Assessment of spatiotemporal patterns of gross primary productivity in the arctic and boreal ecosystems using multi-source products

Since an increasing number of global gross primary productivity (GPP) products have become available and been applied in climate change research, there is an urgent need to compare their performance in capturing spatial and temporal variability, especially in the regions where the number of training data is limited or model parameters are of relatively larger uncertainty. Here, we investigated the spatial patterns of interannual trends and variations, and seasonal-cycle amplitudes of GPP in the arctic and boreal zones, and explored the differences across various GPP products during the overlapping period (2000&#8722;2010). We compared three main types of state-of-the-art GPP products, including simulations derived from terrestrial biosphere models of the Multi-scale Synthesis and Terrestrial Model Intercomparison Project using drivers under different scenarios, 3 datasets up-scaled from FLUXNET eddy covariance measurements based on machine-learning algorithms, and 2 semi-empirical or empirical remotely sensed products based on different satellite data. We also examined the differences of GPP variability across the main ecosystem types, mainly including tundra and taiga, and assessed the contributions of different ecosystems to the temporal variations of total GPP in this zone. The results showed all the products could capture the interannual and seasonal variability of GPP, but the spatial patterns varied largely, which was in-deep discussed. This study will benefit the usage of the GPP products in the carbon cycle research for the arctic and boreal ecosystems.

Download Full-text

Mapping (un)certainty of machine learning-based spatial prediction models based on predictor space distances

10.5194/egusphere-egu2020-8492 ◽

2020 ◽

Author(s):

Hanna Meyer ◽

Edzer Pebesma

Keyword(s):

Machine Learning ◽

Spatial Patterns ◽

Environmental Science ◽

Prediction Models ◽

Learning Algorithms ◽

Predictor Variable ◽

Spatial Prediction ◽

Machine Learning Algorithms ◽

Training Data ◽

Field Samples

Spatial mapping is an important task in environmental science to reveal spatial patterns and changes of the environment. In this context predictive modelling using flexible machine learning algorithms has become very popular. However, looking at the diversity of modelled (global) maps of environmental variables, there might be increasingly the impression that machine learning is a magic tool to map everything. Recently, the reliability of such maps have been increasingly questioned, calling for a reliable quantification of uncertainties.Though spatial (cross-)validation allows giving a general error estimate for the predictions, models are usually applied to make predictions for a much larger area or might even be transferred to make predictions for an area where they were not trained on. But by making predictions on heterogeneous landscapes, there will be areas that feature environmental properties that have not been observed in the training data and hence not learned by the algorithm. This is problematic as most machine learning algorithms are weak in extrapolations and can only make reliable predictions for environments with conditions the model has knowledge about. Hence predictions for environmental conditions that differ significantly from the training data have to be considered as uncertain.To approach this problem, we suggest a measure of uncertainty that allows identifying locations where predictions should be regarded with care. The proposed uncertainty measure is based on distances to the training data in the multidimensional predictor variable space. However, distances are not equally relevant within the feature space but some variables are more important than others in the machine learning model and hence are mainly responsible for prediction patterns. Therefore, we weight the distances by the model-derived importance of the predictors.&#160;As a case study we use a simulated area-wide response variable for Europe, bio-climatic variables as predictors, as well as simulated field samples. Random Forest is applied as algorithm to predict the simulated response. The model is then used to make predictions for entire Europe. We then calculate the corresponding uncertainty and compare it to the area-wide true prediction error.&#160;The results show that the uncertainty map reflects the patterns in the true error very well and considerably outperforms ensemble-based standard deviations of predictions as indicator for uncertainty.The resulting map of uncertainty gives valuable insights into spatial patterns of prediction uncertainty which is important when the predictions are used as a baseline for decision making or subsequent environmental modelling. Hence, we suggest that a map of distance-based uncertainty should be given in addition to prediction maps.

