CTCD Terrestrial Carbon Cycle and Earth Observation summer school 18-26th August, 2005 Editors Mat Williams (University of Edinburgh, UK) & Andreas Heinemeyer (University of York, UK) Organisers Mat Williams Andreas Heinemeyer Staff Phil Ineson (University of York), John Grace, Dave Reay (University of Edinburgh) Phil Lewis, Paul Bowyer, Tristan Quaife (UCL) Andrew Fox (University of Sheffield) Dave Reay (University of Edinburgh) Steve Dobson and John Jones (University of York) Acknowledgments: We would like to thank all funding bodies who made this summer school possible: (i) NERC (UK) (ii) ESF (iii) QUEST (UK) Pictures: Most of the pictures in this manual are produced from licensed data – no photocopying of this material is allowed. Cover picture: Pen-y-ghent field site looking towards Blishmire and Overdale Moor; inset: Malham Tarn, Field Studies Council field centre (Andreas Heinemeyer). Index Introduction ................................................................................................ 2 Course structure and timing ....................................................................... 4 Eddy Covariance ...................................................................................... 13 Measuring Soil Carbon Stocks ................................................................. 17 Measuring Vegetation Carbon Stocks and Leaf Area ............................... 20 NEP chamber........................................................................................... 25 Soil Respiration ........................................................................................ 29 Methane in the Global Carbon and Greenhouse Gas Budget .................. 38 Spatial measurement of Leaf Area using Earth Observation .................... 42 Introducing Ground-Penetrating Radar .................................................... 51 1 Introduction The goal of the summer school is for students to investigate the processes that drive the terrestrial carbon (C) cycle, and to learn the tools available to quantify the associated processes and measure carbon stocks. We will work in an upland heather catchment of a few hectares, small enough to survey and measure, but large enough to incorporate much of the complexity associated with the current science of terrestrial C dynamics. The field work will mainly focus on a 1 km2 area around an Eddy flux tower site, based on a UK national grid reference used for the UK national carbon inventory. However, some work might be done in a slightly larger 4 km2 area around the core site. You will work in a team of students, rotating among various scientific activities within the catchment, so that you gain experience with, and data from, each. Synthetic questions At the end of the week, you will have worked with an array of instrumentation and gathered associated data. Your task is to synthesise these data to produce a summary of the catchment C exchanges over the previous seven days, answering the following questions: Was the catchment a net sink or source of C? How certain are you of your estimate of C source or sink strength? Was the catchment relatively uniform in its behaviour, or were there significant differences within it? What fraction of the total C stored in the catchment turned over during the week? In terms of global warming potential, was CO2 or CH4 the more significant form of exchanged C? On the final day of the summer school, you and your team will present your synthesis to the staff and other students. You will also be presented with a CD containing all the information acquired during the week and all the presentations given. Learning Objectives of the Summer-School The summer school has a series of learning objectives – at the end of the school you should be able to do the following: 1. Measure C fluxes using multi-chamber soil respiration systems, eddy flux systems, and chamber CO2 flux systems and to understand the linkage of fluxes to key environmental factors. 2. Measure C stocks in soils and vegetation. 3. Determine the spectral properties of canopies and relate these to remotely sensed data. 4. Synthesise data collected during the field campaigns to make inter-comparisons between different measurement approaches. 5. To understand the major uncertainties of different key methods used in carbon budgeting. Scientific Background Importance of organic carbon-rich soils in the global carbon cycle One of the greatest problems facing mankind is the impact of environmental change on fundamental aspects of the biogeochemical cycle. Soil is a major component in the global carbon (C) cycle, with about 1,500 Gt of organic C to a depth of 1 m and a further 900 Gt from 1-2 m[1], and vulnerable to impacts of human activity. Globally twice as much carbon is stored in soils as in the atmosphere[2] with peatlands contributing a third of this[3]. Thus even small changes in soil C stocks might contribute significantly to global climate change, for example, due to a negative 2 feedback as a result of global warming[4]. Whereas above ground carbon cycling is well understood, there is great uncertainty in climate impacts on soil carbon cycling[5]. For example, the biological significance and relevance of the most critical C-turnover parameter used in soil organic matter (SOM) sub-models of Dynamic Global Vegetation Models (DGVMs), the Q10 value, representing soil respiration responses of different C-pools to temperature, is not sufficiently understood[6, 7]. More importantly, the understanding and process representation of C cycling in organic C-rich soils is so far insufficiently included in only a few SOM models and not incorporated in any DGVM. Critical feedback: How important is the recalcitrant SOM pool? Recent literature suggests that the decomposition of the more recalcitrant SOM pools is at least as temperature-sensitive as the decomposition of the more labile carbon pools[9,10]. Model predictions, however, show that the response of this “slow” carbon pool will determine the degree to which terrestrial ecosystems will cause a positive feedback to global warming[4,11]. This slow pool represents very large carbon stores within terrestrial ecosystems, most notably in peatlands. Therefore there is the potential for peatlands to respond in a much less transient way to elevated temperature than other low-C soils. However, a number of recent isotopic studies have highlighted the fact that a large percentage of the CO2 being respired by soils appears to have been recently fixed by plants[8]. There appears to be little data on what determines the relationship between photosynthesis and below-ground respiration. Tackling uncertainties in C stocks and fluxes Even relatively small uncertainties in soil C-stock and flux estimates lead to large uncertainties in model predictions. Despite containing the largest soil C stocks, C cycling in organic C-rich soils and its links to the environment (e.g. temperature and photosynthesis) are not incorporated into global model predictions due to insufficient process understanding. Thus, there is an obvious need to link scientists from different key research areas to work on an organic-rich C site (e.g. a heather peatland) to: improve data acquisition, compare C flux measurement accuracies, improve our general understanding about the uncertainties in the terrestrial carbon cycle, and find measures to reduce them, all of which are key strengths of CTCD. This summer school will provide up-to-date equipment as well as expertise gained from links to the CarboEuroflux group. The course will focus around a field experimental campaign designed to address a series of critical scientific and technical questions: Science Questions You will be working alongside scientists from the NERC Centre for Terrestrial Carbon Dynamics (CTCD), and their colleagues, who - aside from training you - are also interested in collecting data during the week to answer several key science questions related to C cycle science. 1. What are the natural scales and processes behind ecosystem CO2 flux variability? We want to know how heterogeneous the landscape is, as any variability influences the resolution required for accurate assessments of regional C exchanges. Approach: Compare chamber flux measurements with eddy-flux tower. 2. Can we distinguish overall CO2 efflux into fluxes from specific pools? (for instance, separate net exchange of carbon in photosynthesis and respiration). Approach: Soil respiration data, chamber data, & tower data. 3. Can canopy spectral data explain carbon flux variability? We want to know if satellite observations can improve regional assessments of C exchange. Approach: Compare measurements of leaf reflectance with gas exchange. 3 4. What is the uncertainty on assigning vegetation parameters across a landscape? Approach: Compare detailed soil and vegetation surveys with remotely sensed data. 5. How effective is our up-scaling of ecosystem C dynamics? Can we divide the landscape into functional units and upscale data from these to produce a reasonable estimate of landscape activity? Approach: Characterisation of flux footprint into dominant vegetation types via mapping and remote sensing, up-scaling of flux predictions from leaf and chamber data, comparison to eddy-flux tower.Technical challenges There are considerable technical difficulties in quantifying C exchanges which you will learn during the week. CTCD scientists and their colleagues are interested in isolating and reducing problems. We will test some new methods (e.g. multiplexed soil respiration chambers) and inter-compare existing approaches. We will also discuss the best approaches to simulation and modelling. 1. Inter-comparison of chambers and eddy-flux system approaches to CO2 flux measurement. Both approaches have advantages and problems. 2. Inter-comparison of a new multiplex system with a single chamber approach for soil CO2 efflux measurement. 3. What is the best chamber design to best measure spatial and temporal variability in soil respiration (e.g. how do chamber size and measurement disturbance affect soil C efflux?) 4. How do we make best use of remote sensing data within the terrestrial carbon cycle framework (including introductions to data analyses, evaluation and data assimilation)? 5. What are the limitations and advantages of different databases (e.g. spatial scale) and model structures (e.g. accuracy of processes)? Course structure and timing Teams and general organisation: 1. The students are split into 6 teams (A-F), each with 5 or 6 members. 2. Each team is tasked with answering the synthetic questions (see above), and the science questions if possible. 3. We have 6 activities (see below). 4. Each team spends 1 day on each activity. 5. There is one day to generate the synthesis. However, some data might be measured (e.g. weight) the following day by another group. 6. On the final morning each team presents their conclusions to the group. Activities 1. 2. 3. 4. 5. 6. Met station and eddy flux [Andy Fox, John Grace] Soil respiration chambers [Andreas Heinemeyer] NEP chamber [Mat Williams] Soil and vegetation C stocks [Phil Ineson, John Grace] Methane fluxes [Dave Reay] Canopy characteristics and surface reflectance [Phil Lewis] 4 Also peat depth measurement with Ground Penetrating Radar (GPR) [Steve Dobson/ John Jones] Each activity is described in detail later in this manual. Fieldwork organisation Each day in the field you will be familiarised with equipment and trained in usage. You will then collect data and, back at the lab, work with the staff to understand and analyse the data. You will be encouraged to develop your own sampling design and scaling strategies to answer the synthetic questions. There will be time each afternoon for data analysis - by the end of each day you need to make sure you have completed the critical analyses. Overall time table There are six days allocated to the activities, then a day of synthesis, and presentations on the final day. Day Date Assignment Thursday 18 August Students arrive late. 8pm introductory meeting Friday 19 1st activity Saturday 20 2nd activity Sunday 21 3rd activity Monday 22 4th activity Tuesday 23 5th activity Wednesday 24 6th activity Thursday 25 Synthesis Friday 26 AM presentations, conclude at lunch, depart. Teams and Activities This table indicates which team (A-F) does which activity (1-6) on each day (19-24 August) Date\Team 19 (Fri) 20 (Sat) 21 (Sun) 22 (Mon) 23 (Tue) 24 (Wed) A 1 2 3 4 5 6 B 2 3 4 5 6 1 C 3 4 5 6 1 2 D 4 5 6 1 2 3 5 E 5 6 1 2 3 4 F 6 1 2 3 4 5 Daily timetable 07:30 08:30-09:20 09:30 10:00 12:30-13:00 Breakfast Lecture (activity related) Depart for the field Fieldwork Pack lunch 19:00 20:00 21:00 Return to FSC (~13:30-15:30) Data analysis (until 18:00) Dinner Guest lecture Bar Lectures To supplement the focussed experimental programme, we have two lecture series. Each morning, from 0830-0920, there will be a lecture relating to one of the field activities, providing more detailed insight into the techniques and the science. On five evenings there are guest lectures on a broader range of topics: Date Lecturer Title Wolfgang Knorr Modelling the terrestrial carbon cycle 19 (Fri) Jens Arne Subke C flux components and the environment 20 (Sat) Uwe Franko Modelling regional C-cycling and its uncertainties 21 (Sun) The role of peatlands in the global carbon cycle (CH4 and 22 (Mon) Mike Billett DOC) Werner Kutsch Challenges in ecosystem flux measurements: experiences 23 (Tue) within CarboEurope References: [1] Miko UF Kirschbaum (2000) Will changes in soil organic C act as a positive or negative feedback on global warming? Biogeochemistry 48 (1): 21-51. [2] Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K, Johnson CA (eds). (2001). Climate Change 2001: The Scientific Basis. Cambridge, UK: IPCC, Cambridge University Press. [3] Gorham, E (1991). Northern peatlands: Role in the carbon cycle and probable responses to climatic warming. Ecological Applications 1, 182-195. [4] Cox PM, Betts RA, Jones CD, Spall SA, Totterdell IJ (2000). Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature, 408, 184-187. [5] Valentini R, De Angelis P, Matteucci G, Monaco R, Dore S, Scarascia Mungnozza GE (1996). Seasonal net carbon dioxide exchange of a beech forest with the atmosphere. Global Change Biology, 2, 199207. [6] Giardina CP and Ryan MG (2000). Evidence that decomposition of organic carbon in mineral soil do not vary with temperature. Nature, 404, 858-861. [7] Grace J and Rayment M. (2000). Respiration in the balance. Nature, 404, 819-820. [8] Ekblad A, Hogberg P (2001) Natural abundance of C-13 in CO2 respired from forest soils reveals the speed of link between tree photosynthesis and root respiration. Oecologia, 127, 305-308. [9] Knorr W, Prentice IC, House JI, Holland EA (2005) Long-term sensitivity of soil carbon turnover to warming. Nature, 433, 298-301. [10] Fang C, Smith P, Moncrieff JB, Smith JU (2005) Similar response of labile and resistant soil organic matter pools to changes in temperature. Nature, 433, 57-59. 6 [11] Eliassson PE, McMurtrie RE, Pepper DA, Stromgren M, Linder S, Agren GI (2005) The response of heterotrophic CO2 flux to soil warming. Global Change Biology, 11, 167-185. 7 Site information The study site is located in one of the most beautiful regions within England, the North Yorkshire Dales National Park (established 1954). Please bear this in mind when sampling; we are here to do science and not to destroy an area of outstanding natural beauty! Field work will mainly focus on a 1 km2 grid (green square) on Blishmire and Overdale where several measurement stations such as the Eddy flux tower (FT) have been set up. Most of this area is on sedimentary Carboniferous limestone (~300 million years ago) and under the upland climate (wet and cold most of the year) developed an important soil carbon sink with large areas covered by peaty soils (mostly peaty gley and deep peat soils such as the Winterhill series). Please find below some images showing details of data accessible to you for a 4 km 2 area (red square): (i) OS map; (ii) land cover map; (iii) soil types; soil carbon estimates over (iv) 30 cm and (v) 100 cm soil depth. 