[1] We describe results of a project known as OptIC (Optimisation InterComparison) for comparison of parameter estimation methods in terrestrial biogeochemical models. A highly simplified test model was used to generate pseudo-data to which noise with different characteristics was added. Participants in the OptIC project were asked to estimate the model parameters used to generate this data, and to predict model variables into the future. Ten participants contributed results using one of the following methods: Levenberg-Marquardt, adjoint, Kalman filter, Markov chain Monte Carlo and genetic algorithm. Methods differed in how they locate the minimum (gradient-descent or global search), how observations are processed (all at once sequentially), or the number of iterations used, or assumptions about the statistics (some methods assume Gaussian probability density functions; others do not). We found the different methods equally successful at estimating the parameters in our application. The biggest variation in parameter estimates arose from the choice of cost function, not the choice of optimization method. Relatively poor results were obtained when the model-data mismatch in the cost function included weights that were instantaneously dependent on noisy observations. This was the case even when the magnitude of residuals varied with the magnitude of observations. Missing data caused estimates to be more scattered, and the uncertainty of predictions increased correspondingly. All methods gave biased results when the noise was temporally correlated or non-Gaussian, or when incorrect model forcing was used. Our results highlight the need for care in choosing the error model in any optimization.

We present a first analysis of data (June 1998 to December 2000) from the long-term eddy covariance site established in a Pinus sylvestris stand near Zotino in central Siberia as part of the EUROSIBERIAN CARBONFLUX project. As well as examining seasonal patterns in net ecosystem exchange (N E), daily, seasonal and annual estimates of the canopy photosynthesis (or gross primary productivity, G P) were obtained using N E and ecosystem respiration measurements. Although the forest was a small (but significant) source of CO 2 throughout the snow season (typically mid-October to early May) there was a rapid commencement of photosynthetic capacity shortly following the commencement of above-zero air temperatures in spring: in 1999 the forest went from a quiescent state to significant photosynthetic activity in only a few days. Nevertheless, canopy photosynthetic capacity was observed to continue to increase slowly throughout the summer months for both 1999 and 2000, reaching a maximum capacity in early August. During September there was a marked decline in canopy photosynthesis which was only partially attributable to less favourable environmental conditions. This suggests a reduction in canopy photosynthetic capacity in autumn, perhaps associated with the cold hardening process. For individual time periods the canopy photosynthetic rate was mostly dependent upon incoming photon irradiance. However, reductions in both canopy conductance and overall photosynthetic rate in response to high canopy-to-air vapour differences were clearly evident on hot dry days. The relationship between canopy conductance and photosynthesis was examined using Cowan's notion of optimality in which stomata serve to maximise the marginal evaporative cost of plant carbon gain. The associated Lagrangian multiplier (λ) was surprisingly constant throughout the growing season. Somewhat remarkably, however, its value was markedly different between years, being 416 mol mol −1 in 1999 but 815 mol mol −1 in 2000. Overall the forest was a substantial sink for CO 2 in both 1999 and 2000: around 13 mol C m −2 a −1. Data from this experiment, when combined with estimates of net primary productivity from biomass sampling suggest that about 20% of this sink was associated with increasing plant biomass and about 80% with an increase in the litter and soil organic carbon pools. This high implied rate of carbon accumulation in the litter soil organic matter pool seems unsustainable in the long term and is hard to explain on the basis of current knowledge.

We present a first analysis of data (June 1998 to December 2000) from the long‐term eddy covariance site established in a Pinus sylvestris stand near Zotino in central Siberia as part of the EUROSIBERIAN CARBONFLUX project. As well as examining seasonal patterns in net ecosystem exchange (NE), daily, seasonal and annual estimates of the canopy photosynthesis (or gross primary productivity, GP) were obtained using NE and ecosystem respiration measurements. Although the forest was a small (but significant) source of CO2 throughout the snow season (typically mid‐October to early May) there was a rapid commencement of photosynthetic capacity shortly following the commencement of above‐zero air temperatures in spring: in 1999 the forest went from a quiescent state to significant photosynthetic activity in only a few days. Nevertheless, canopy photosynthetic capacity was observed to continue to increase slowly throughout the summer months for both 1999 and 2000, reaching a maximum capacity in early August. During September there was a marked decline in canopy photosynthesis which was only partially attributable to less favourable environmental conditions. This suggests a reduction in canopy photosynthetic capacity in autumn, perhaps associated with the cold hardening process. For individual time periods the canopy photosynthetic rate was mostly dependent upon incoming photon irradiance. However, reductions in both canopy conductance and overall photosynthetic rate in response to high canopy‐to‐air vapour differences were clearly evident on hot dry days. The relationship between canopy conductance and photosynthesis was examined using Cowan's notion of optimality in which stomata serve to maximise the marginal evaporative cost of plant carbon gain. The associated Lagrangian multiplier (λ) was surprisingly constant throughout the growing season. Somewhat remarkably, however, its value was markedly different between years, being 416 mol mol−1 in 1999 but 815 mol mol−1 in 2000. Overall the forest was a substantial sink for CO2 in both 1999 and 2000: around 13 mol C m−2 a−1. Data from this experiment, when combined with estimates of net primary productivity from biomass sampling suggest that about 20% of this sink was associated with increasing plant biomass and about 80% with an increase in the litter and soil organic carbon pools. This high implied rate of carbon accumulation in the litter soil organic matter pool seems unsustainable in the long term and is hard to explain on the basis of current knowledge.

Current methods for estimating vegetation parameters are generally sub-optimal in the way they exploit information and do not generally track uncertainties. We look forward in the future to operational data assimilation schemes to track land surface processes and exploit multiple types of observation. Data assimilation schemes seek to combine observations and models in a statistically optimal way taking into account uncertainty in both, but have not yet been much exploited in this area. The EO-LDAS scheme and prototype, developed under ESA funding is designed to exploit the can be reduced by treating the data to normalise for such effects (e.g. Rochdi and Fernandes, 2010). Ultimately though, however much care is taken to treat such effects, methods assuming such fixed mappings from VIs with 'statistical' models are open to many criticisms, some of the more significant of which could be considered: (i) they fail to make full use of the information content of the observational data; (ii) they (often) fail to make use of our understanding of the physics of the situation; (iii) they need recalibration if conditions change (e.g. sensor band pass functions or scale of observation); (iv) they tend not to treat uncertainty in the mapped product in any rigorous way (mostly, they fail to consider this at all).This is a judgement call. An alternative stratagem has been to build mathematical models of the physics of radiation interactions with vegetation canopies and the intervening atmosphere, phrased as functions of 'control' variables (polarisation, wavebands, viewing and illumination angles etc.) and (bio) physical parameters or 'state variables' (LAI, leaf chlorophyll concentration etc. for the canopy, and aerosol optical depth, ozone concentration etc. for the atmosphere), and to use these to attempt to interpret the satellite signal. We may call these radiative transfer (RT) models. To tie in with discussions below and to provide consistency with the data assimilation literature, such models are called here 'observation operators' (denoted H x ()) in that they map from the state variable vector x to the EO signal (as a vector) R for a given set of control variables, so the modelled signal vector R = H x (). The 'remote sensing inverse problem' then is to obtain an estimate of some function of x , F x () from measurements R. How this may be achieved is discussed in more detail below. Much effort has been devoted to producing information from EO data about specific biophysical quantities that are relevant to science and society. A major focus of this has been to attempt to provide estimates of (green) LAI. Garrigues et al. (2008) consider four representative EO-derived global LAI products, with core spatial resolutions of 1 km or coarser, that use what might be