 |
Carbon Assimilation and Modelling of the European Land Surfaces |
 |
 |
|
Third CAMELS consortium Meeting held at ALTERRA, Wageningen, Netherlands, between 11th November and 12th November 2003 Present: Met Office: Peter Cox, Richard Betts, Venkata R Jogireddy LSCE: Roger Dargaville, Diego Santeren MPI-BGC: Martin Heimann, Jens Kattge, Kristina Trusilova ALTERRA: Ronald Hutjes, Isabel van der Wyngaert, Mart-Jan Schelhaas, Bart Kruijt UNITUS: Riccardo Valentini, Dario Papale, Markus Reichstein EFI: Ari Pussinen CEH: Chris Huntingford Apologies: Wolfgang Knorr (MPI-BGC), Pierre Friedlingstein, Philippe Peylin, Andrew Friend, Philippe Ciais (LSCE), Richard Harding (CEH), Thomas Kaminski (FastOpt), Michel Verstraete, Nadine Gobron, Bernard Pinty (JRC), Claus Bruning (EC) Date of next meeting: Met Office, Exeter, UK, during June 2004
IvdW described progress on the preparation of site-level data for use in WP2. 20 sites have now been prepared with timeseries ranging from 1 to 4 years, and covering a range of plant functional types. Data is available at a temporal resolution of 30 minutes, with continuous (gap filled) driving data, but no gap-filling of fluxes. Formats are as follows: i. Flux and met data in netCDF ii. Site-information in ASCII iii. Soil and vegetation data in Excel or ASCII. There was some discussion of the relationship between the 20 sites currently available and the original sites of Nicolas Viovy. DP also described the TACOS dataset (http://gaia.agaria.unitus.it/database) which covers 39 sites. There was agreement that a mega-table of sites listing those in each subset (TACOS, CAMELS, Viovy) should be prepared for linking to the CAMELS website. ACTION 3(a): ALTERRA to circulate a description of the available flux sites for linking to the CAMELS website, including distinction between the various flux site lists. b. Atmospheric CO2 dataset for use in nowcasting system (LSCE) Nobody at the meeting was able to report on this, but RD promised to ask for an update on this deliverable from Philippe Ciais (see action 3(n)). c. Land-use and nitrogen deposition historical datasets 1900-2000 (Hadley) RB described his work on merging together the crop dataset of Ramankutty and Foley (1999) and the pasture data processed from the HYDE dataset of Klein Goldewijk (2001), to produce a dataset of land-use change from 1700-1990 (this dataset will also be used by C4MIP). There was some discussion of the need to interpolate this dataset to 2000 (for use in WP3 and WP4). Action 2(a) is ongoing. No nitrogen deposition datasets have yet been acquired, partly because none of the current CAMELS TEMs can utilise these (as they do not include interactive N cycling). However, there are plans to include N effects in MOSES. KT described the N dep datasets used with BIOME-BGC and offered to provide details on these. PC took an action to check whether N dep datasets are available from IGAC. ACTION 3(b): KT to provide link to N dep datasets (two snapshots). ACTION 3(c): Hadley (PC) to decide best way forward on N deposition datasets. d. Datasets of recent change in European land carbon (EFI) M-JS presented results from EFI on recent changes in European forest carbon. He pointed out that these changes are dominated by changes in land management (e.g. extensive afforestation in Denmark 1881-1990). Even apparently large natural disturbance events result in little impact on the overall carbon budget due to responses by the forestry sector. There was a discussion of how this information could be used within the CAMELS TEMs of WP3. AP explained that harvesting was generally a small part of the forest carbon budget (~5% of litterfall) and could therefore be neglected in the first instance. However, a proper treatment of the impacts of land-management changes certainly requires inclusion of age-class effects in the TEMs. In this instance the most useful inventory information would be regeneration rate (as a function of year), and harvesting rate (as a function of age-class and year). RB suggested that his land-use dataset should be checked for consistency with the EFI inventory data. ACTION 3(d): EFI to provide link to inventory data for the historical period. ACTION 3(e): Hadley(RB) to check for consistency between inventory data and land-use dataset. ACTION 3(f): EFI to report on availability of regeneration rate (as a function of year), and harvesting rate (as a function of age-class and year). e. Dataset of fAPAR for Europe, and other remote sensing products (JRC) PC gave a JRC presentation prepared by NG (without a French accent - as instructed!). Good progress has been made on both the fAPAR retrievals for the flux sites (11 will shortly be available through the CarboData website), and the global fAPAR dataset. NG asked for guidance on the additional requirements for fAPAR products, and after some discussion it was decided that priorities should be; (i) to extend the site-level fAPAR data to each of the 20 CAMELS flux sites, and (ii) to produce global fAPAR datasets for other years (to capture interannual variability). NG also posed the question how are we (JRC), in practice, associated to the scientific interpretation of these products ?. It was agreed that the use of fAPAR data in CAMELS should be on the same basis as the use of CAMELS model outputs, and flux data, i.e. that all providers of data to a study (model outputs, direct observations or remote sensing) should be offered joint authorship on any publications which emerge. ACTION 3(g): JRC to provide extra code for reading of the FAPAR timeseries when there is missing data. ACTION 3(h): UNITUS to check on availability of FAPAR ground-based measurements from flux sites, for validation of JRC product. 