8 9 However, little is known about the carbon source/sink functioning of these upland regions and our work will try to investigate this mainly within a 1 km2 grid (green square) around the Eddy 10 flux tower site (FT) which also contains a lime vs. non-lime treatment area (Lime). All the above datasets have been licensed to CTCD and are not to be used outside the summer school. The study site is part of a larger upland area, the Yorkshire Dales, which has large areas of peaty soils mainly due to its relatively high elevation and the oceanic climate. Consequently, there are major soil carbon stocks. The following picture shows the best UK estimate of soil carbon up to 1 m depth within the wider study area superimposed on a digital elevation model; soil carbon stocks are very large and increase with altitude to up to 100 kg C/ m2. This is the result of a changes in both, inputs as net primary productivity (NPP) and outputs through decomposition, which we will measure as soil respiration. However, there are many other factors involved such as changes in litter quality and soil pH. Figure showing the wider study area within the Yorkshire Dales National Park. Soil carbon estimates for 1 m depth overlaid onto a DEM. The study site (red square) is approximately 10 miles from the field centre (blue circle) at Malham Tarn. However, it is important to note that vegetation carbon stocks are only a fraction of this, reaching values of less than 15 kg C/ m2. Furthermore the soil carbon stocks do not consider peat depth of greater than 1 m. This is of particular importance as we do not know how these deep soil carbon stocks will respond to climate change. They certainly have a great potential for a positive feedback – just think about small changes in big numbers. For clarification the two graphs below show (i) the global carbon cycle as proposed by the IPCC and (ii) the world’s vegetation and soil carbon stock estimates. 11 12 Eddy covariance Eddy Covariance (Instructor: Andrew Fox) Introduction to the technique Eddy covariance (or the analogous, older term, eddy correlation) flux measurements have become an increasingly popular method for assessing ecosystem carbon exchange, and complement traditional methods assessing inter-annual changes in biomass and soil carbon (e.g. Baldocchi et al., 1988; Running et al., 1999; Geider et al., 2001; Lee et al., 2004). This popularity is due to a number of factors. First, and most importantly, the measurements are made at the most appropriate ‘canopy-scale’, assessing net CO2 exchange for a whole ecosystem. Second, it is a direct measurement of net CO2 exchange across the canopy-atmosphere interface, using micrometeorological theory to derive a flux from the covariance between vertical wind velocity and scalar concentration fluctuations. And thirdly, the technique is capable of measuring net ecosystem CO2 exchange across a wide range of temporal scales, ranging from hours to days to years. Consequently, it is a particularly good technique for investigating ecosystem physiology, able to assess whole ecosystem response to environmental perturbations and when utilised in paired studies it can be used to assess the effects of disturbance, stand age or plant functional types (Anthoni et al., 2002; Baldocchi et al., 1997; Chen et al., 2002). Since 1997 regional networks of eddy covariance flux towers over a variety of principally forest ecosystems have been in operation in Europe (CarboEuro-Flux, Aubinet et al., 2000) and North America (Ameriflux, Running et al., 1999) and there are currently over 180 sites world wide as part of the FLUXNET program (Baldocchi et al., 2001) involving new regional networks in Canada, Brazil, Asia, Australia and Africa. Principles The eddy covariance technique works by sampling the turbulent motions of upward and downward moving air parcels that transport trace gases such as CO2 across the canopyatmosphere interface. Statistical analysis (Reynolds’ rules of averaging (Reynolds, 1895)) is used to determine mean flux density of CO2 (F, mol m-2 s-1) averaged over some time span (normal 30-60 minutes) as the covariance between fluctuation in vertical wind velocity (w) and the CO2 mixing ratio (c = c/a, where c is CO2 density and a air density): F a w' c' (1) In Eqn 1 the over bars denote time averaging and primes represent fluctuations from the mean (e.g. c’ = c - c̄ ). Positive covariance indicates a net CO2 transfer into the atmosphere. The equation defining the conservation of mass means that over flat, homogeneous landscapes simultaneous measurements of vertical wind speed made with a three-dimensional sonic anemometer and trace gas concentration from a suitable sensor made in the surface boundary layer several metres above the canopy-atmosphere interface can deduce the exchange of carbon into and out of the underlying ecosystem. The frequency and duration with which these measurements are required is determined by the requirement to sample the cospectrum of turbulent motions that exist in the atmosphere. Sampling rates of 10 times per second usually capture the high frequency component of the cospectrum. To capture the low frequency contribution to the flux covariance the sampling duration must be long enough to encompass the atmospheric motions associated with convective boundary layer activity, but not so long as to be affected by diurnal variations in CO2 13 Eddy covariance concentrations. This tends to limit result in measurement and averaging periods between 30 to 60 minutes. In practise a number of instrumental, sampling and turbulence issues prevent a complete sampling of the cospectrum and this spectral filtering can be addressed by using theoretical or empirically derived transfer functions to correct the eddy covariance flux measurements, and in general such correction factors have been found to range between 1.04 and 1.25 for CO2 flux densities in the CarboEuro-Flux network (Aubinet et al., 2000). The type of instrument used to measure the scalar flux can also effect the computation of the covariance. For example, in the case of CO2 an infrared gas analyzer (IRGA) is used which does not actually measure mixing ratio (c), but rather molar density (c). This can vary in response to adding/removing molecules from the control volume, or changing the size of the control volume, which will occur in response to pressure, temperature and humidity variations in the boundary layer. When deriving the net CO2 flux density from these molar density measurements a correction has to be applied which takes these variation in temperature and humidity into account (Webb et al., 1980). This is particularly important for ‘open path’ sensors, such as the Li-Cor 7500 we will be using. Practical operation The University of Edinburgh roving eddy covariance flux measurement equipment used here consists of a Campbell Scientific CSAT3 sonic anemometer and a Li-Cor LI-7500 open path IRGA. The CSAT3 is a three dimensional sonic anemometer that uses advanced pattern recognition techniques to determine the time that the sonic signal takes to travel between pairs of transducers on three non-orthogonal axes. From these measurements, orthogonal wind speed and sonic temperature can be computed. The LI-7500 Open Path CO2/H2O gas analyzer consists of two components: the analyzer sensor head and the control box which houses the electronics. The sensor head has a 12.5 cm open path, with single-pass optics and a 1 cm diameter optical beam. The Infrared Source emits radiation, which is directed through a chopper filter wheel, focusing lens, and then through the measurement path to a cooled lead selenide detector. The absorption at wavelengths centred at 4.26 µm and 2.59 µm are used for measurement of CO2 and water vapour, respectively. Reference filters centred at 3.95 µm and 2.40 µm provide for attenuation corrections at non-absorbing wavelengths minimizing sensitivity to drift and dust, which can accumulate during normal operation. Campbell Scientific CSAT3 and Li-Cor LI-7500 mounted adjacently 14 Eddy covariance An important decision to be made at the start of an investigation is the height at which to mount the sensors. This needs to be well above the surface roughness sub-layer but within the well mixed surface boundary layer. The actually height will then depend on the nature of the underlying canopy, its zero-plane displacement height and roughness length. Above short, moor land vegetation this corresponds with a height of about 3m. Once the flux sensors have been set up and calibrated, they generally require very little attention from the operator. Normal field operation consists of little more than: 1. Checking the general physical condition of the sensors, possible cleaning the windows at both ends of the optical path 2. Checking any real time observations are giving sensible values for concentrations, fluxes, temperature, wind direction and time 3. Ascertaining the battery status/power supply 4. Downloading the logged data Analysis - how to convert the raw data into something usable To get a flux density measurement from the raw 10 Hz data over a given observation period can be daunting. Statistical analysis is required to determine the covariance, coordinate rotation is often used to ensure mean vertical and some time cross wind velocities are zero, and there is the application of correction factors discussed above (and others). This requires the use of custom software, such as that developed at the University of Edinburgh (EdiRe, http://www.geos.ed.ac.uk/abs/research/micromet/EdiRe/). EdiRe is a flexible, and (relatively!) user friendly software tool. It is adaptable to most eddy covariance raw data formats and has a graphical user interface simplifies development of processing routines and allows rapid redesign of routines to enhance question/answer cycle of data analysis. For reasons discussed above it is common practice to average flux densities over 30 minute intervals, but even then this often produces a time series with numerous ‘spikes’ and curiosities, which may have to be removed. These, and any other gaps in the record due to issues like instrument breakdown, calibration, or measurements over range, will need to be filled to generate a continuous time series of fluxes. In addition, measurements are also often discarded when the wind is coming from an unfavourable sector, or the sensors are wet. A number of approaches to gap filling have been tried, including empirical derived algorithms driven by meteorological variables, interpolation between adjacent data points (only useful for small gaps!) or binning data by hour over a one or two week period and using a time dependent mean to gap fill. Caveats - problems and weaknesses in the method To address ecological relevant questions we need to sum eddy covariance fluxes over extended periods to determine daily, seasonal or annual carbon balances. The accuracy of doing this will depend on a set of random and systematic bias errors caused by flaws in measurements, sampling and theoretical issues. With a well set up system measurement errors should be small, and is generally less than 7% during the day, and less than 12% at night. Natural variability in turbulence is in the order of 10-20%, which will cause variability in flux measurements under similar conditions. Constructing long-term averages reducing this sampling error to +5% (Moncrieff et al., 1996). The eddy covariance flux community continues to address two specific systematic bias errors. The first is the widely experienced failure to close the energy balance when eddy covariance measurements of latent and sensible heat do not match independently measured available energy (radiation and ground heat flux) (Wilson et al., 2002). The second apparent bias is towards 15 Eddy covariance underestimation of nocturnal ecosystem efflux, particularly during periods with low wind velocities (Moncrieff et al., 1996). Historically, eddy covariance flux measurements have been made over homogeneous vegetation types. To interpret CO2 fluxes from heterogeneous vegetation mosaics additional information is required about the ‘flux footprint’, that is the area of the canopy-atmosphere surface “seen” upwind of the sensor and for which the measurements are valid. This issue is beginning to be addressed using footprint models (e.g. Schmid and Lloyd, 1999). Anthoni, P.M. et al., 2002. Seasonal differences in carbon and water vapor exchange in young and old-growth ponderosa pine ecosystems. Agricultural and Forest Meteorology, 111(3): 203-222. Aubinet, M. et al., 2000. Estimates of the annual net carbon and water exchange of forests: The EUROFLUX methodology, Advances in Ecological Research, Vol 30. Advances in Ecological Research, pp. 113-175. Baldocchi, D. et al., 2001. FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bulletin of the American Meteorological Society, 82(11): 2415-2434. Baldocchi, D.D., Hicks, B.B. and Meyers, T.P., 1988. Measuring Biosphere-Atmosphere Exchanges of Biologically Related Gases with Micrometeorological Methods. Ecology, 69(5): 1331-1340. Baldocchi, D.D., Vogel, C.A. and Hall, B., 1997. Seasonal variation of carbon dioxide exchange rates above and below a boreal jack pine forest. Agricultural and Forest Meteorology, 83(1-2): 147-170. Chen, J.Q. et al., 2002. Biophysical controls of carbon flows in three successional Douglas-fir stands based on eddy-covariance measurements. Tree Physiology, 22(2-3): 169-177. Geider, R.J. et al., 2001. Primary productivity of planet earth: biological determinants and physical constraints in terrestrial and aquatic habitats. Global Change Biology, 7(8): 849882. Lee, X., Massman, W.J. and Law, B., 2004. Handbook of Micrometerology A guide for surface flux measurement and analysis. Kulwer Academic Publishers, Dordrecht, 250 pp. Moncrieff, J.B., Malhi, Y. and Leuning, R., 1996. The propagation of errors in long-term measurements of land- atmosphere fluxes of carbon and water. Global Change Biology, 2(3): 231-240. Running, S.W. et al., 1999. A Global Terrestrial Monitoring Network Integrating Tower Fluxes, Flask Sampling, Ecosystem Modeling and EOS Satellite Data. Remote Sensing of Environment, 70(1): 108-127. Schmid, H.P. and Lloyd, C.R., 1999. Spatial representativeness and the location bias of flux footprints over inhomogeneous areas. Agricultural and Forest Meteorology, 93(3): 195209. Webb, E.K., Pearman, G.I. and Leuning, R., 1980. Correction of Flux Measurements for Density Effects Due to Heat and Water-Vapor Transfer. Quarterly Journal of the Royal Meteorological Society, 106(447): 85-100. Wilson, K. et al., 2002. Energy balance closure at FLUXNET sites. Agricultural and Forest Meteorology, 113(1-4): 223-243. 16 Soil stocks Measuring Soil Carbon Stocks (Instructor: Phil Ineson) Introduction The objective of this exercise is to quantify the organic C stock held in the soils within the ‘1 to 4’ square km grid cells. This will be achieved through a programme of field sampling, followed by subsequent analyses of samples in the laboratory to determine total organic C content. Soils are extremely heterogeneous and this creates considerable problems when measuring C stocks or attempting to detect change. This heterogeneity is the result of many contributory factors, including local geology, hydrology and management and has to be allowed for as much as possible when planning and executing the field samplings. According to existing soil survey maps there are several (~5) soil associations represented within the ‘1 to 4’ km and these maps will have been produced based largely on distribution of vegetation distribution coupled with a very limited number of soil test pits. Surface vegetation frequently provides a very useful, rapid and non-invasive method of locating the extent of a particular soil type and assists in defining boundaries between soil associations. Unfortunately, soil organic C content frequently changes down a soil profile, yet tends to be fairly consistent throughout a particular soil horizon within a particular soil type. It is therefore necessary to dig appropriate soil pits (or take cores) and to quantify the depths and C contents of each horizon going down to a depth of 1 m. So, it is important that horizons are identified, measured and sampled to a depth of 1 m. Additionally, since you will report the final C stocks on an area basis (kg C per m2) it is important to have an appropriate means of converting between soil volumes and weights; this is achieved by measuring bulk density (BD; soil mass per unit volume) by soil association and horizon. Furthermore to relate the soil carbon to the national database we might have to consider stoniness as to obtain final C density values. Fig 1 and Table 1-1 (below) provide a written and visual representation describing the general nature of soil horizons and it may help you in planning how to sample. The actual analysis of soil organic C content is ideally performed by combustion of the sample in the presence of oxygen and measuring the amount of CO2 produced, using either gas chromatography (GC) or some form of infra-red gas analysis (IRGA). A further alternative method is to use wet combustion (e.g. the Walkley-Black method) yet the most frequently used indirect measure for soil organic C determination is weight loss resulting from combustion at 17 Soil stocks high temperature; this method is usually referred to as loss-on-ignition (LOI). LOI analyses are frequently performed on dry soil samples combusted for several hours in a muffle