Selecting informative bands for partial least squares regressions improves their goodness-of-fits to estimate leaf photosynthetic parameters from hyperspectral data

The plant photosynthetic capacity determines the photosynthetic rates of the terrestrial biosphere. Timely approaches to obtain the spatiotemporal variations of the photosynthetic parameters are urgently needed to grasp the gas exchange rhythms of the terrestrial biosphere. While partial least squares regression (PLSR) is a promising way to predict the photosynthetic parameters maximum carboxylation rate (V_cmax) and maximum electron transport rate (J_max) rapidly and non-destructively from hyperspectral data, the approach, however, faces a high risk of overfitting and remains a high hurdle for applications. In this study, we propose to incorporate proper band selection techniques for PLSR analysis to refine the goodness-of-fit (GoF) in estimating V_cmax and J_max. Different band selection procedures coupled with different hyperspectral forms (reflectance, apparent absorption, as well as derivatives) were examined. Our results demonstrate that the GoFs of PLSR models could be greatly improved by combining proper band selection methods (especially the iterative stepwise elimination approach) rather than using full bands as commonly done with PLSR. The results also show that the 1st order derivative spectra had a balance between accuracy (R² = 0.80 for V_cmax, and 0.94 for J_max) and denoising (when a Gaussian noise was added to each leaf reflectance spectrum at each wavelength with a standard deviation of 1%) on retrieving photosynthetic parameters from hyperspectral data. Our results clearly illustrate the advantage of using the band selection approach for PLSR dimensionality reduction and model optimization, highlighting the superiority of using derivative spectra for V_cmax and J_max estimations, which should provide valuable insights for retrieving photosynthetic parameters from hyperspectral remotely sensed data.

How eddy covariance flux measurements have contributed to our understanding of Global Change Biology

A global network of long-term carbon and water flux measurements has existed since the late 1990s. With its representative sampling of the terrestrial biosphere's climate and ecological spaces, this network is providing background information and direct measurements on how ecosystem metabolism responds to environmental and biological forcings and how they may be changing in a warmer world with more carbon dioxide. In this review, I explore how carbon and water fluxes of the world's ecosystem are responding to a suite of covarying environmental factors, like sunlight, temperature, soil moisture, and carbon dioxide. I also report on how coupled carbon and water fluxes are modulated by biological and ecological factors such as phenology and a suite of structural and functional properties. And, I investigate whether long-term trends in carbon and water fluxes are emerging in various ecological and climate spaces and the degree to which they may be driven by physical and biological forcings. As a growing number of time series extend up to 20 years in duration, we are at the verge of capturing ecosystem scale trends in the breathing of a changing biosphere. Consequently, flux measurements need to continue to report on future conditions and responses and assess the efficacy of natural climate solutions.

Estimating forest carbon fluxes using four different data-driven techniques based on long-term eddy covariance measurements: Model comparison and evaluation

With the recent availability of large amounts of data from the global flux towers across different terrestrial ecosystems based on the eddy covariance technique, the use of data-driven techniques has been viable. In this study, two advanced techniques, namely adaptive neuro-fuzzy inference system (ANFIS) and extreme learning machine (ELM), were developed and investigated for their viability in estimating daily carbon fluxes at the ecosystem level. All the data used in this study were based upon the long-term chronosequence observations derived from the flux towers in eight forest ecosystems. Both ANFIS and ELM methods were further compared with the most widely used artificial neural network (ANN) and support vector machine (SVM) methods. Moreover, we also focused on probing into the effects of internal parameters on their corresponding approaches. Our estimates showed that most variation in each carbon flux could be effectively explained by the developed models at almost all the sites. Moreover, the forecasting accuracy of each method was strongly dependent upon their respective internal algorithms. The best training function for ANN model can be acquired through the trial and error procedure. The SVM model with the radial basis kernel function performed considerably better than the SVM models with the polynomial and sigmoid kernel functions. The hybrid ELM models achieved similar predictive accuracy for the three fluxes and were consistently superior to the original ELM models with different transfer functions. In most instances, both the subtractive clustering and fuzzy c-means algorithms for the ANFIS models outperformed the most popular grid partitioning algorithm. It was demonstrated that the newly proposed ELM and ANFIS models were able to produce comparable estimates to the ANN and SVM models for forecasting terrestrial carbon fluxes.