Download Full-text

Atmospheric temperature variability in the Arctic as revealed in a TOVS data record

Polar Record ◽

10.1017/s003224740001370x ◽

1995 ◽

Vol 31 (177) ◽

pp. 199-210

Author(s):

Siri Jodha Singh Khalsa ◽

Jeffrey R. Key

Keyword(s):

Spatial Patterns ◽

Principal Component ◽

Atmospheric Temperature ◽

Temperature Variability ◽

The Arctic ◽

Spatial And Temporal Variability ◽

Climate Change Scenarios ◽

Tropospheric Temperature ◽

Data Record ◽

Rawinsonde Data

AbstractThe Earth's high-latitude regions are of critical importance in many climate-change scenarios, but a time continuous, spatially complete, and well-calibrated record of tropospheric temperatures is needed in order to assess past and future climate changes. Studies of recently compiled upper-air data sets show no evidence of CO2-induced warming, but the spatial pattern of tropospheric temperature variability in the Arctic has not been thoroughly examined. This study analyzes a 108-month segment of the data record from the TIROS Operational Vertical Sounder (TOVS) aboard NOAA polar-orbiting satellites to examine both the spatial and temporal variability of atmospheric temperature in the Arctic.Temperature retrievals based on clear-column radiances archived at NOAA/NESDIS were done using algorithms tuned to Arctic conditions. The retrieved temperatures compared well with Arctic rawinsonde data, and include lowlevel inversions that are often problematic for satellite retrievals. The amplitude of the seasonal cycle in 500 mbar temperatures from the TOVS, NMC, and rawinsonde data generally agreed, whereas the phase comparisons produced mixed results. Principal component analyses of the TOVS and NMC temperatures revealed both monopole and dipole spatial patterns in the component loadings. Spatial patterns of the correlation between the upper-air data and the TOVS retrievals were similar to the principal component loading patterns. Whereas no significant trends were found in the station data for the same period as the TOVS record, a significant negative trend could be seen in the first principal component scores of the TOVS retrievals.

Download Full-text

Satellite-based Sun-Induced Chlorophyll Fluorescence in the Greater Alpine Space: Spatial Patterns and Relationship to Gross Primary Productivity

10.5194/egusphere-egu2020-20467 ◽

2020 ◽

Author(s):

Ulisse Gomarasca ◽

Gregory Duveiller ◽

Alessandro Cescatti ◽

Georg Wohlfahrt

Keyword(s):

Remote Sensing ◽

Chlorophyll Fluorescence ◽

Land Cover ◽

Spatial Patterns ◽

Primary Productivity ◽

Climate Models ◽

Gross Primary Productivity ◽

Accurate Estimation ◽

Small Scale ◽

Alpine Space

Accurate estimation of terrestrial gross primary productivity is essential for the development of credible future carbon cycle and climate simulations. Current remote sensing techniques allow retrieval of sun-induced chlorophyll fluorescence (SIF) as a valid proxy for GPP, but low resolution, sparse coverage, or resolution mismatches between the different satellite sensors hinder our ability to effectively link SIF to many environmental variables at fine scales. In order to better characterize heterogeneous landscapes, several attempts to downscale SIF products to higher resolutions have been made. We investigate the ability of the downscaled GOME-2 product developed by Duveiller et al. (2019), to capture the differences in spatiotemporal dynamics over the Greater Alpine Space. We analyse SIF in connection to land cover and elevation, and calculate land phenology metrics based on seasonal SIF time series. Ground-based GPP validation suggests biome-specific SIF-GPP relationships, but the comparison was hindered by the resolution mismatch of the data. Moreover, missing data are present at high elevations, diminishing the suitability of current SIF products to analyse cloud-prone mountainous areas. Important insights into spatial patterns and seasonal trends could be inferred at forest and other large-area land cover types, typical of mid elevations in the Alps, but many anthropogenic habitats at low elevations, as well as high elevation grasslands and other small-scale heterogeneous environments could not be thoroughly investigated and are likely to be underrepresented or prone to biases. Similar downscaling procedures might be applied at finer scales to e.g. TROPOMI products, or alternative advanced remote sensing SIF techniques and instruments might be needed in order to enable detailed and systematic evaluations of the Alpine region or similar highly heterogenous landscapes, before a process-oriented monitoring and unbiased implementation into climate models may be performed.