furnace. The procedure we describe here follows the method described by Schulte and Hopkins (1996) in which samples of soil are sieved, oven-dried and then weighed sub-samples are placed in crucibles and subjected to heat at 360oC. Various factors are used for the conversion of organic matter (LOI) to organic carbon. The figure for the ratio organic matter/organic carbon which has often been accepted is 1.72 and is based on the assumption that soil organic matter contains 58% carbon. However, there are different factors suggested for different soils (Howard, 1966): Mor: 1.77-1.93; alluvial: 1.81-1.83; peat: 1.90-1.95 and mull: 1.97-2.07. However, weight loss at each stage is calculated and, for organic soils such as these, we can use the following equation to convert LOI to soil organic C content (Konen et al. 2002): Soil organic C content (g C kg-1) = 0.6094 * LOI (g kg-1) + 0.1949 Materials & Methods Spades Peat corer Trowels Plastic bags Soil maps Soil profile descriptions Tape measures Rulers Plastic cores Field mallet Knives Wooden boards Sample bags Marker pens GPS system Muffle furnace (360oC) Drying oven (105oC) Crucibles Balance Sampling. It is necessary to decide the optimal sampling strategy prior to going into the field. Using the maps and images provided you should decide a stratified sampling system for the soil associations in the km2 under study. We are not providing you with a prescriptive methodology but are leaving it to each group to devise an optimal sampling strategy. The main constraint is that you have only 90 crucibles available to you for LOI determination and you are therefore limited to a total of 90 soil samples for organic C analysis. You are advised to consider carefully the descriptions of the soil associations to decide on a profile sampling strategy and the number of cores/pits to sample. Note that the sampling of peat soils can best be achieved using the peat corer provided; mineral soils require a more extensive soil profile to be dug and for soil sampling plastic pipes to be inserted. Adequate plastic sample bags are provided. Care must be taken not to compress soils when BDs are being determined. Always label any material appropriately with an identifier and date. Analysis. Samples of known weight and volume are placed in the drying oven to determine moisture content and to prepare samples for the muffle furnace. Samples are oven-dried at 105°C overnight, cooled in a desiccator, and weighed. Equivalent volumes (approximately 8 g) of < 2mm air-dry soil are then placed into 15 cm3 crucibles and samples combusted at 360°C for 2 h in a muffle furnace. Samples are then transferred after the 2-h combustion period to he drying oven at 105°C for several hours. Samples are then cooled in a desiccator and re-weighed. Loss-onignition is calculated using the following equation: LOI (g C kg-1) = (oven dry soil wt – soil wt after combustion) * 1000 oven dry soil wt 18 Soil stocks Reporting. Using a combination of available maps, group surveying, sampling and subsequent LOI analysis the group should derive a single figure for the soil C stock in the sample km 2. Remember to subtract any stone volume if necessary. This value will then be compared with data from the UK national data base for the same square. We also want the group to report on where they consider the greatest sources of error and to provide some consideration of how to derive an uncertainty estimate for the final figure. References: Schulte, E.E. and B.G. Hopkins (1996). Estimation of organic matter by weight loss-onignition. p. 21–31. In F.R. Magdoff et al. (ed.) Soil organic matter: Analysis and interpretation. SSSA Spec. Publ. 46. SSSA, Madison, WI. Michael E. Konen, Peter M. Jacobs, C. Lee Burras, Brandi J. Talaga and Joseph A. Mason (2002). Equations for Predicting Soil Organic Carbon Using Loss-on-Ignition for North Central U.S. Soils. Soil Science Society of America Journal 66:1878-1881 (2002) 19 Vegetation Carbon Stocks Measuring Vegetation Carbon Stocks and Leaf Area (Instructor: John Grace) Introduction Terrestrial Ecosystems of the world represent a carbon store of about 2100 Gt of carbon, and a single hectare of land may have from ten to several hundred tons of carbon (Table 1). It is important to know the size of the carbon stocks, and especially to characterise the changes in stocks when land is managed or converted from one use to another. The carbon stores in NW European dwarf shrublands are not well known, and during this course we hope to produce some provisional figures. For land-based measurement (as opposed to measurements from remote sensing) we generally use the simple method of destructive sampling. Often, we measure Leaf Area Index at the same time, as this describes the fundamental capacity of ecosystems to capture energy from sunlight. Table 1. The estimated carbon stocks of ecosystems of the world, and the corresponding values of Net Primary Productivity, NPP (based on Saugier et al. 2001). 1 Pg = 1015 g = 1 gigatonne= 109 tonnes (1Gt). Biome Area (million km2) Average C stocks (tC ha-1) Total C pool (Pg C) Average NPP (tC ha-1 year-1) Total NPP (Pg C year-1) Trop. forests Temp. forests Boreal forests Arctic tundra Med. shrublands Crops Trop. savanna & grasslands Temp. grasslands Deserts Ice TOTAL 17.5 10.4 13.7 5.6 2.8 13.5 27.6 316 280 288 208 314 11 118 553 292 395 117 88 15 326 12.5 7.7 1.9 0.9 5.0 3.1 5.4 21.9 8.1 2.6 0.5 1.4 4.1 14.9 15 27.7 15.3 149.3 121 61 182 169 3.7 1.2 5.6 3.5 2137 62.6 Principles Carbon stocks To measure the carbon stocks in herbaceous or shrubby ecosystems, we generally harvest the biomass, necromass and soil from a suitable number of sample plots, dry it in an oven set to 80 o C, determine the mass, and convert from mass to carbon content using a particular conversion factor. It is best to determine the carbon percentage using a mass spectrometer but most biomass is close to 50% carbon, and so it is usually good enough to assume a conversion factor of 0.50. When dealing with woodlands or forests, this direct method is not practical and then we find empirical relationships between mass and diameter (‘allometric relationships’) and estimate the mass of trees from their diameter. There is a large and rapidly growing literature on this subject, in which allometric relationships are given (Zianis & Mencuccini 2004). 20 Vegetation Carbon Stocks Leaf Area Index Leaf Area Index (LAI) is defined as leaf area per land area, and it is a dimensionless quantity. To measure LAI directly we simply remove the leaves from square sample plots and measure their combined area. This could be done with graph paper but is more conveniently and rapidly done using a mechanical device known as an area meter. For really rapid work, or work in woodlands and forests, we use an optical method. There are two optical methods available to us: (i) use a special camera that can view an entire hemisphere to capture an image of the canopy viewed from the ground and (ii) use a Li-Cor canopy analyser which takes a measurement and makes the calculation in real time based on the light that penetrates the canopy at different angles. Both of these methods depend on the same ‘gap-fraction’ theory: the amount of gap in the canopy for light to penetrate declines in a predictable manner as the LAI increases. By measuring the gap in zones of varying angular elevation, and after making a few assumptions, we get the LAI (Norman & Taylor, 1989).The photographic method is not very useful for dwarf shrub communities because the camera cannot ‘fit’ under the canopy. The canopy analyser can easily be poked under the canopy, and readings taken in a number of locations. Note that the data logger of the Li-Cor canopy analyser does the calculations of mean and standard error for us. Fig. 1 The Li-Cor LAI-2000 canopy analyser. The right diagram shows the optical arrangement of the sensor whereby an image of the canopy is projected onto a photo-detector made of concentric rings. The left diagram shows sensor and hand-piece connected to the logger Li-Cor datasheet. Practical Aspects Sampling Once the study site has been identified, decide on the sampling strategy required to capture the pattern of variability (Fig. 2). For a more or less homogenous moorland, you might want to define a number of random stations for intensive sampling. However, if there is a visual pattern of well-drained slopes and, for instance, valley bog, you may wish to find their areas in m 2 and designate these as the two strata from which you sample at random. This would be a simple form of stratified random sampling. For unbiased quantitative results it is better to use stratified random sampling. 21 Vegetation Carbon Stocks Fig. 2. Two ways to lay out the sample stations in a landscape with five sorts of vegetation. Left, purely random (determined using random co-ordinates); right, systematic sampling. A third type (not shown) is random sampling within the designated sorts of vegetation (stratified random sampling). Irrespective of the sampling strategy, you need to address the issue of sampling intensity. For above-ground vegetation, it is usual to lay out n sample squares of area a (m2) in the study area and then proceed to harvest them. The question arises, how large does n and a need to be? Of course, ‘the more samples the better’ but what is ‘reasonable’? For herbaceous and shrubby vegetation the squares need to be large enough to encompass the basic pattern, otherwise the population variance will be too large. For dwarf shrubs a is typically 0.5 m2. To determine how large n should be for a stated precision, we should first do a pilot study to estimate the population variance (s2). Do this by sampling 5-10 plots. s2 ( x x) 2 n 1 Then we can find the find the estimated standard deviation s by taking the square root of the variance. And we can find the standard error of the mean (S) for any value of n as: S s n And we can find the 95% confidence interval Ic for any value of n as: I c tn S where tn is found from t-tables (Annex 1). For example, using a two-tailed t table with a probability level of 0.05, t decreases from 2.66 at n=6 to 1.98 at n=120. These calculations are conveniently carried out on a spreadsheet (please try them with a dummy data set). In fig. 3 we show the result for a data set of biomass measurements from a pilot study: six determinations gave 57, 120, 98, 87, 110, 111 g m-2 . 22 Vegetation Carbon Stocks 95% confidence interval 30 25 20 15 10 5 0 0 50 100 150 n, sample size Fig. 3. Dependency of the 95% confidence interval on the number of samples taken, for the data set given in the text (mean = 97.2, s estimated as 22.7). This leads us to the conclusion that we would need 100 samples to get within 5 g m-2 of the true value. Sorting Please try to identify the species. You will be required to sort the harvested material into fractions for the main species, corresponding to leaf, woody biomass, and litter. There are problems of classification that we need to agree upon at the start. Here are two issues that we need to discuss. (i) In many communities the litter does not decay rapidly, and the soil profile has a recognised organic layer on the surface. Is this regarded as litter or part of the soil? (ii) the bog moss Sphagnum has a photosynthesising canopy of unknown LAI. Pull out one plant and see that the distinction between green biomass and litter is not a sharp one. How shall we treat Sphagnum and other bryophytes? Drying Biomass and necromass contain a variable fraction of water, and so we begin by driving off the water in an oven set at 80 oC. Usually it takes only 8 hours to dry leaves and fine shoots, but much longer for bulky wood samples such as limbs of trees. One ought to ‘dry to constant mass’ which requires repeated removal from the oven, cooling and weighing until you are sure that all water has been driven off. We might have to pass on some of the weighing work to the group of the next day as to make sure that the samples are ‘dry enough’. Always label any material appropriately with an identifier and date. Caveats - problems and weaknesses in the method Please bear in mind that any measurement of LAI using optical equipment will always count the dead leaf biomass as well. Consequently, for canopies with large amounts of leaf litter within the canopy this will lead to overestimating LAI. Just think about a Pine forest with dead leaves hanging in between the branches or have a look at the heather plants – do they have dead leaves still attached to their branches?! 23 Vegetation Carbon Stocks References IPCC (2004) Land Use and Land Use Change, Good Practice Guidelines. http://www.ipccnggip.iges.or.jp/public/gpglulucf/gpglulucf_files Norman JM & Campbell GS (1989) Canopy Structure. Pp 301-325 in Pearcy RW, Ehleringer J, Mooney HA & Rundel PW. Plant Physiological Ecology, Field Methods and Instrumentation. Chapman & Hall, Lndon. Saugier, B., Roy, J. & Mooney, H.A. (2001) Estimations of global terrestrial productivity: converging toward a single number? Pp 543-557. In Roy J, Saugier B & Mooney HA Terrestrial Global Productivity. Academic Press, San Diego. Zianis D, Mencuccini M (2004) On simplifying allometric analyses of forest biomass. Forest Ecology & Management 187, 311-332. 24 Net Ecosystem Productivity NEP chamber (Instructor: Mathew Williams) Introduction to the technique Our goal is to measure the net exchange of CO2 from the soil and vegetation, and record how C exchange responds to changing environmental conditions. Our approach is to seal a unit of vegetation and soil under a clear Perspex chamber and record the changes in CO2 concentration within the chamber, via an infra-red gas analyser. By manipulating the conditions in the chamber (for example, shading the chamber or darkening it completely) we can determine how vegetation responds to changes in light, and also determine the respiration of the vegetation and soil. The ability to easily manipulate the environment and to separate photosynthesis from respiration are distinct advantages of this chamber approach, compared to eddy flux techniques. Of course, the scale of measurement by chamber (1m2) is much smaller than the footprint of an eddy flux system (1ha+), and manual chambers are not suitable for continuous measurements. Your goal is to characterise the light response of photosynthesis, and explores relationships between photosynthesis and respiration for the key vegetation cover types within the catchment. Each complete measurement set takes ~1 hour, so you can collect ~4 data sets. You must decide where to collect your data. By determining light response curves, you will be able to construct a simple models of photosynthesis, and estimate daily GPP for the entire day using radiation data from the meteorological station. Principles Perspex is transparent to photosynthetically active radiation, so plants and soil can be enclosed in a clear chamber and their activity monitored by measuring the changes in the concentration of gases within the chamber. In the light, photosynthesis will draw down CO2, while transpiration will increase water vapour concentration. Both CO2 and water vapour are greenhouse gases, so their concentration can be determined by the absorption of distinct components of the infra-red spectrum. The technique of infra-red gas analysis (IRGA) is well developed, and we use a system constructed by LI-COR to monitor the concentration changes in the chamber head space. The Perspex chamber itself has internal dimensions of 1 m wide x 1 m deep x 0.25 m tall. The chamber sits on an flat aluminium base with itself rests on 4 steel legs which are driven into the ground. The legs’ lengths are selected based on the height of the vegetation. The air tight seal between chamber and base is achieved with a foam tape gasket. The aluminuium base has an opaque plastic skirt long enough to drape down to ground level. A heavy chain weights the skirt to the ground to ensure an air-tight seal. The chamber has 4 small internal fans powered by a separate 12v battery to circulate the air. Practical operation - a step-by-step field protocol 1. Locate the measurement site, and then set up the base, with each of the corner pegs slotted onto a hollow leg, driven into the ground. Make sure the base is level. 2. Seal the chamber to the ground with the skirt plastic and heavy chain. Take care when sealing the chamber so that small hummocks or depressions are tightly sealed. 3. Once the chamber base is set up, lay the chamber upon it and check the seal visually. Then remove the chamber to avoid moisture build-up. Connect the fans to their power source and start them up. 4. Seal the chamber to the base and initiate the closed programme on the LICOR 6400. When light conditions are high (PAR at 1000 or greater) and steady, take 2 measurements at high light, followed by 2 each at 3 levels of light reduction. The light reduction is achieved by using mosquito netting laid over the chamber. Finally, take 4 dark measurements, with the 25 Net Ecosystem Productivity blackout blanket over the chamber. At each measurement, also record the temperature in the chamber and the PAR. Each measurement lasts 60s and data is recorded at a 2s interval. 