Towards physiologically meaningful water-use efficiency estimates from eddy covariance data

Intrinsic water-use efficiency (iWUE) characterizes the physiological control on the simultaneous exchange of water and carbon dioxide in terrestrial ecosystems. Knowledge of iWUE is commonly gained from leaf-level gas exchange measurements, which are inevitably restricted in their spatial and temporal coverage. Flux measurements based on the eddy covariance (EC) technique can overcome these limitations, as they provide continuous and long-term records of carbon and water fluxes at the ecosystem scale. However, vegetation gas exchange parameters derived from EC data are subject to scale-dependent and method-specific uncertainties that compromise their ecophysiological interpretation as well as their comparability among ecosystems and across spatial scales. Here, we use estimates of canopy conductance and gross primary productivity (GPP) derived from EC data to calculate a measure of iWUE (G₁ , "stomatal slope") at the ecosystem level at six sites comprising tropical, Mediterranean, temperate, and boreal forests. We assess the following six mechanisms potentially causing discrepancies between leaf and ecosystem-level estimates of G₁ : (i) non-transpirational water fluxes; (ii) aerodynamic conductance; (iii) meteorological deviations between measurement height and canopy surface; (iv) energy balance non-closure; (v) uncertainties in net ecosystem exchange partitioning; and (vi) physiological within-canopy gradients. Our results demonstrate that an unclosed energy balance caused the largest uncertainties, in particular if it was associated with erroneous latent heat flux estimates. The effect of aerodynamic conductance on G₁ was sufficiently captured with a simple representation. G₁ was found to be less sensitive to meteorological deviations between canopy surface and measurement height and, given that data are appropriately filtered, to non-transpirational water fluxes. Uncertainties in the derived GPP and physiological within-canopy gradients and their implications for parameter estimates at leaf and ecosystem level are discussed. Our results highlight the importance of adequately considering the sources of uncertainty outlined here when EC-derived water-use efficiency is interpreted in an ecophysiological context.

Gross primary production responses to warming, elevated CO2 , and irrigation: quantifying the drivers of ecosystem physiology in a semiarid grassland

Determining whether the terrestrial biosphere will be a source or sink of carbon (C) under a future climate of elevated CO₂ (eCO₂ ) and warming requires accurate quantification of gross primary production (GPP), the largest flux of C in the global C cycle. We evaluated 6 years (2007-2012) of flux-derived GPP data from the Prairie Heating and CO₂ Enrichment (PHACE) experiment, situated in a grassland in Wyoming, USA. The GPP data were used to calibrate a light response model whose basic formulation has been successfully used in a variety of ecosystems. The model was extended by modeling maximum photosynthetic rate (A_max ) and light-use efficiency (Q) as functions of soil water, air temperature, vapor pressure deficit, vegetation greenness, and nitrogen at current and antecedent (past) timescales. The model fits the observed GPP well (R²  = 0.79), which was confirmed by other model performance checks that compared different variants of the model (e.g. with and without antecedent effects). Stimulation of cumulative 6-year GPP by warming (29%, P = 0.02) and eCO₂ (26%, P = 0.07) was primarily driven by enhanced C uptake during spring (129%, P = 0.001) and fall (124%, P = 0.001), respectively, which was consistent across years. Antecedent air temperature (Tair_ant ) and vapor pressure deficit (VPD_ant ) effects on A_max (over the past 3-4 days and 1-3 days, respectively) were the most significant predictors of temporal variability in GPP among most treatments. The importance of VPD_ant suggests that atmospheric drought is important for predicting GPP under current and future climate; we highlight the need for experimental studies to identify the mechanisms underlying such antecedent effects. Finally, posterior estimates of cumulative GPP under control and eCO₂ treatments were tested as a benchmark against 12 terrestrial biosphere models (TBMs). The narrow uncertainties of these data-driven GPP estimates suggest that they could be useful semi-independent data streams for validating TBMs.

