from dispersed measurement sites. Eddy-covariance data from across the NEON network are expected to amount to 100 Gigabytes per day. The large throughputs of data between the database, the processing environment, and the data portal require an efficient file format. The capability to process large data sets is reliant upon efficient input and output of data, data compressibility to reduce compute resource loads, and the ability to easily package and access metadata. The Hierarchical Data Format (HDF5) is a file format that can meet these needs. A NEON standard HDF5 file structure and metadata attributes allow users to explore larger data sets in an intuitive “directory-like” structure. Additionally, HDF5 expands possibilities for data provenance where various levels of data products can be packaged together in a single file.

#29 NEON’s Streaming Processing Pipeline for Eddy-Covariance Raw Data

Greg Holling*, David Durden, Andy Fox, Hongyan Luo, Natchaya Pingintha-Durden, Cove Sturtevant, and Stefan Metzger Contact: NEON

Automatically retrieving, analyzing and storing complex datasets along their contextual information can pose a substantial challenge. Tower networks such as ICOS, Ameriflux, and NEON exemplify this challenge through the growing size of datasets from dispersed measurement sites: Eddy-covariance raw data across the NEON network are expected to amount to 100 Gigabytes per day. The streaming processing and dissemination of this data requires a seamless integration of reproducible, extensible and portable scientific code, in combination with efficient data flows, scalable processing and data discovery tools. Hence, the eddy-covariance raw data processing at NEON differs from a pre-existing pipeline in some fairly significant ways:

  1. The eddy-covariance code is being developed by a consortium of scientists and software engineers both inside and outside of NEON; most other NEON data products rely on code solely developed by NEON software engineers.
  2. The eddy-covariance code is written as publicly available R-packages; most other NEON data product code is written in Java and used NEON-internally only.
  3. The eddy-covariance R-packages and workflows are compiled into portable, publicly available Docker images and run in parallel Docker containers.
  4. I/O for the eddy-covariance code uses self-describing HDF5 files; most other NEON code interacts directly with the NEON production database and outputs data and contextual information in separate plain text files.

We will discuss the general architecture used for processing eddy-covariance code in the NEON environment, along with how tools such as HDF5 and Docker will be used for algorithmic processing of the eddy-covariance data.

#30 Investigating the effects of landscape heterogeneity on flux tower eddy-covariance measurements using high resolution images

Giannico V*, Shao C., Ouyang Z, John R, Lafortezza R., Chen J. (Michigan State University)

Eddy-covariance (EC) technique relies on the assumption that the area contributing to the fluxes is covered by homogeneous vegetation. Such assumption is often violated due to the difficulties in finding such condition in natural areas. Uncertainty in the contributing area (footprint) can thus be a source of misinterpretation in EC data analysis. In this study, we aim to investigate the effects of landscape heterogeneity on flux tower eddy-covariance measurements using high resolution images. We selected two flux tower located in Mongolia with the purpose of representing two different levels of heterogeneity. For each tower, we calculated EC fluxes and the relative contributing area using the Kormann and Meixner (2002) footprint model. We extracted multiple texture metrics from WorldView-2 images covering the surrounding area of the two towers. After dividing the area around each tower in 40 sectors we analyzed the relationship between the landscape heterogeneity and the CO2 fluxes. Preliminary results are presented.

#31 AmeriFlux Management Project Tech Team

Sébastien C. Biraud1, Stephen W. Chan1, Sigrid Dengel1, Chad Hanson2, Dave Billesbach3, Margaret S. Torn1 • 1 Lawrence Berkeley National Laboratory / 2 Oregon State University / 3 University Nebraska Lincoln

AmeriFlux’s goal is to develop a network of long-term CO2 flux sites for quantifying and understanding the role of the terrestrial biosphere in global climate change. The network currently includes more than 100 sites started by many scientists and supported by multiple U.S. federal agencies (DOE, NSF, USDA, USFS). The AmeriFlux Management Program (AMP) Tech Team at LBNL strengthens the entire AmeriFlux Network and Core Sites by— standardizing operational practices, calibration, and maintenance routines, setting clear data quality goals, helping resolve instrument failure promptly. To ensure inter-comparability in the network, we conduct site comparisons with portable eddy covariance (PEC) systems, provide calibration gas standards and lab-quality sensors to check instrument performance, and identify uncertainties associated with data processing using data diagnostics and gold-standard files. Our team is experienced in testing new instruments and working with manufacturers, and is building relationships with both vendors and investigators. We have completed 33 site visits in since the beginning of the project.

Last updated Thursday 9/29/16 – marilyn