3.Progress on WP2 a. Report from WP2 modellers meeting (Paris, 16/5/03): TEM parameters and a priori estimates (MPI-BGC) JK reported on the WP2 meeting in Paris. The aim was to come up with a list of TEM parameters and default values for these parameters, where possible along with a priori error bars based on citeable literature. The different modelling groups had been tasked with doing the literature searches for various aspects of the TEMs. JK showed good progress at MPI-BGC on deriving ranges for the parameters of leaf photosynthesis, but the other WP2 groups (LSCE, Hadley-CEH) had not yet provided input on the areas of TEM parameterisation allocated to them. PC wondered whether progress might be faster if group were instead responsible for defining a priori uncertainties for their own model parameters. However, JK felt that the original subdivision of tasks would work and should be retained. A hard deadline of end Dec was agreed for this task. ACTION 3(i): MODELLING GROUPS to provide uncertainties on parameters, as agreed in the Paris sub-meeting (by end Dec). b. Preliminary parameter optimisation for ORCHIDEE including Bordeaux, Bray-site (LSCE) DS presented his results on optimisation of ORCHIDEE using a Bayesian cost function. The optimisation was shown to improve the fit to the amplitude of the daily cycle in the CO2 and H2O fluxes. However, the optimisation algorithm appeared to be stopping short of the global minimum error. This bug was identified by DS by single parameter variation about the optimum identified by the algorithm. RD also described his work optimising the simple light-use efficiency TEM, called TURC. TURC is sufficiently fast to be an excellent test bed for a range of data assimilation and parameter optimisation approaches (including those which are candidates for optimisation of ORCHIDEE). c. Preliminary parameter optimisation for MOSES (CEH, Hadley) CH described optimisation of MOSES for two coniferous forest sites (Loobos and ?). A NAG routine had been used to optimise MOSES parameters associated with leaf-level photosynthesis, stomatal conductance and soil water stress. For both sites it had been impossible to get simultaneously good fits to the diurnal cycles of both the CO2 and H2O fluxes. CH explained that the source of this systematic error appeared to be in the Sellers big-leaf scaling-up from leaf to canopy. PC described the work which is underway at Hadley to replace this part of MOSES with a more explicit treatment of canopy radiation absorption, which at least recognises separate light-limited and non-light-limited leaf classes. d. Monte Carlo simulations with BETHY (MPI-BGC) JK described work at MPI-BGC on the use of a Bayesian approach and the Metropolis Monte Carlo method to estimate parameter PDFs for BETHY. This work was carried out initially for the Loobos coniferous forest site. JK had tested the extent to which the derived global optimum was dependent on the starting point in the BETHY parameter space. He generally found that most starting points ended up at the same optimum paraemeter set, but some did end up within local minima of the error function. PC asked whether we should be including such local minima in our definitions of the parameter PDFs, rather than considering these as failures of the optimisation algorithms. It was suggested that a simple model like TURC could be used to explicitly map its entire error function (within the parameter phase space), to help assess the contribution of local minima to the total PDF. 4 Estimates of measurements errors (ALTERRA) BK presented his work on estimating the uncertainties in flux measurements. He produced an Excel spreadsheet quantifying the sources of error and producing uncertainties for daytime and night-time measurements in various conditions. It was widely felt that this study offered an excellent basis for defining prior uncertainties for flux measurements. BK agreed to make his spreadsheet accessible from the CAMELS website. 5 Progress on WP5 a. Update on the CAMELS website (Hadley) PC displayed the mock CAMELS website, with links to project documents, minutes of meetings and updates on deliverables. The CAMELS web address will be www.camels.co.uk, but this website is not currently available owing to the move of the Hadley Centres webserver to the new Met Office site in Exeter. PC hoped to have the CAMELS website live by the end of the year at the latest. VJ will be the CAMELS webmaster. b. Links to CarboEurope and the European Commission (UNITUS) RV explained that CAMELS had a central role in CarboEurope, especially as full carbon accounting is becoming more favoured amongst policymakers. RV described various dissemination activities which he had undertaken, in which CAMELS was relevant and/or featured. These included (i) the Global Carbon Project (RV and PC are on the GCP scientific steering committee); (ii) an EU workshop on climate policy; (iii) contribution to the IPCC special report on good practice guidelines. RV will also fly the CAMELS flag at the CoP9 UNFCCC in Milano (Dec 2003) and the CarboEurope GHG workshop on terrestrial carbon accounting (Jan 2004). c. Plans for CAMELS brochure at end of 2nd year (UNITUS) RV described his plans for a high level overview paper to display CAMELS results (amongst others), on Energy Requirements and the Vulnerability of Land Carbon. This would use the recent uncertainty range in the projections of the future global land carbon sink, as estimated by Gruber et al. (SCOPE report), as its starting point. The critical point to make is that uncertainties in the land carbon sink (now and in the future) feed through into uncertainties in the technological development required to stabilise CO2 at non-dangerous levels. The CAMELS project is required to reduce these uncertainties! RV hoped that a paper of this type would form the basis for the CAMELS brochure which is to be delivered by the end of year 2. MH went on to explain the role of CAMELS in the new FP6 CarboEurope integrated project. CAMELS will be a key part of the Integration component of the CarboEurope-IP, which aims to quantify the western Europen carbon budget to within 10-20%! PC asked whether the CarboEurope machinery would be available for publication of CAMELS brochures etc., and MH confirmed that it would be. 6 A priori model parameter uncertainties (MPI-BGC) JK reiterated the requirement for the modelling groups to provide input on prior parameter uncertainty estimates (see action 3(i) which is now urgent!). JK showed MPI-BGC work on estimating uncertainties in longwave emissivities (which are close to 1 for vegetation) and Farquhar-type photosynthesis parameters. 7 Plans for WP3 a. Discussion of design of WP3 model experiments, including potential symbioses with other projects such as C4MIP (LSCE) RD introduced the discussion of this topic based on some questions from PF (an absent, proud, and tired new father!). It was decided that the CRU climate data would be suitable for the historical simulations, along with some algorithms to derive surface LW and SW from cloud cover, and possibly a stochastic weather generator (to be provided by CEH). It was felt that the initial TEM state should correspond to 1700 so that the full history of land-use changes from 1700 (as given in the land-use dataset, see item 2c) can be incorporated. In order to achieve this some estimate of the climate from 1700 to 1900 (the first year in the CRU dataset) will be required. The simplest approach is to use constant 1900 climate for this period (perhaps with some artificial interannual variability added). However, there is also the possibility of applying climatic anomalies (relative to 1900) from a long historical GCM simulation (e.g. Simon Tetts 500 year HadCM3 run). ACTION 3(j): Hadley (RB) to check on the availability of climate data from Simon Tetts run, for model spin-up in WP3. ACTION 3(k): LSCE (PF) to design the protocol for WP3 simulations, based on the discussion summarised in the minutes of the 3rd meeting. 8 Plans for WP4 a. Discussion of meteorological data to drive TEMs, including potential symbioses with other projects such as GSWP2 (Hadley) PC introduced the discussion on the meteorological data for WP4. It was generally agreed that it would be best to adopt the driving data prepared for the Global Soil Wetness Project 2 (GSWP2). 9 Any other business a. Annual Report PC explained the urgency with which CAMELS needed to provide an annual report to Claus Bruning. CAMELS started on 1st November 2002, and so has now completed its first year (happy birthday to all CAMELS!!). RV provided a copy of the reporting guidelines for FP projects. Hadley will write the management report, an put together the scientific report based on inputs as outlined in the actions below. Please note the tight deadlines and do all you can to meet these! ACTION 3(l): ALTERRA to collate WP1 annual report based on reports on deliverables from participating groups, and forward to RB (by mid Dec!) ACTION 3(m): ALTERRA to write 2-3 page report on D1.1 (by end Nov) ACTION 3(n): LSCE to send 1-2 page report on D1.2 to ALTERRA (by end Nov) ACTION 3(o): Hadley to send 2-3 page report on D1.3 to ALTERRA (by end Nov) ACTION 3(p): EFI to send 2-3 page report on D1.4 to ALTERRA (by end Nov) ACTION 3(q): MPI-BGC to collate WP2 annual report based on estimates of measurement and model parameters uncertainties, and forward to RB (by mid-Dec) ACTION 3(r): Hadley to produce annual report based on reports on WP1 and WP2 (by end Dec). b. CAMELS Publications There was a discussion of possible early CAMELS publications, based on the use of eddy flux data to calibrate the TEMs. Two initial papers were envisaged: ACTION 3(s): MPI-BGC(JK) to lead on paper on Use of eddy flux data to reduce uncertainties in TEM parameterisation (for submission in July 04) ACTION 3(t): CEH(CH) and ALTERRA (IvdW) to lead on paper on Robustness of land-surface parameterisation within plant functional type (for submission Nov 04) Meeting closed: 2pm on 12th November. |