5. After and between each measurement, lift the chamber to restore ambient conditions – watch the graphs of CO2 and H2O until they have stabilized at pre-measurement levels. 6. When light conditions are low or unsteady, alter the order of the measurement series, and do your best to cover a range of PAR values as well as take measurements when the light is at its steadiest. 7. Measure the volume within the base by taking depth measurements at 36 points using a grid (PVC frame with strings tied across) laid on top of the base. In plots where the substrate is unstable, an initial and final volume can be measured to account for any sinking due to the weight of the chamber. 8. To estimate the vegetation cover in the chamber, use the Skye 2-channel sensor. This sensor measures reflectance in the red and near-infra red. Make sure to record both the channel readings as you will need these to calculate the NDVI, the normalised difference vegetation index. Make 9 measurements in a 3 x 3 grid, so that you record the heterogeneity of the vegetation. 9. Record the soil temperature at 9 locations in a 3 x 3 grid. 10. Estimate the soil and vegetation percentage cover, and note the dominant species. Analysis - how to convert the raw data into something usable Your goal is to generate the parameters for a simple model of photosynthetic light response, and to correlate the model parameters with your measurements of canopy cover (NDVI) and soil respiration. You can also relate your estimates of soil respiration to soil temperature. Calculate fluxes from chamber concentrations recorded by the Li-Cor 6400, according to the formula Fc = w V dC/dt A (1) where Fc is net CO2 flux (μmol m-2 s-1), w is molar air density (mol m-3), V is the chamber volume (m3), dC/dt is the slope of chamber CO2 concentration against time (mol mol-1 s-1) and A is the chamber surface area (m2). The chamber volume is simply its area (1 m2 multiplied by 0.25 (chamber height) plus the average in m of the 36 depth measurements). w = P/TK where P is pressure (Pa), Tk is air temperature (K), and is the gas constant (8.3144 Pa m3 mol-1 K-1). Quality check your data by plotting light response curves. Are they smooth? Are your dark measurements reproducible (i.e. give similar estimates)? If not then you likely have problems with leaks. Fit photosynthesis model to light response curves. Convert all NEP data to GPP using dark respiration data to provide and estimate of Re [all data are assumed scalars]. GPP = NEP + Re (2) The light response model is 26 Net Ecosystem Productivity GPP Pmax I kI (3) where I is photosynthetic photon flux density (mol m-2 s-1) from the PAR sensor, and Pmax and K are parameters, the maximum photosynthetic rate (mol m-2 s-1) and the half-saturation value (mol m-2 s-1) respectively. Enter your light response curve data into an excel spreadsheet with 3 columns for PAR, GPP, temperature. Label a fourth column “model GPP”. To the right of the data & model columns, create two new columns, one for parameter names, and to its right, a column for parameter values. In the name column, enter the parameter names Pmax and K. In the value columns, enter initial guesses for the two parameters, e.g. 10 and 200. You can “name” these value cells for ease – highlight the four cells with the names and values, then click on “Insert”, “name”, “create” and tick “make names in left column”. Click OK. Now you can simply write “Pmax” or “K” in any formula, and the numbers in the values column will be substituted. Make sure that solver is activated in your spreadsheet. Go to “Tools”...”add-ins” and then click on the solver add-in. In the “model GPP” column, for each row with data, write the model equation (3), making sure that “Pmax” and “K” refer to the named parameter cells, while I refers to the specific irradiance for the row. Once complete, each row will have a modelled GPP based on the initial values you set for “Pmax” and “K”, and the measured irradiance. To provide the best estimate of the model parameters, we minimise the sum-of-square differences between “model GPP” and “GPP”. Create a new column for the squared difference calculations, with each row corresponding to the squared difference between model and observation. In another cell elsewhere, sum these squared differences. Use the solver to determine the best parameter estimates. Go to “Tools”, “Solver”. Set the target cell to the cell with the sum of square differences (SSD). Click the “min” button, as the goal is to minimise the SSD. Click in the “by changing cells” box, and then highlight the two initial values set for “Pmax” and “K”. Then click solve. Excel will use optimisation methods to find the best parameter estimates. Check these are good by plotting the modelled light curve over the data you collected. Collate the daily time series of PPFD. Construct a model to estimate instantaneous GPP for each half hour of the day. Integrate these date to generate a daily GPP estimate for each of your chamber locations. Decide how to aggregate your data to produce a catchment estimate. Plot your Pmax values against your NDVI NDVI = (RIR- RVIS)/( RNIR + RVIS), (3) Plot Re against soil temperature and against Pmax - are there any signs of relationships? 27 Net Ecosystem Productivity Caveats - problems and weaknesses in the method Closed systems are inherently problematical because the conditions within the chamber change immediately the system is disconnected from the environment. Thus, within the illuminated chamber there is an immediate draw down in CO2 concentration and a rise in water vapour concentration. These changes in conditions will alter the functioning of the vegetation over time. Thus, the longer the chamber is left in position, the less the measurements reflect the activity of the ecosystem to the ambient conditions outside the chamber. For this reason, measurements must be made rapidly so the chamber is only in place for a minute or two, and internal conditions change only by a small amount during this time. Reading list Johnson L.C., Shaver G.R., Cades D., Laundre J.A., Nadelhoffer K.J. & Rastetter E.B. (1996) Effects of drainage and temperature on carbon balance of tussock tundra microcosms. Oecologia, 108, 737-748. Shaver G.R., Johnson L.C., Cades D.H., Murray G., Laundre J.A., Rastetter E.B., Nadelhoffer K.J. & Giblin A.E. (1998) Biomass and CO2 flux in wet sedge tundras: responses to nutrients, temperature and light. Ecological Monographs, 68, 75-97. Vourlitis GL, Oechel WC, Hope A, Stow D, Boynton B, Verfaillie J, Zulueta R, Hastings SJ. 2000. Physiological models for scaling plot measurements of CO2 flux across an arctic tundra landscape. Ecological Applications 10: 60-72. Williams M., Street L., Wijk M.T.V. & Shaver G.R. (in press) Identifying differences in carbon exchange among arctic ecosystem types. Ecosystems. Equipment list Perspex chamber (1 m x 1m x 0.25 m) Aluminium Base (1 m x 1 m) Base legs (various lengths in groups of 4) Hammer Thick opaque plastic skirt Duct tape 12V battery to power fans Small saw or secateurs to cut stems Heavy chain (4 m +) LICOR LI-6400 Photosynthesis system PAR sensor Shade netting Blackout blanket Skye 2-channel light sensor for NDVI Soil thermometer PVC frame with grid for depth measurement Spirit level Straight ruler and measuring tape. Acronyms GPP = gross primary production NEP = net ecosystem production Re = ecosystem respiration NDVI = normalised difference vegetation index 28 Soil Respiration Soil Respiration (Instructor: Andreas Heinemeyer) Introduction to the technique Our aims are (i) to measure the soil CO2 efflux (Rs), (ii) to determine how much this flux contributes to the total carbon budget, (iii) to understand measurement uncertainties and (iv) to assess how Rs is controlled by changing environmental conditions such as temperature and light. The system we are going to use is a LI-COR 8100 closed-dynamic system (LI-COR 8100) which uses chambers placed on soil inserted PVC collars. The approach is based on the increase in headspace CO2 from a given soil surface over time within the chamber volume, measured with an infra-red gas analyser; where air is circulated between the chamber and the IRGA. We will use two different systems: (i) a single portable survey system either for a 4 inch or 8 inch collar size and (ii) a purpose build automated system which has been multiplexed by CTCD using 12 long-term LI-COR (8 inch) chambers. The complete system thus enables us to perform (A) frequent measurements on 12 plots around the flux tower using the long-term automated system and (B) many spot measurements for spatial coverage within the flux tower footprint and at an experimental site. A major ‘unknown’ in eddy flux tower measurements is Rs. Consequently, the first aim of our team will contribute to a better estimate of the contribution of soil respiration to the measured NEE by obtaining frequent and direct measurements of soil respiration within the flux tower footprint area (1ha+). Of course, the scale of measurement by the 4 and 8 inch chambers (0.0083 m2 and 0.0346 m2, respectively) is much smaller than the footprint of an eddy flux system (1ha+), and thus the second aim will be to use a combination of 12 long-term plots with many spot measurements around the flux tower to enable us to up-scale soil respiration from spot to ecosystem scale and estimate its uncertainty. Obviously, certain key environmental variables, such as soil temperature, soil moisture and water table will need to be monitored during the soil respiration measurement period in order to assess the impact of those factors on soil CO2 efflux. The third aim is therefore to obtain a correlation matrix for soil respiration data vs. soil temperature (e.g. Q10) and soil moisture. The fourth aim will be to assess land use impacts such as liming and burning on soil C cycling in organic rich soils. It is also intended to address certain technical issues in the correct chamber design such as the role of pressure vents. Your goal is to characterise the response of soil respiration to key environmental factors, and to explore relationships between soil respiration and land use management in the catchment. Each complete measurement set takes 2 - 5 minutes, so you can collect ~12 data points per hour using the automated system and 20 data points per hour using either survey chamber. As to avoid disturbance effects, 40 collars (20 cm depth) for the small survey chambers and 12 collars (20 cm depth) for the long-term chambers were placed into the ground around the flux tower (100 x 100 m; Overdale) already in July. You will also find some more collars in burnt heather areas near the tower. Further, within a 100 x 100 m area away from the flux tower site (Blishmire) there are 4 plots (5 x 5 m) each to which 1 lime application (at 500g/ m2) were given on 18th July which can be compared to 4 control plots (no added lime); within this plot area there are 24 small survey collars (3 within each plot; 20 cm depth). Placing thermistor (soil temperature), theta probes (soil moisture) and water table PVC tubes (each 1.5 cm diameter; 50-100 cm depth) within the measurement areas will enable us to determine response curves, which we will use to construct simple models of soil respiration, and estimate daily soil respiration losses for the wider flux footprint area using other data available to you such as land use and soil type areas. 29 Soil Respiration Principles Carbon dioxide movement from soils is the result of a concentration gradient between the soil and the open atmosphere but mass flow can also occur under windy conditions caused by pressure fluctuations induced by atmospheric turbulence. Soils release CO 2 mainly as a result of biological activity mainly from three pools i.e. microbial soil, microbial litter and root (n.b. in most cases mycorrhizal) respiration. The combined flux can be measured by one of several methods: (1) closed dynamic system - increase in headspace CO2 within a known enclosed volume of air where air is circulated between chamber and IRGA, (2) open dynamic system – increase of chamber air compared to a reference incoming ambient air line, (3) chemical system – fixing released CO2 to NaOH (4) soil CO2 profile – flux rate based on CO2 gradient over certain soil depth or less common (5) micrometeorological – either measuring the CO2 air profile or the vertical fluxes (eddies) of CO2, both are based on mixing of soil and ambient air. We are going to use method 1 (e.g. LI-COR 8100) in which soil flux is determined by measuring the change in CO2 concentration with time in an enclosure where the soil surface is isolated from the atmosphere (see Fig. 1, LI-COR Fig. 1 manual). By assuming that the rate of change in gas concentration is constant, the appropriate equation for calculating the flux is: Q = ∆c V ∆t S (1) where ∆c is the gas concentration (μmol m-3) difference within the chamber in the time interval ∆t (s), V is the volume of air within the entire enclosure (i.e. chamber + tubing + IRGA), S is the soil surface area covered by the chamber (m2). The chambers volumes and soil area are given in the software. However, as we use collars in the ground we have to correct for additional air volume between chamber bottom and soil surface (see below). The technique of infra-red gas analysis (IRGA) is well developed, and we use a system constructed by LI-COR to monitor the concentration changes in the chamber head space. Measurements of dry CO2 (corrected for water vapour) are made every second and data can be stored on an internal memory or a flash card and are transferred to a computer at the end of each field day. N. b. Dry CO2, 10 s flux rates and many other data can be monitored in real time using wireless technology with either a Palm handheld or a laptop. You will be shown how to use this equipment in the field. Obviously, you have to be careful not to breathe into the chamber when placing it on the collar, you should also avoid putting too much pressure on the chamber when placing it onto the collar as this will cause air to be pressed out of the ground into the chamber. Remember: The weight of the chamber itself is enough to form a good seal but you do have to check for branches etc. lying on the soil collar. 30 Soil Respiration Practical operation - a step-by-step field protocol 11. Locate the measurement site (see Annex 3). Open the yellow LI-COR 8100 box and check that either the power inlet (automated system) or battery (survey system) are properly connected, then turn on the IRGA (press ON/OFF once). Close the box (the IRGA-optical bench needs to warm up to >51°C before any measurement can be taken). 12. Wait until both the READY (communication with chamber OK) and the IRGA READY (ready for measurements) light comes on and check that the bench temperature is >51°C. This will take about 5-10 minutes (keep on checking by opening the box or better by interrogating with the Palm (will be explained to you). The ACTIVE lamp will only come on once you start a measurement. 13. Whilst you are waiting check that: the chamber is attached properly and clean on the inside, survey chamber has the auxiliary sensor interface (soil temperature probe) attached, the Delta-T data logger is still logging (soil temperature and moisture) and possibly relocate the probes (press READ and use the up and down arrow). 14. Once the bench temperature has reached >51°C (IRGA READY) you can start measuring: Start the LI-8100 software (or the Palm OS LI-COR interface). Place the chamber onto the collar (survey chamber). Connect via Serial Port or TCP/IP; n.b. communication with the long-term system will be via serial port only (click) Connect Check the data and time Utilities Remote Date and Time Choose New Measurement and set Observation Length (~ 3 minutes), Observation Count (whatever but for long-term system 12!), Observation Delay (~1-2 minutes) and Dead Band (~ 35 s). Leave RH box unchecked. If using the long-term system then check the Repeat box as 1 hour repeat x-amount of times (daytime as many hours you will be there to do and overnight run about 20); leave Turn off flow pump box unchecked. Next Log data on the Onboard Internal flash memory (n.b. but external flash card for longterm system!), log every second and check the Log raw data records box and check the Voltage 1 and Thermocouple 4 and all other right hand column data. Next Enter a File Name, Treatment Label (e.g. flux tower, footprint, burn or lime) and Comments such as sunny, few clouds, very windy, Start immediately, Chamber Offset to whatever appropriate (ask but ideally the same for all e.g. 5 cm!). Edit Choose Instrument Name e.g. survey 4 inch, IRGA averaging = 1 sec, change Instrument Settings to 4 inch or 8 inch default accordingly, set flow rate for 4 inch ~ Med and for 8 inch chamber ~ High click apply and OK. Start Measurement. The chamber should start to close immediately! 