Instantaneous-to-daily GPP upscaling schemes based on a coupled photosynthesis-stomatal conductance model: correcting the overestimation of GPP by directly using daily average meteorological inputs

Daily canopy photosynthesis is usually temporally upscaled from instantaneous (i.e., seconds) photosynthesis rate. The nonlinear response of photosynthesis to meteorological variables makes the temporal scaling a significant challenge. In this study, two temporal upscaling schemes of daily photosynthesis, the integrated daily model (IDM) and the segmented daily model (SDM), are presented by considering the diurnal variations of meteorological variables based on a coupled photosynthesis-stomatal conductance model. The two models, as well as a simple average daily model (SADM) with daily average meteorological inputs, were validated using the tower-derived gross primary production (GPP) to assess their abilities in simulating daily photosynthesis. The results showed IDM closely followed the seasonal trend of the tower-derived GPP with an average RMSE of 1.63 g C m(-2) day(-1), and an average Nash-Sutcliffe model efficiency coefficient (E) of 0.87. SDM performed similarly to IDM in GPP simulation but decreased the computation time by >66%. SADM overestimated daily GPP by about 15% during the growing season compared to IDM. Both IDM and SDM greatly decreased the overestimation by SADM, and improved the simulation of daily GPP by reducing the RMSE by 34 and 30%, respectively. The results indicated that IDM and SDM are useful temporal upscaling approaches, and both are superior to SADM in daily GPP simulation because they take into account the diurnally varying responses of photosynthesis to meteorological variables. SDM is computationally more efficient, and therefore more suitable for long-term and large-scale GPP simulations.

Seasonal, diurnal and vertical variation in photosynthetic parameters in Phyllostachys humilis bamboo plants

In recent years, temperate bamboo species have been introduced in Europe for multiple uses such as renewable bio-based materials (wood, composites, fibres, biochemicals…) and numerous ecological functions (soil and water conservation, erosion control, phytoremediation…). Despite their interesting potential, little is known on the ecophysiology of these plants in their new habitat. Therefore, we studied gas exchange parameters on a full soil bamboo plantation of Phyllostachys humilis on a test field in Ireland (Europe). We evaluated the seasonal, diurnal and vertical variation of the parameters of two commonly used photosynthetic models, i.e. the light response curve (LRC) model and the model of Farquhar, von Caemmerer and Berry (FvCB). Furthermore, we tested if there were environmental effects on the photosynthetic parameters of these models and if a correlation between photosynthetic parameters and fluorescence parameters was present, fluorescence parameters can be easily and fast determined. Our results show that the gas exchange parameters do not vary diurnally or vertically. Only seasonal variations were found and should, therefore, be taken into account when using the LRC or FvCB model when modelling canopy growth. Therefore, a big-leaf model or a sunlit-shade model can be used for modelling bamboo growth in Western Europe. There is no straightforward relation between environmental variables and the photosynthetic parameters. Although fluorescence parameters showed a correlation with the photosynthetic parameters, application of such correlation may be limited.

Controls on seasonal patterns of maximum ecosystem carbon uptake and canopy-scale photosynthetic light response: contributions from both temperature and photoperiod

Most models of photosynthetic activity assume that temperature is the dominant control over physiological processes. Recent studies have found, however, that photoperiod is a better descriptor than temperature of the seasonal variability of photosynthetic physiology at the leaf scale. Incorporating photoperiodic control into global models consequently improves their representation of the seasonality and magnitude of atmospheric CO2 concentration. The role of photoperiod versus that of temperature in controlling the seasonal variability of photosynthetic function at the canopy scale remains unexplored. We quantified the seasonal variability of ecosystem-level light response curves using nearly 400 site years of eddy covariance data from over eighty Free Fair-Use sites in the FLUXNET database. Model parameters describing maximum canopy CO2 uptake and the initial slope of the light response curve peaked after peak temperature in about 2/3 of site years examined, emphasizing the important role of temperature in controlling seasonal photosynthetic function. Akaike's Information Criterion analyses indicated that photoperiod should be included in models of seasonal parameter variability in over 90% of the site years investigated here, demonstrating that photoperiod also plays an important role in controlling seasonal photosynthetic function. We also performed a Granger causality analysis on both gross ecosystem productivity (GEP) and GEP normalized by photosynthetic photon flux density (GEP n ). While photoperiod Granger-caused GEP and GEP n in 99 and 92% of all site years, respectively, air temperature Granger-caused GEP in a mere 32% of site years but Granger-caused GEP n in 81% of all site years. Results demonstrate that incorporating photoperiod may be a logical step toward improving models of ecosystem carbon uptake, but not at the expense of including enzyme kinetic-based temperature constraints on canopy-scale photosynthesis.