15. Once the chamber closes check the running data with the Palm or laptop (n.b. serial port only with the long-term system). (View Monitoring). You can change the displayed data (such as dryCO2, RH, 10 s flux, T4 (survey system only: soil temperature, V1 (long-term system only: chamber position) etc. 16. Write down the time of the start of the measurement and note any disturbance (trampling, wind etc.) or a wrong measurement (badly placed chamber, twig on collar etc.). When using the survey or long-term chamber system: 31 Soil Respiration Download any data first using the laptop (n.b. long-term system: (i) communication only via serial port and (ii) data transfer will be by using the external flash card which can be removed after turning off the LI-COR box and then reinserting it into the laptop!). (Connect Utilities File Manager Transfer. N.b. you might have to click Stop Measurement first in case the long-term system is still active. Just download to the Desktop Look at the data with the File Viewer (double click on the file) and check that they are OK! Now transfer the files to the appropriate folder: E:\CTCDSS\Respiration\LICOR. Delete the transferred data from the internal LI-COR memory only once you have saved them on the laptop Close the File Manager! Long-term system: Reset the black control box! (Open press Stop press Start). The numbers 1 and 12 should alternatively flash slowly. Close the black box. Reset the LI-COR as above by connecting via TCP/IP then Start Measurement. For chamber offset determination: For 4 inch collars: measure the distance between soil surface and collar rim. For 8 inch collars: measure the distance between the soil surface and the lower edge of the chamber metal base (use two rulers for this!). For data logger download: Connect logger cable to serial port of the laptop. Open MyLogger wait until connection has been established Stop logging Dataset Retrieve name appropriately e.g. MTfluxtower16/08/05 and save on desktop wait until the dataset is displayed check if all data are OK close this data window. Delete all records Delete Retrieved Records. Restart logger Logger Start. 32 Soil Respiration Analysis - how to convert the raw data into something usable Flux data You are lucky as the LI-COR software does all the calculations for you. However, sometimes you need to recalculate the fluxes due to a different chamber offset (e.g. 1.5 cm and not 2.0 cm) or the need to alter the time frame used for the flux calculations (e.g. 80 – 160 sec. instead of the intended 40-120 sec.) due to a delayed initial mixing. A look at the data in the file viewer will help to decide if everything went OK: Open the File Viewer, open the wanted file double click on a data record Regression Analysis. You will see a plot of all dryCO2 (water corrected) data against time. The green and red line show begin and end of the flux calculation, this should be a near linear slope. You can move the two lines in order to capture the most appropriate time and watch how the flux rate changes. You can compare the linear rate calculation (Mean Flux) vs. the exponential rate (IV flux). Quality control checking… . Look at all the data (are they good or were there any problems?!) and then decide on a common time frame for all data (or do only individual rows) and then Edit Recompute… . Check all rows ( or only selected) and set new Curve Fit timings Recompute. N.B. you should do this anyway as only this way it will give you the R 2 (only 0.0000 so far) for the exponential and linear fit, so even if the timings are OK do this recomputation but set the timings slightly different e.g. 40-121 sec. (make sure you either check All or Selected). If needed you can also adjust the Chamber Volume or Soil Area or the Chamber Offset by checking the individual box BUT note that if All is checked it will do this for all observations. It is more likely that you want to do the chamber offset individually for a certain chamber. Then you first have to select those rows in the File Viewer data table (hold Ctrl and click individual rows), then Edit Recompute… check the Selected box and fill in new Chamber Offset value. After post-processing the data we can now export the final dataset as a .txt file which we will then import into Excel as a spreadsheet. Do File Export and check All and Define which Columns to be exported: I recommend: Orig. File; Obs; Date; Flux; ExpR2; Mean Flux; LinR2; T chamb; RH; V1 (for long-term system only); T4 (for survey chamber only). Give the file a name and Save as a .txt file. Open Excel Open (make sure you select All Files (*.*) and select delimited Next select Tab Next Finish! You should now see all the data (in case not…maybe you have to enlarge some column width!). Data logger data The environmental data collected with the logger have been saved as .bin files on the desktop. Transfer these into the appropriate folder (see below). Then open Excel File ImportDataset(s) Import data from dataset file(s) Next. This will ask for the data file (on desktop) and then subsequently needs confirming of what time frame and what sensors – just always click Next and Finish until the Data appear. The graphs are on sheet 5 and the data set is on sheet 4. 33 Soil Respiration Temperature function: Q10 For the temperature sensitivity of soil respiration we need to combine the Delta-T soil temperature data with the LI-COR soil CO2 flux data. Think about how to do this (e.g. exponential curve fit of flux vs. soil temperature). N.b. For this we need to make sure that the logger and the Li-COR have the same timings! We will use a spreadsheet for this purpose: E:\CTCDSS\Respiration\LICOR\Temperature Soil moisture impacts In order to test for soil moisture impacts on soil respiration we also need to combine the Delta-T soil moisture data with the LI-COR soil CO2 flux data. Think about how to do this (e.g. curve fit of flux vs. soil moisture and possibly also with soil temperature). N.b. For this we need to make sure that the logger and the Li-COR have the same timings! We will use a spreadsheet for this purpose: E:\CTCDSS\Respiration\LICOR\Moisture Scaling-up We certainly want to now how many sample points we need to reduce the measurement uncertainty due to spatial heterogeneity to less than 10%. For this purpose we will use an already prepared spreadsheet (from the vegetation carbon analysis), which will help us with the calculations: E:\CTCDSS\Respiration\LICOR\Spatial. For spatial up scaling we need to have a mean value for each land use type. However, we will not be able to acquire this much information in one week but will get close to a mean value by assuming that our land use is the dominant one for the area (also possibly including % burnt vs. un-burnt flux area). A major problem with up scaling over time is much more difficult as we cannot assume that the temperature function (Q10) will be the same throughout the year! In fact the temperature response changes with substrate supply from the roots to the soil and for peatlands mainly changes in the water table will likely alter temperature responses. To verify any spatial up-scaling methodology, the estimated (scaled-up) soil CO2 flux should be compared with the flux measured by another technique that integrates over a larger scale. For example, when chamber measurements are scaled-up to provide an estimate of the flux over a large spatial scale, the flux should agree with eddy covariance measurements that measure directly at the larger scale. We will also assess the variability related to distance. Prediction of the spatial process at unsampled locations by techniques such as ordinary kriging requires a theoretical semivariogram or covariance. We will use a semivariogram spreadsheet in the same directory. 34 Soil Respiration Caveats - problems and weaknesses in the method Closed systems are inherently problematical because the conditions within the chamber start to change immediately the system is disconnected from the environment. Closed automated systems are also problematic as their cover will partially shade the surrounding area thus possibly causing changes in vegetation around collar or changing root growth under the collar. As seen in equation 1, we rely on the assumption that the CO2 diffusion rate is proportional to the CO2 concentration gradient between the soil surface and the chamber atmosphere. It is important to note that with such a system we have to operate at near ambient CO2 concentrations and consequently closure times have to be as short as possible (e.g. < 5 minutes) as otherwise the increase in chamber CO2 would alter the concentration gradient between soil and chamber air and thus suppress diffusion, resulting in a lower flux. Accordingly, our LI-COR system gives us two options to analyse the CO2 data to obtain a flux, either using a linear (assuming no suppression) or an exponential fit (taking into account possible suppression). We also assume that other factors driving diffusion such as barometric pressure, temperature and moisture of the soil remain constant during the measurement. Whereas the latter two are unlikely to change over such a short time scale there might be pressure changes due to either wind or internal pump speed changes (i.e. in vs. outflow speed). This is why there has to be a major focus on the design of the pressure vent for closed dynamic systems, allowing for equilibration of pressure changes but preventing Venturi Effects. According to the Bernoulli equation, for a free air stream, P + 0.5ρ v2 + ρ gh = constant (2) Static pressure (P) + dynamic pressure (p v2)+ gravitational pressure (p gh) = constant. At constant height, when wind speed increases, dynamic pressure increases as the square of wind speed and static pressure declines an equal amount to keep the sum constant. Thus, still air inside a vent tube will have a higher static pressure than moving air outside a vent tube. Air will flow out of the chamber down this pressure gradient, reducing pressure in the chamber and drawing CO2 out of the soil. Thus, soil CO2 efflux is over-estimated when the open end of a vent tube is exposed to wind. Li-COR designed a new vent (Fig. 2; LI-COR patent pending) based on the Bernoulli equation that reduces the speed of wind as it passes over the vent opening, thus increasing the static pressure of the air at the vent opening, and eliminating the pressure difference between chamber and air. Fig. 1 LI-COR pressure vent preventing Venturi effects. Another issue is that equation 1 ignores dilution effects of water vapour, which interferes with CO2 absorbance. Rh >90% One problem with all chamber flux measurements is the soil collar on which the chamber is placed; a collar is needed to form a good seal between chamber and soil air. Apart from soil disturbance effects which will increase the flux rate it is likely that roots are cut which will also alter any flux rates. Thus the soil collars should be in place several weeks before any 35 Soil Respiration measurements are made. Chamber offset needs to be measured correctly. Please note the different procedure for survey vs. long-term chambers (see above)! Further, for the long-term system we have to allow for a full flush of the air within the long tubing system (2 x 10 m @ chamber) as to prevent CO2 carry-over between lines. We have shown that allowing about 2 minutes observation delay between each measurement will prevent this. The below extract (edited) is taken from an European Science Foundation (LESC Exploratory workshop Investigating the role of soils in the terrestrial carbon balance – harmonising methods for measuring soil CO2 efflux; Edinburgh, 6-8 April 2000) workshop (see Annex 1) provides a valuable summary on most systems, like closed or open dynamic, closed static, chemical CO 2 trapping or micrometeorological systems; all have their strength and weakness. Reading list Nakayama F.S. (1990) Soil Respiration. Remote Sensing, 5, 311-321. Grace J. & Raymont M. (2000) Respiration in the balance. Nature, 404, 819-820. Reichstein M., Rey A., Freibauer A. et al. (2003) Modelling temporal and large-scale spatial variability of soil respiration from soil water availability, temperature and vegetation productivity indices. Global Biogeochemical Cycles, 17 (4) 15.1-15.15. Kirschbaum M.U.F. 2004 Soil respiration under prolonged soil warming: are rate reductions caused by acclimation or substrate loss? Global Change Biology, 10, 1870-1877. Eliasson P.E., McMurtie R.E., Pepper D.A., Strömgren M., Linder S. & Ågren G.I. 2005 The response of heterotrophic CO2 flux to soil warming. Global Change Biology, 11, 167181. Fang C., Smith P., Moncrief J.B. & Smith J.U. 2005 Similar responses of labile and resisatant soil organic matter pools to changes in temperature. Nature, 433, 57-59. Knorr W., Prentice I.C., House J.I. & Holland E.A. 2005 Long-term sensitivity of soil carbon turnover to warming. Nature, 433, 298-301. Hanson P.J., Edwards N.T., Garten C.T. & Andrews J.A. (2000) Separating root and soil microbial contributions to soil respiration: A review of methods and observations. Biogeochemistry, 48, 115-146. Fitter A.H., Graves J.D., Self G.K., Brown D.S. & Taylor K. (1998) Root production, turnover and respiration under two grassland types along an altitudinal gradient: influence of temperature and solar radiation. Oecologia, 114, 20-30. Information list: LI-COR web-side: http://www.licor.com/env/Products/li8100/8100.jsp Equipment list LICOR LI-8100 with 12 long-term chambers LICOR LI-8100 with either of the two survey chambers Palm Tungsten handheld (wireless communication) 4 Delta T data loggers with ~30 thermistor (soil temperature) and 3 theta probes (soil moisture) 50 x (50 cm) PVC tubes for soil water table measurement 40 x 4 inch collars 55 x 8 inch collars Lime 8.0 kg Hammer (for inserting the PVC tubes) GPS (for site locations) Two rulers (for chamber offset determination) 36 Soil Respiration Acronyms NEE Rs Q10 Rh = net ecosystem exchange = soil respiration = increase in Rs over a 10 degree temperature rise = relative humidity 37 Methane Fluxes Methane in the Global Carbon and Greenhouse Gas Budget Instructor (Dave Reay) Introduction Methane concentrations in the atmosphere have more than doubled since the start of the industrial revolution. Wetland and peat bog methane emissions dominate the natural sources of methane. Global emissions from natural sources total around 250 million tonnes each year, with wetland/peat bog areas comprising around 80% of this. Natural emissions of both methane and carbon dioxide can be greatly affected by climate change and increased release of both gases from peat bogs, with increasing global temperatures, is a big potential positive feedback to global warming. Energy related and ruminant methane dominate human-made methane sources. Global human-made methane emissions are estimated to total about 320 million tonnes each year. Tropospheric destruction of methane by hydroxyl (OH) radicals is the dominant sink for atmospheric methane. With stratospheric destruction and oxidation by soil bacteria, the total sink for methane is estimated to be between 500 and 600 million tonnes of methane each year. Principles Methane (CH4) contains carbon and so its flux is part of the overall carbon cycle; there are natural and human-made sources of methane. In addition to this, methane is a powerful greenhouse gas having a global warming potential (GWP) 21 times greater, molecule per molecule, than that of CO2. Our aim is to measure the methane flux from the peatland using a combination of a giant tunnel chamber and a state-of-the-art laser system. The frames for the tunnel chambers are already in place. Once covered with the plastic sheeting and secured around the edges these tunnels become large flux chambers in which the change in methane concentration over time can be measured. The laser system inside the tunnel measures the changing methane concentration by firing an infrared beam down the tunnel and back. Any molecules of methane which get in the way of the beam are therefore measured. We can therefore estimate how much methane the peatland is belching out and its contribution in terms of carbon and, more importantly, its climate forcing effect. To compare our methane emissions with CO2 in terms of climate forcing we will multiply our methane fluxes by 21 to give ‘CO2-equivalents’. 38 Methane Fluxes Practical Operation For detailed use of the Boreal Laser system, refer to the manuals (see Dave) or ask for help. In brief: 1. Position laser on tripod or stable mount at one end of tunnel 2. Position reflector at other end of tunnel (i.e. tape to frame) 3. Turn on the laser 4. Once warmed up it will probably say ‘Low light’ 5. The supervisor will help remedy this. 6. Measure distance between laser and reflector and input data 7. Laser should now be measuring background methane concentration 8. Set laptop to log data 9. Draw plastic cover across chamber and fasten around sides 10. If a laptop is being used, any change in methane concentration can be seen using GasView2. 11. The cover should be left in place for up to 30 mins at a time – keep in mind the need for replication, good linearity and avoidance of superambient heating 12. Maximise the number of incubations, aiming for at least three per site 13. Make a note of the times and filenames used for each logging period 14. Having recorded changes in methane concentration within the chamber(s) the surface area the chambers cover and the total volume of the chamber must be calculated (see below). 15. Now, open the data in Excel (use comma delimiting) and plot the various datasets as ppmm against time to view linearity. 16. To calculate the methane flux for the chambers use only data where a linear increase is apparent, then… the maths Analysis Calculate change in methane concentration in ppm (ppmm divided by m) To get the flux rate divide change in methane by change in time. Multiply this by the volume (v) of the chamber (in m3) divided by the area (A) it covers (in m2). (In case of acute memory loss: the area of a circle is calculated by pi x radius2) Multiply this by the Molecular weight (M) of the gas (16) divided by the universal gas constant (V) (0.0224 m3) Finally, multiply by 273/temperature in Kelvin during measurement (i.e. 273 + temp. in chamber in oC) to correct for ambient temperature As an example: Say change in CH4 ppmm was 100 in 10 minutes So flux was 10ppmm per minute Path length was 10m So flux was 1ppm per minute Say chamber volume = 15m3, area = 18m2, then v/A = 0.83 Multiply 1ppm per minute by 0.83 (i.e. 0.83 micromoles per mole) Then multiply this by 16/0.0224 (M/V). (NB if we wanted µmoles rather than µg CH4 it would be 1/0.0224) 39 Methane Fluxes Finally, given temperature at site of a balmy 20oC, Multiply by 273/293 Gives methane flux of = 552.4 µg CH4 per m2 min-1 = 92 µg CH4 m-2 s-1 = 331.2 mg CH4 m-2 h-1 = 3.312 kg CH4 ha-1 h-1 If we wanted the carbon (C) emission we just need to multiply this by (12/16): = 69 µg C m-2 s-1 The final step is to compare these CH4 fluxes on the basis of CO2-equivalents. As CH4 has a global warming potential of 21 times that of CO2 we can calculate our CO2equivalent flux as 92 µg m-2 s-1 x 21 So for our example: = 1.932 mg CO2-equivalents m-2 s-1 = 6.955g CO2-equivalents m-2 h-1 Greenhouse gas flux from the peatland is = 69.55 kg CO2-equivalents ha-1 h-1 (NB there are 10,000 m2 in a ha) We should have several flux measurements from which to derive a mean flux rate and a standard error. Additionally, if you have data from more than one site you must decide how to extrapolate these data for the peatland as a whole to give the most representative CH4 flux for the area. Caveats and potential problems Spatial variation is our enemy here. The tunnel chambers are large so they negate some of this problem but we must be aware that methane fluxes may be significantly different in other parts of the peatland. Methane fluxes tend to increase as the soils get warmer and wetter, so as a rule of thumb the wetter the soil the more methane we’ll get coming out. The amount of soil carbon, in our case the depth of the peat, is also a key factor – the more carbon then the greater potential for methane emission. Soils not only produce methane but can also oxidise it. As such, when the soils are dry and oxic (large acrotelm) the flux of methane may be into the soil rather than out of it. This can be hard to detect, but keep an eye out for it. As regards the above, there may be little or no change in methane concentrations at some sites – methane oxidation is equalling production Creating a good seal around the chambers can be tricky and care should be taken. If the methane accumulation is clearly non-linear then it is likely you have a leak somewhere. The use of chambers can affect the methane flux through increasing the temperature and reducing air movement. For short incubations this shouldn’t be a problem but it is always worth keeping a record of external wind speeds and external and internal temperatures. The rate of methane flux is likely to change over the course of the day and certainly seasonally. By making several measurements at the same site we can overcome some of this temporal variability, but when citing our ‘Site-wide methane flux’ we need to make clear the assumptions this makes about both spatial and temporal variability. 40 Methane Fluxes Readings and websites IPCC 2001 Climate Change: The Scientific Basis See: http://www.grida.no/climate/ipcc_tar/wg1/134.htm#4211 Conrad R. 1996. Soil microorganisms as controllers of atmospheric trace gases (H-2, CO, CH4, OCS, N2O, and NO). Microbiological Reviews 60:609 (634 pages). Dunfield. P., Knowles, R., Dumont, R., Moore, T.T., 1993. Methane production and consumption in temperate and subarctic peat soils: Response to temperature and pH. Soil Biology & Biochemistry 25, 321-326. GreenHouse Gas Online http://www.ghgonline.org/ Houweling, S., Kaminski, T., Dentener, F., Lelieveld, J., Heimann, M., 1999. Inverse modelling of methane sources and sinks using the adjoint of a global transport model. Journal of Geophysical Research 104, 26137-26160. Panikov, N.S., Dedysh, S.N., Kolesnikov, O.M., Mardini, A.I., Sizova, M.V., 2001. Metabolic and environmental control on methane emission from soil: mechanistic studies of mesotrophic fen in West Siberia. Water, Air and Soil Pollution: Focus 1, 415-428. Reay D.S., McNamara, N. Nedwell, D.B., 2001. Physical determinants of methane oxidation capacity in a temperate soil. Water, Air and Soil Pollution: Focus 1, 404-414. Equipment List Boreal Gas Laser Laser Tripod or stable base Laser power pack Reflectors Measuring tapes and rule Thermometers Laptop with GasView2 and connection cable Calculators Sticky tape 41 Remote Sensing & Earth Observation Techniques Spatial measurement of Leaf Area using Earth Observation Instructor (Phil Lewis) Introduction Providing data on terrestrial ecosystems over large areas is only feasible using Earth Observation (EO) data. However, EO does not generally (except perhaps in the case of LiDAR) make a direct measurement of the properties we are interested in: it measures scattered or emitted radiation at some set of wavelengths. The properties of the ecosystems must therefore be inferred from such measurements. EO data can of course be used for ‘simple’ applications such as land cover classification, where the differing spectral properties of land cover types or the temporal trajectories of these ‘signatures’ are used to discriminate between the cover types of interest and provide a spatial map of such data. Such data are extremely useful in ecosystem modelling, for instance in allocating properties of specific Plant Functional Types (PFTs) over spatial grids to model the state and carbon exchange of the land surface. Perhaps more powerfully still, we can use EO data to infer physical and biochemical properties of ecosystems and provide spatial maps of these to: (i) test model runs; or (ii) reduce uncertainty in model outputs by calibrating model components (e.g. phenology) against EO data or by assimilating EO data into the models. To provide this inference of biophysical properties from EO data, we use various forms of models. The two main approaches are: (i) empirical models relating the biophysical quantities to EO data; or (ii) physically-based models which treat the physics of the radiation scattering for a given set of ecosystem parameters, either to provide an ‘EO operator’ to an assimilation scheme or to provide an estimate of the biophysical quantities through model inversion. The former of these two approaches is much more common in EO applications as it is simpler and faster to apply. The main drawbacks of this method are: (i) the empirical relationship used in the model is only valid over similar conditions to those under which it was originally developed, and so may not apply to different cover types, at different spatial scales, or under different illumination and viewing geometries; (ii) most of these empirical relationships use only a limited subset of the information content of the data; (iii) most of these empirical relationships rely on correlations between biophysical quantities (e.g. leaf area index and biomass) which may not be precise enough for some applications. The use of physically-based models is therefore generally preferable and more powerful, but these generally involve much more stringent requirements on the data, such as the need for well-calibrated measurements and the ability to account for atmospheric scattering and absorption. In addition, the inversion of the typically non-linear models used can be time-consuming and a complex task for multidimensional models (a typical physically-based model requires at least seven parameters). A useful modern text on physicallybased models and their applications is Liang (2003)1. In this exercise, we will concentrate on simpler, practical methods for providing spatial datasets on a particular biophysical quantity (Green Leaf Area Index – gLAI, which is defined as the one-sided leaf area per unit ground area (dimensionless)). gLAI is an important quantity in ecosystem modelling and monitoring as it is related to the ability of the vegetation to intercept shortwave radiation and convert this energy to carbohydrates through photosynthesis. 1 Liang, S., Quantitative Remote Sensing of Land Surface, John Wiley and Sons, Inc., 534 pages, Nov. 2003. 42 Remote Sensing & Earth Observation Techniques Materials and Methods You will be using the following data and equipment to develop an LAI map of the study area: 1. 2. 3. 4. GPS to locate your sample sites; Spectroradiometers to measure surface spectral reflectance at the sample sites; LAI2000 instruments to measure LAI; Optical satellite imagery to provide the spatial mapping of LAI Ground measurement of LAI and radiometry The aim of this is to provide observations of LAI (LAI2000) and surface spectral reflectance (in at least two wavebands: red and near infrared; spectroradiometers) at known ground locations (GPS). The measurements should be averaged over the spatial resolution of the satellite data (15m or 30m) to provide data on a comparable scale to the satellite data. Since we will be using these data to calibrate an empirical relationship, the range of LAI data measured should cover the full range that we are expecting in the landscape. Since the empirical relationship can also depend on cover type, the cover type at each location should be noted, and ideally, LAI measured over the full range of values for each cover type. The detailed operation of the equipment will be explained to you in the field class. Some details on one of the radiometers (ASD Fieldspec Pro) we will use are given on http://fsf.nerc.ac.uk/instruments/asd_fieldspec.shtml. Such equipment measures spectral reflectance in a large number of narrow wavebands across the solar reflective spectrum (usefully, around 400-2400 nm) and costs upwards of £50,000. You will also find details at http://fsf.nerc.ac.uk on how similar equipment can be borrowed for NERC-funded projects. We will also be using ‘cheaper’ more mobile Skye radiometers2 (which cost a few £1000) that measure in only a few (broader) wavebands. Measurement of reflectance is achieved by measuring the spectral radiance of a target area and rationing this by a measurement of incident irradiance. This latter task is achieved either through a radiance measurement on a near-perfect diffusing white panel (e.g. made of spectralon) (ASD) or through a cosine receptor that directly measures the incoming irradiance from the sun and sky (Skye). To achieve measurements comparable to the satellite data, radiometry should be conducted under clear, stable illumination conditions, preferably with the solar zenith angle similar to that of the image acquisition. LAI is generally most accurately measured by destructive sampling and measuring leaf area through some scanning system (this can be done with a flatbed scanner, although specific instruments have been designed along these lines for this task). This is however, rather a slow method to achieve much spatial sampling, so we tend to use indirect estimates of LAI. These methods rely on inferring LAI through some measure of gap probability in the canopy. For an idealised homogeneous canopy, the probability of a gap, Pgap z, in the canopy at depth z at a zenith angle of is: Pgap z, e LAI ( z ) G ( ) cos( ) where: LAI(z) is the downward cumulative LAI; and G() is the so-called ‘G-function’, the areanormalised mean projection of leaves in the direction defined by . For a spherical leaf angle distribution, G()is 0.5, although it will vary as a function of in the more general case. If 2 http://www.skyeinstruments.com/ 43 Remote Sensing & Earth Observation Techniques measurements of gap probability are made at a range of zenith angles, the leaf angle distribution can be inferred along with LAI. Note that the equation defined above is for a homogeneous canopy. Further complications arise for a clumped canopy. What impact might this have on gap probability and inferred LAI? Two main methods are used for measuring gap probability: (i) a radiometric measurement using an instrument such as an LAI2000 from Li-COR3 which measures the incoming radiance at the top and bottom of the canopy in a number of azimuthal and zenith bins to determine gap probability, thence LAI; or (ii) film or high resolution digital cameras with fisheye lenses which can effectively measure gap probability through image classification in either upward-looking (taller canopies, in which one classifies ‘sky’ or ‘vegetation’ to calculate gaps) or downward-looking (for low canopies, in which gaps are found by distinguishing soil and vegetation). The reliability of both methods somewhat depends on the incident illumination conditions, and are best conducted under overcast skies. Note that these non-destructive methods are less accurate for increasing LAI due to the exponential nature of the relationship with gap probability. This effect also means that there is effectively a saturation in the relationship for high LAI values. Pre-processing of the satellite imagery We will be using ‘high’ spatial resolution satellite data to provide the mapping of LAI. Such data are available from instruments such as quickbird4 at around 2.5m resolution, but a relatively high data cost, or from (cheaper) coarser resolution sensors such as ASTER (15m)5, SPOT HRVIR (20m)6, or Landsat TM (30m)7. The data you should use for any particular application will depend on the resolution you require, the date for which you require coverage, the area required, and, not unimportantly, data availability. This latter point is particularly relevant for optical sensors such as those noted above, as we generally require cloud-free scenes for mapping. Typical repeat coverage times of these sensors are a few weeks (this depends on latitude and can be less for pointable sensors such as SPOT HRVIR), but the frequency of cloud-free images is likely to be much less than this. An alternative to using satellite images over relatively small study areas is to obtain data from airborne sensors, such as those made available to NERC projects through the NERC ARSF8. A calibrated satellite image is able to measure the radiance received at a sensor, but this is not an intrinsic property of the surface being observed as it depends on the illumination and atmospheric conditions at the time of imaging. We generally therefore wish to process this to obtain an image of at-surface spectral reflectance. This can be achieved in two main ways: (i) by modelling the impact of atmospheric scattering and absorption on the measured signal and inverting the model to estimate surface reflectance; or (ii) empirical calibration of reflectance by deriving a model of atmospheric influence for each waveband based on ground measured reflectance – the so-called empirical line method. Various codes are available to achieve the former, the most typical of which is 6s9. This method has the requirement that you are able to characterise the atmospheric state at the time of imaging. In the simplest sense, this means that you need information of the atmospheric ozone, water vapour and aerosol type and concentration. These can be measured with ground instruments10, and some of the properties are 3 http://www.glenspectra.co.uk/glen/licor/lai2000.htm http://www.digitalglobe.com/about/quickbird.html 5 http://www.gds.aster.ersdac.or.jp/gds_www2002/index_e.html 6 http://www.spot.com/html/_401_402_.php 7 http://edc.usgs.gov/products/satellite/tm.html 8 http://www.nerc.ac.uk/arsf/Home.htm 9 http://www-loa.univ-lille1.fr/SOFTWARE/Msixs/ 10 http://aeronet.gsfc.nasa.gov/ 4 44 Remote Sensing & Earth Observation Techniques available from other, e.g. satellite sources (MODIS aerosol properties11; MODIS water vapour12, TOMS ozone13), though these are only available for the locations these sensors happened to be viewing at any particular time and are quite coarse resolution. The empirical line method requires nominally contemporaneous measurements of ground reflectance, ideally over bright and dark targets. A linear relationship is then calibrated between the ground measured reflectance and the satellite measurement and applied to the whole image. This method is usually reasonable, provided such measurements are made at the same time as imaging, the atmospheric conditions are near constant across the scene, and topographic variation is not too large (otherwise the optical thickness varies across the scene). The assumed linear relationship is reasonable, except for very bright targets or relatively high aerosol loadings (in which case multiple scattering between the ground and atmosphere results in a non-linear relationship). For this exercise, we will not have contemporaneous ground reflectance or atmospheric measurements, so we will have to perform an empirical line correction based on the ground reflectance measurements you make. Note that it is important that you measure reflectance over as large a range of targets (bright, dark) as possible. Since the images are not exactly the same time as your ground surveys, there will be some uncertainty in the mapping and you may have to remove outliers in the relationship should the ground properties have changes significantly. Note also that since the bandpass functions of the sensors will be different to those of the ground instruments, this will introduce additional uncertainty. Calibration of the relationship between radiometric measurements and LAI Healthy green vegetation tends to strongly absorb incident radiation at red wavelengths (around 650nm) for use in photosynthesis. Near infrared radiation (around 750 nm) on the other hand is mostly not absorbed. This is a distinctive feature of vegetation and allows it to be discriminated from other cover types. Thus, satellite measurements of vegetation appear bright in the near infrared and dark in the red. It is found that a combination of information in these bands is therefore strongly related to vegetation amount. We typically term such a relationship a vegetation index. If we plot measured reflectance in red and near infrared as a scatter plot (below), we can note that increasing vegetation amount is generally given in the direction indicated. The most commonly-used such index is the NDVI, the Normalised Difference Vegetation Index, which is a measure in this ‘direction’. It is defined as: NDVI nir red nir red where nir ( red ) is the reflectance in the near infrared (red) waveband. It is a normalised measure lying between -1 and 1. Positive values are related to vegetation amount, although they can also be influenced by various extraneous factors such as soil brightness variations, topographic effects and variations in viewing and illumination conditions. 11 http://modis-atmos.gsfc.nasa.gov/MOD04_L2/ http://modis-atmos.gsfc.nasa.gov/MOD05_L2/ 13 http://toms.gsfc.nasa.gov/ozone/ozone.html 12 45 Remote Sensing & Earth Observation Techniques Line of increasing vegetation amount Near-infrared-red scatterplot Once you have measured a set of reflectance data, examine a scatterplot in red-near infrared feature space as above to confirm the behaviour of the observations. You can then simply calculate NDVI in a spreadsheet and plot this as a function of LAI. Note the form of this relationship and fit an appropriate function to it. Is the form of relationship what you expected? Why might it be of this particular form? What implications does this have (if any) for uncertainty as a function of LAI? If you have sufficient samples for this relationship, you should use some of them to assess the uncertainty in the derived LAI data. If the sample size is small, you might consider using a jackknifing method (step through the dataset, removing one sample at a time and re-fitting the relationship to the remaining points). Alternatively, you might consider using data collected by other groups in assessing this (if available). As with all datasets during this summer school you can join and link datasets acquired during the entire week! Mapping of LAI Once you have the relationship between NDVI and LAI defined from above (along with an assessment of uncertainty), you can apply this relationship to NDVI calculated from the satellite data. Ideally, you should have a further set of independent measurements of LAI to further test your relationship. Practical Aspects Sampling There are various aspects to consider in defining your sampling strategy. First of all, you have to be able to relate your sample sites to the satellite imagery. This will be achieved by measuring the location of the sites with GPS. Second, you need to be able to match the resolution of your satellite data to the area you will sample on the ground (e.g. 15 m x 15 m for ASTER or 30 m x 30 m for Landsat TM). You then need to think about the sampling pattern you will use within this area. The geolocation of the satellite imagery is likely to be around 1 pixel (i.e. 15 m or 30 46 Remote Sensing & Earth Observation Techniques m). What impact might this have on your results? Third, you will need to consider sampling density. Since you require a ‘reasonable’ number of samples to define your empirical relationship and you have only limited time in this exercise, there will be a trade-off between the sampling intensity at each site and the number of sites. Remember that the sample sites you chose should cover the range of variability of the quantity we are trying to map. How might you use the satellite data to help you decide on this? Figure 1. Example Landsat 5 TM imagery over the study area, the image was acquired on the 14 May 1988. The top image is a false colour composite, where vegetated areas appear in shades of red, while the bottom image is of the NDVI computed from the same 1988 image. 47 Remote Sensing & Earth Observation Techniques Finally, remember also that the ‘optimal’ conditions for doing radiometric measurements are rather different from those required to use the LAI2000 and that this might impact you results (ideally, you would have a longer period of time in which to perform your measurements). Figure 2. Example Landsat 7 ETM+ imagery over the study area, the image was acquired on the 7 May 2000. The top image is a false colour composite, where vegetated areas appear in shades of red, while the bottom image is of the NDVI computed from the same 2000 image. 48 Remote Sensing & Earth Observation Techniques References Baret, F., and Guyot, G., 1991, Potentials and limits of vegetation indices for LAI and apar assessment, Remote Sensing of Environment, 35, 161-173. Chen, J.M., Liu, J., Leblanc, S.G., Lacaze, R., Roujean, J.L., 2003, Multi-angular optical remote sensing for assessing vegetation structure and carbon absorption, Remote Sensing of Environment, 84, 516-525. Duncan, J., Stow, D., Franklin, J., and Hope, A., 1993, Assessing the relationship between spectral vegetation indices and shrub cover in the Jornada Basin, New Mexico, International Journal of Remote Sensing, 14, 3395-3416. Fang, H., and Liang, S., 2005, A hybrid inversion method for mapping leaf area index from MODIS data: experiments and application to broadleaf and needleleaf canopies, Remote Sensing of Environment, 94, 405-424. Gower, S.T., Kucharik, C.J., and Norman, J.M., 1999, Direct and indirect estimation of leaf area index, FAPAR, and net primary production of terrestrial ecosystems, Remote Sensing of Environment, 70, 29.51 Huete, A.R., and Jackson, 1987, Suitability of spectral indices for evaluating vegetation characteristics on arid rangelands, Remote Sensing of Environment, 23, 213-232. Jonckheere, I., Fleck, S., Nackaerts, K., Muys, B., Coppin, P., Weiss, M., and Baret, F., 2004, Review of methods for in situ leaf area indexdetermination. Part 1. Theories, sensors, and hemispherical photography, Agricultural and Forest Meteorology, 121, 19-35. Liang, S., Fang, H., and Chen, M., 2001, Atmospheric Correction of Landsat ETM+ Land Surface Imagery-Part I: Methods, IEEE Transactions on Geoscience and Remote Sensing, 39, 11, 2490-2498. Leprieur, C., Kerr, Y.H., and Pichon, J.M., 1996, Critical assessment of vegetation indices from AVHRR in a semi-arid environment, International Journal of Remote Sensing, 17, 13, 25492563. Myneni, R.B., Maggion, S., Iaquinta, J., Privette, J.L., Gobron, N., Pinty, B., Kimes, D.S., Verstraete, M.M., and Williams, D.L., 1995, Optical remote sensing of vegetation: modeling, caveats, and algorithms, Remote Sensing of Environment, 51, 169-188. Myneni, R.B., Hoffmann, S., Knyazikhin, Y., Privette, J.L., Glassy, J., Tian, Y., Wang, Y., Song, X., Zhang, Y., Smith, G.R., Lotsch, A., Friedl, M., Morisette, J.T., Votava, P., Nemani, R.R., and Running, S.W., 2002, Global products of vegetation leaf area and fraction absorbed PAR from year one of MODIS data, Remote Sensing of Environment, 83, 214-231. Running, S.W., and Coughlan, J.C., 1988, A general model of forest ecosystem processes for regional applications I. Hydrologic balance, canopy gas exchange and primary production processes, Ecological Modelling, 42, 125-154. 49 Remote Sensing & Earth Observation Techniques Running, S.W., Nemani, R.R., Peterson, D.L., Band, L.E., Potts, D.F., Pierce, L.L., and Spanner, M.A., 1989, Mapping regional forest evapotranspiration and photosynthesis by coupling satellite data with ecosystem simulation, Ecology, 70 (4), 1090-1101. Sellers, P.J., Dickinson, R.E., Randall, D.A., Betts, A.K., Hall, F.G., Berry, J.A., Collatz, G.J., Denning, A.S., Mooney, H.A., Nobre, C.A., Sato, N., Field, C.B., and Henderson-Sellers, A., 1997, Modelling the exchanges of energy, water, and carbon between continents and the atmosphere, Science, 275, 502-509. Verstraete, M. M., and Pinty, B., 1996, Designing optimal spectral indices for remote sensing applications, IEEE Transactions on Geoscience and Remote Sensing, 34, 1254-1265. Verstraete, M.M., Pinty, B., and Myneni, R.B., 1996, Potential and limitations of information extraction on the terrestrial biosphere from satellite remote sensing, Remote sensing of environment, 58, 201-214. Walthall, C., Dulaney, W., Anderson, M., Norman, J., Fang, H., and Liang, S., 2004, A comparison of empirical and neural network approaches for estimating corn and soybean leaf area index from Landsat ETM+ imagery, Remote Sensing of Environment, 92, 465-474. Weiss, M., Baret, F., Myneni, R.B., Pragnere, A., and Knyazikhin, Y., 2000, Investigation of a model inversion technique to estimate canopy biophysical variables from spectral and directional reflectance data, Agronomie, 20, 3-22. Weiss, M., Baret, F., Smith, G.F., Jonckheere, I., and Coppin, J., 2004, Review of methods for in situ leaf area index (LAI) determination, Part II. Estimation of LAI, errors, and sampling, Agricultural and Forest Meteorology, 121, 37-53. 50 Ground Penetrating Radar and Peat Depth Introducing Ground-Penetrating Radar (Instructor: Steve Dobson and John Jones) Ground-Penetrating Radar (GPR) provides a means of inspecting and mapping sub-surface geology, archaeology and modern buried features such as structures or services. High-frequency electromagnetic radio (radar) is transmitted into the ground and the time of transmission, reflection and return to the antenna is calculated by the instrument. The radar signal typically is reflected by minute changes in the electrical properties of soil sediment, water content, geological nature, density or stratigraphic change. The antenna then receives the reflected waves and stores them in the digital control unit. The ground penetrating radar antenna (transducer) is pulled along the ground by hand or behind a vehicle. Electromagnetic waves travel at a specific velocity that is determined primarily by the electrical permittivity of the material. The velocity is different between materials with different electrical properties, and a signal passed through two materials with different permittivity over the same distance will arrive at different times. The interval of time that it takes for the wave to travel from the transmit antenna to the receive antenna is simply called the transit time. The basic unit of electromagnetic wave travel time is the nanosecond (ns), where 1 ns=10-9 s. Since the velocity of an electromagnetic wave in air is 0.3 m/ns, then the travel time for an electromagnetic wave in air is approximately 3.3333 ns per m travelled. The velocity is proportional to the inverse square root of the permittivity of the material, and since the permittivity of earth materials is always greater than the permittivity of the air, the travel time of a wave in a material other than air is always greater than 3.3333 ns/m. The penetration depth of GPR is based on many factors - one of the most influential of these is the frequency of the antenna. Typically antenna frequencies range between 10-1000 MHz with lower frequencies achieving much greater depths, however due to the longer wavelength of these frequencies, vertical resolution is much lower. Another major factor is the nature of the materials that the radar wave passes through. Ground penetrating radar waves can reach depths up to 100 feet (30 meters) in low conductivity 51 Ground Penetrating Radar and Peat Depth materials such as dry sand or granite. Clays, shale, and other high conductivity materials, may attenuate or absorb GPR signals, greatly decreasing the depth of penetration to 3 feet (1 meter) or less. In this case the electromagnetic energy of the signal needs to be able to pass through a material without being absorbed or dissipated – this attribute is described as the Relative Dielectric Permittivity of a material and outlines the velocity of radar waves through the material. Examples of high RDP materials: Water Saturated Silt Clay 80-88 RDP 10-40 RDP 5-40 RDP Examples of low RDP materials: Air Dry Sand Concrete 1 RDP 3-5 RDP 6 RDP Consequently, large reflections will be observed at the interface between thick layers of very different RDP, conversely gradual change or small changes ever few centimetres will generate very small responses. Other factors that affect the ‘visibility’ of subsurface features or geological interfaces are the effects of focussing/dispersion and the alignment of the feature or interface to the signal. Focusing or dispersion occurs when transmitted energy passes through a series of materials with increasing (focusing) or decreasing (dispersion) RDP. FOCUSING RDP 1 RDP 5 RDP 50 52 Ground Penetrating Radar and Peat Depth DISPERSION RDP 50 RDP 5 RDP 1 The alignment of a surface or interface will also play a large part in its detection since if it is steeply sloping transmitted waves will be deflected away from the antenna. Since the return signal in this case will be absent (or greatly weakened) the feature will be ‘invisible’. DEFLECTION 53 Ground Penetrating Radar and Peat Depth GPR EXAMPLE PLOT: This example plot shows a number of very clear return signals from localised buried features. The GPR approach has been used in the past to determine peat depth (have a look at : http://nesoil.com/gpr/gprgis.htm) and we will try to use this technique at our field site and compare estimates to real peat core measurements. Remember, a problem might always be the high water content (i.e. high RDP) of the peat. Establishing depth scales: As mentioned earlier, the vertical scale of the radar profile is a time scale which shows the time it takes for the signal to travel through the subsurface and return to the antenna. This time scale can be converted to a depth scale if the signal propagation velocity is known. Propagation velocities can be either calculated or estimated if ground truthing is not performed. Calculating the propagation velocity is usually performed by burying a reflector at a known depth and determining the velocity using the following formula: Vm = 2D/t Where: D T Vm = measured depth to reflecting interface. = elapsed time between transmitted and received pulse (nanoseconds). = effective propagation velocity (feet/nanosecond) If propagation velocities are known, depths can be estimated using the following formula: Depth = Vm(t)/2 Propagation velocities are often estimated using standard published values for various materials (see RDP table above). Anyhow, the software will do this for us! 54 Ground Penetrating Radar and Peat Depth 55 Annex 1 Annex 1 - Two-tailed t-table t-tables n p=0.05 p=0.01 p=0.001 n p=0.05 p=0.01 p=0.001 1 12.706 63.656 636.578 22 2.074 2.819 3.792 2 4.303 9.925 31.6 23 2.069 2.807 3.768 3 3.182 5.841 12.924 24 2.064 2.797 3.745 4 2.776 4.604 8.61 25 2.06 2.787 3.725 5 2.571 4.032 6.869 26 2.056 2.779 3.707 6 2.447 3.707 5.959 27 2.052 2.771 3.689 7 2.365 3.499 5.408 28 2.048 2.763 3.674 8 2.306 3.355 5.041 29 2.045 2.756 3.66 9 2.262 3.25 4.781 30 2.042 2.75 3.646 10 2.228 3.169 4.587 32 2.037 2.738 3.622 11 2.201 3.106 4.437 34 2.032 2.728 3.601 12 2.179 3.055 4.318 36 2.028 2.719 3.582 13 2.16 3.012 4.221 38 2.024 2.712 3.566 14 2.145 2.977 4.14 40 2.021 2.704 3.551 15 2.131 2.947 4.073 45 2.014 2.69 3.52 16 2.12 2.921 4.015 50 2.009 2.678 3.496 17 2.11 2.898 3.965 75 1.992 2.643 3.425 18 2.101 2.878 3.922 100 1.984 2.626 3.39 19 2.093 2.861 3.883 20 2.086 2.845 3.85 1.96 2.576 3.29 21 2.08 2.831 3.819 56 ANNEX 2 Annex 2 – Soil Respiration system inter-comparison: LESC Exploratory workshop Investigating the role of soils in the terrestrial carbon balance – harmonising methods for measuring soil CO2 efflux; Edinburgh, 6-8 April 2000) workshop Closed static systems Three types of closed static systems are distinguished: 1. Chemical systems, using NaOH, KOH or soda lime to precipitate or adsorb CO2, 2. Physical systems, where CO2 is sampled from the headspace of the chamber, 3. Automated physical systems, where the closure and headspace sampling is automated. Chemical systems Limitations (-) and Advantages (+) Accuracy When the chemicals become saturated or when CO2 concentrations within the chamber rise much above ambient levels, large underestimation errors can occur. Strong CO2 uptake by the chemicals can lead to a change in the naturally occurring diffusion gradient. Long closure times and edge effects, especially in small, shallow chambers, can lead to increased errors. Long exposure times might cause heating of the chamber, affecting rate constants for chemical reactions and root or microbial activities. Resolution + Method offers potential to integrate over time, e.g. at night. + Lots of replicates can be used. Thus, good method for inhomogeneous sites. Fast measurements are impossible because exposure times are long. Convenience + Method is cheap. + Laboratory preparation is fast, about 5 - 10 minutes per sample. + Easy experimental control in the field (non-emitting surface, same set-up as "real" chambers). + Off site analysis of sample. Other There might be problems for subsequent isotope analyses. Bad prestige among scientists. Minimum (•) and recommended (*) technical specifications • Use NaOH, KOH or sodalime. • Solution in tray, 7 cm above the soil surface, with tray covering ~1/3 soil area. • Chemical must not be saturated more than 10 - 50 % at the end of the measurement. • Chambers up to 5 l in volume, 20 cm in diameter, 15 - 20 cm height. • Exposure time between 8 and 10 hours. • Liquids are better than granules, granule size should be as small as possible. • The bigger the area of chemical, the better the representation of the actual flux. Summary Useful for highly diverse sites where a large number of measurements is needed to capture spatial variation. Method should be cross-compared to other measurements at individual sites. Could be the method of choice around a small number of automated chambers. Physical systems and automated physical systems Accuracy 57 ANNEX 2 + High accuracy. If non-linear increase of headspace CO2 concentration is not accounted for by multiple sampling and non-linear regression, the flux will be underestimated. Resolution + Chambers can be made very large, minimising the problems associated with small scale spatial variability. Extensive temporal sampling (e.g. diurnal cycle) is labour intensive. Convenience + Easy to learn and do. + Bottles can be stored for weeks or months. + Off site analysis of sample. Requires a GC in the lab. Other + Parallel analyses of other trace gases are possible. + Parallel analyses of isotopic composition are possible. Minimum (•) and recommended (*) technical specifications • Chamber at least 20 cm height. • Sample stored in 20 ml gas-tight glass vial. • At least two headspace samples per measurement. • Three or more headspace samples per measurement gives increased confidence in extrapolation of flux to intercept (at t = 0.) Summary If a GC is available this method is very powerful, particularly if the research is focussed on a limited number of sites or if fluxes of gases other that CO2 are being measured. Closed dynamic chambers Two types of closed dynamic systems are distinguished: 1. Manual system where a chamber is placed in position and removed after measurement. 2. Automated system where a chamber remains in place for a long time (open & closed). Limitations (-) and Advantages (+) Accuracy + Because no differential analyser is used, IRGA calibration less important. If a fan is used for mixing, the speed of the fan can alter flux rates (pressure changes). The calculated efflux rate depends on the chamber volume and soil air volume (?!). As not in the steady-state chamber effects might cause measurement artefacts (wind!). Resolution + Measurement can be taken in a very short time, spatial sampling can be extensive. Extensive temporal sampling (e.g. diurnal cycle) is labour intensive. Convenience + Commercial systems are available and easy to use. Other + Commercial systems could make this method a “standard” methodology. Minimum (•) and recommended (*) technical specifications • The chamber should be fitted with a pressure vent. The tube should be long and thin (possibly at ground level) to minimise diffusive loss from the chamber. • A correction should be made for the volume of air in the soil below the chamber. • Ensure good mixing inside the chamber. • If using a mixing fan, the effect of fan speed should be investigated. • Chamber should be opaque as a 1°C increase in temperature might lead to ~ +300 Pa. • The linearity of the increase in concentration over time should be investigated. • At least three readings at each measurement spot. 58 ANNEX 2 * Minimise the ‘trampling’ around the chamber (causing mass flow of air into chamber). Summary Simplicity of use and low price make this system an attractive choice. Open dynamic systems Again, two types of open dynamic systems are distinguished: 1. Manual system where a chamber is placed in position and removed after measurement. 2. Automated system where a chamber remains in place for a long time (open & closing). Accuracy + High, if artefacts removed. + All necessary variables can be measured accurately. + Steady state measurement. Depends on clever design to overcome pressure effects. Prone to condensation (when running continuously). Resolution + Allows continuous measurements. + High temporal resolution, once chamber is in steady-state. Chamber can take up to 10 minutes to reach steady-state. Small number of chambers (when running automatically). Convenience Needs more electrical power than closed chambers (pumps). Other Requires costly differential gas analyser and mass flow controller. Minimum (•) and recommended (*) technical specifications • Only steady-state measurements should be reported. • The pressure difference should be less than 0.3 Pa (preferably less than 0.1 Pa). • The volume of the analysis and reference sample lines should be the same. * Care should be taken to prevent condensation in the sample tubes. * Smaller chamber volume gives a faster response time. Summary If chamber pressurisation artefacts can be eliminated, the steady-state nature of the measurement means that this system potentially gives the most meaningful chamber data. Eddy covariance Accuracy + Does not disturb system being measured, especially important in terms of • Local pressure fluctuations and natural transport processes, • CO2 concentration, • Temperature, • Soil moisture. + Probably ± 25% most of the time. Difficult to test accuracy of method against standard. Many uncertainties (frequency response, mean flow direction, non-turbulent motions). Resolution + Spatially integrated measurement, scale 10 - 100m. + Continuous measurement. + Potential for identifying turbulence effects such as pressure pumping. Low spatial resolution. Measures bulk properties, i.e. cannot separate soil from ground vegetation. Variable sample area (footprint). Trade off between time averaging and representativeness. 59 ANNEX 2 Convenience Complex, not plug and play. Understanding the data requires more measurements/knowledge. High power requirements. Other Expensive. Limited applicability on slopes or under dense canopies. Minimum technical specifications • Standard protocol of additional measurements/corrections important. • Deployment protocol is site dependent. • Special attention must be paid to frequency related aspects. Summary This method gives the most accurate representation of the naturally occurring soil CO2 flux, but this might not be the same as soil respiration because it includes understorey assimilation etc. Collecting, correcting and interpreting the data is demanding and can be problematic. General considerations - summary All important variables must be measured accurately, (i) closed chambers: volume, concentration, time, chamber area; (ii) open chambers: flow rate, concentration differential, chamber area. There is a general trade off between sampling and resolution, a few locations can be measured continuously (with an automatic system) or many locations can be measured less frequently (with a manual system). To ensure no alteration of the soil environment (temperature, moisture, litter input) chambers should not be left in place for more than 24 hours unless they are of the automatically opening type. Closed static chambers should cover at least 400cm2 of soil. For other systems chamber size should be chosen according to requirement. Small chambers can be placed between plants or in different micro topographical positions. Larger chambers can integrate over larger areas and might give a more representative measurement. For open systems, very large chambers (> 1m2) require large air flows and are power intensive. Collars should be placed at least 24 hours prior to measurement to avoid measuring any initial flush of CO2. Collars have two roles, to seal the chamber/soil boundary, and to provide anchorage for the chamber. The depth of insertion into the soil will therefore vary from site to site. Factors that should be considered are, • Soil porosity – more porous soils means deeper insertion • Depth of the root mat • Depth of the water table When measurements are to be made only once, severing roots with the collar will not cause unreliable data. When a collar is inserted and left for a long period, severing the root mat will alter the carbon dynamics of the soil under the chamber and should be avoided. Where this is unavoidable, the collar should be placed in the winter before measurements are made, and should have holes (Swiss cheese collar) through which new roots can grow. An alternative is to have a shallow collar with anchoring legs that are inserted deeper into the soil. 60 Annex 3 Annex 3 - Field Site Soil Respiration Coordinates: Flux Tower Site 1 UK grid: Easting Northing GPS: SD 83535 RNG 71749 383535 471749 Collar locations (always 3 x 8 inch collar each with a 4 inch collar inside): SD RNG Plot1 83549 71751 Plot2 83537 71741 Plot3 83517 71741 Plot4 83530 71732 Plot5 83524 71722 Plot6 83544 71754 FT 83535 71749 Flux Tower Site 2 (incl. burnt vs. un-burnt area) UK grid: Easting Northing GPS: SD 83575 RNG 71588 383575 471588 Lime treatment UK grid: Easting Northing GPS: SD 83711 RNG 71808 383711 471808 Semivariogram collars UK grid: Easting Northing GPS: SD 83519 RNG 7172 383519 471721 61 Annex 4 Annex 4 - List of Participants Record First Name Surname Organisation Type Email Address Tel: 1 Ilaria Inglima Second University of Naples phd ilaria.inglima @unina2.it Department of Environmental Sciences 390823274648 Via Vivaldi, 43 Caserta 81100 2 Daniel Metcalfe University of Edinburgh phd d.b.metcalfe@s ms.ed.ac.uk Flat 2F1, Warrender Park Terrace +44 (0) 7745 492 655 Edinburgh EH9 1EE 3 Tim Jupp CEH Monks Wood pod tju@ceh.ac.uk Abbots Ripton +44 (0)7941 906814 Huntingdon PE28 2LS 4 Sabine Göttlicher Swedish University of Agricultural Sciences phd sabine.goettlic her@sek.slu.se Department of Forest Ecology 0046 (0) 90-786 8361 Petrus Laestadius väg SE - 901 83 Umeå 5 Luiz Eduardo Aragao University of Oxford Pod laragao@ouce. ox.ac.uk 62 School of Geography and the Environment +44 (0)1865 281980 Annex 4 University of Oxford Mansfield Road OX1 3TB 6 Chris Miller University Of Aberdeen phd c.j.miller@abd n.ac.uk 29 Hilton Road 7974945642 Aberdeen AB24 4HR 7 Iain Hartley University of York phd iph106@york.a c.uk Area 2, Biology Department 01904 328594 Heslington, York , YO10 5YW 8 Timothy Hill CTCD, University of Edinburgh phd t.c.hill@sms.ed .ac.uk The University of Edinburgh 7793058251 Crew Building The Kings Buildings EH9 3JN 9 Suresh Kumar Jogi Department of Botany phd jogis@tcd.ie Trinity College 00353-1-6081051 Dublin-2 10 Katherine Chadwick University of Aberdeen phd k.chadwick@a bdn.ac.uk Plant & Soil Science Department 07832 192962 St Machar Drive , Aberdeen, AB24 3UU 11 Luke Spadavecchia Univerity of Edinburgh (CTCD) phd s0198247@sms .ed.ac.uk 63 Flat 2F1 07838 116632 Annex 4 16 Gillespie Crescent Edinburgh, EH10 4HT 12 Ana Prieto University of Wales Swansea phd 304376@swan. ac.uk 13 Clifton Hill +44 7800757671 SA1 6XQ 13 Rui Zhang University of Edinburgh phd R.Zhang2@sms.ed.ac.u k Crew building (Attic) (44)131 6506480 School of Geosciences, Edinburgh University , EH9 3JN 14 Angela Gallego-Sala University of Bristol phd Angela.Gallego Sala@bristol.a c.uk Wills Memorial Building + 44 (0) 117 954 9500 Queens Road, Bristol, BS7 8TP 15 Casey Ryan University of Edinburgh phd me@caseyryan .co.uk 44 Lordship Rd 2088021641; 01603-593516 London, N16 0QT 16 Claudia Di Bene Scuola Superiore Sant-Anna Pisa phd claudine@sssu p.it P.zza Martiri della Libertà 33, 56127 Pisa 393334562263 17 Louise Grøndahl National Environmen tal Research Institute phd lgr@dmu.dk Frederiksborgvej 399 P.O.Box 358, +45 46 30 19 31 DK-4000 Roskilde 64 Annex 4 18 Hildegard Meyer University of Vienna phd loiseleuria@ya hoo.de Dep. of Chemical Ecology and Ecosystem Research 0043 1 42 77 54 251 Althanstrasse 14 Vienna, A-1090 19 Gustavo Saiz Univ.College Dublin, Dep. Env. Resource Manag. phd gustavo.saiz@u cd.ie Faculty of Agri-Food and the Environment, +353 87 925 7794 Dublin 4 20 Diana Hernandez CSIC, Center for Environmen tales Science phd dhernandez@c cma.csic.es C/ Serrano 115 bis. Madrid, 28006 21 Albin Hammerle Universtiy of Innsbruck phd csac3404@uib k.ac.at Tiergartenstr. 25b/3/31 +34 91 411 5301 A-6020 Innsbruck 22 Michael Schmitt University of Innsbruck phd schmitt.michel @gmx.de Fischnalerstrasse 12 43512937953 A-6020-Innsbruck 23 NOEL ROBERTSON UNIVERSIT Y OF SHEFFIELD phd n.a.robertson @sheffield.ac.u k HICKS BUILDING 0114 2223749 HOUNSFIELD ROAD SHEFFIELD S3 7RH 24 Jessica Lopez University of Helsinki phd jessica.lopezbe llido@helsinki. fi Niemenkatu 73, 15140 Lahti - Finland Pääjarventie 320, 16900 Lammi-Finland 65 358-3-89220308 Annex 4 25 Torbjörn Johansson Lund University phd torbjorn.johan sson@nateko.l u.se GeoBiosphere Science Centre (CGB) +46 (0)46 222 39 74 Physical Geography and Ecosystems Analysis (INES) Sölvegatan 12, SE 223 62 Lund 26 Inge Vande Walle Ghent University phd Inge.VandeWal le@UGent.be Coupure links 653, 9000 Gent 9000 +32 9 264 61 26 27 Teresa Bertolini C.N.R.I.S.A.Fo.M. pod t.bertolini@isp aim.na.cnr.it Via Patacca, 85 0039 081 5746606 28 Helen Bowes University of Bristol phd h.l.bowes@bris tol.ac.uk Wills Memorial Building 0117 3315037 Queens Road Bristol BS8 1RJ 29 Helen O-Brien Nottingham Trent University phd helen.obrien@ ntu.ac.uk School of ARES 44 01636 817000 Brackenhurst Southwell Nottinghamshire NG25 0QF 30 Peter Eliasson Swedish University of Agric. Sciences phd peter.eliasson @eom.slu.se Dept. for Ecology & Environmental Research SE 750 07 +46 (90) 672458 phd a.wakefield@la ncaster.ac.uk IENS 07816166702 (mobile) (SLU) 31 Angela Wakefield Lancaster University/C EH Lancaster Lancaster University Lancaster LA1 4YQ 66 Annex 4 32 Martin De Kauwe Edinburgh University Bruce Thomson Lancaster University/C EH Lancaster 33 34 Kevin Jones University of Edinburgh phd mdekauwe@g mail.com brth@ceh.ac.u k Hydrology, Oxford phd phd Centre for Ecology and s0197746@sms .ed.ac.uk The Univeristy of Edinburgh Crew Building The Kings Buildings EH9 3JN 67 +441865281684