Download Full-text

Bayesian uncertainty quantification of spatio-temporal trends in soil organic carbon using INLA and SPDE

10.5194/egusphere-egu2020-9154 ◽

2020 ◽

Author(s):

Nicolas P.A. Saby ◽

Thomas Opitz ◽

Bifeng Hu ◽

Blandine Lemercier ◽

Hocine Bourennane

Keyword(s):

Climate Change ◽

Organic Carbon ◽

Soil Organic Carbon ◽

Spatial Patterns ◽

Temporal Trends ◽

Model Parameters ◽

Spatial And Temporal Variability ◽

Statistical Inferences ◽

Spatio Temporal ◽

General Response

The assumption of spatial and temporal stationarity does not hold for many ecological and environmental processes. This is particularly the case for many soil processes like carbon sequestration, often driven by factors such as biological dynamics, climate change and anthropogenic influences. For better understanding and predicting such phenomena, we develop a Bayesian inference framework that combines the integrated nested Laplace approximation (INLA) with the stochastic partial differential equation approach (SPDE). We put focus on modeling complex temporal trends varying through space with an accurate assessment of uncertainties, and on spatio-temporal mapping of processes that are only partially observed.We model observed data through a latent (i.e., unobserved) smooth process whose additive components are endowed with Gaussian process priors. We use the SPDE approach to implement flexible sparse-matrix approximations of the Mat&#233;rn covariance for spatial fields. The separate specification of the spatially varying linear trend allows us to conduct component-specific statistical inferences (range and variance estimates, standard errors, confidence bounds), and to provide maps to stakeholders for time-invariant spatial patterns, spatial patterns in slopes of time trends, and the associated uncertainties. For observed data following a Gaussian distribution, we add independent measurement errors, but more general response distributions of the data can be implemented. We also include in our model covariate information on parent material, climate and seasonality.The INLA method and its implementation in the R-INLA library provide a rich toolbox for statistical space-time modelling while sidestepping typical convergence problems arising with simulation-based techniques using Markov Chain Monte&#8211;Carlo codes for large and complex hierarchical models such as ours. Uncertainties arising in model parameters and in pointwise spatio-temporal predictions are naturally captured in the posterior distributions computed through INLA using appropriate approximation techniques, and we can communicate on them through maps of various properties. Moreover, INLA also allows for direct simulation from the estimated posterior model, such that we can conduct statistical inferences on more complex functionals of the multivariate predictive distributions by analogy with MCMC frameworks.Soil organic carbon is a major compartment of the global carbon cycle and small variations of its level can largely impact atmospheric CO2 concentrations. In the context of global climate change, it is important to be able to quantify and explain spatial and temporal variability of SOC in order to forecast future changes. In this work, we used this approach to study possible trends in space and time of soil carbon stock of three agricultural fields in France. Fitted models reveal signi&#64257;cant temporal trends with strong spatial heterogeneity. The Mat&#233;rn model and SPDE approach provide a flexible framework with respect to field design.

Download Full-text

Improved Modeling of Gross Primary Productivity of Alpine Grasslands on the Tibetan Plateau Using the Biome-BGC Model

Remote Sensing ◽

10.3390/rs11111287 ◽

2019 ◽

Vol 11 (11) ◽

pp. 1287 ◽

Cited By ~ 3

Author(s):

Yongfa You ◽

Siyuan Wang ◽

Yuanxu Ma ◽

Xiaoyue Wang ◽

Weihua Liu

Keyword(s):

Tibetan Plateau ◽

Primary Productivity ◽

Remotely Sensed ◽

Gross Primary Productivity ◽

Model Parameters ◽

The Tibetan Plateau ◽

Alpine Grasslands ◽

Biogeochemical Models ◽

Sensing Technology ◽

Sensitivity Analysis Method

The ability of process-based biogeochemical models in estimating the gross primary productivity (GPP) of alpine vegetation is largely hampered by the poor representation of phenology and insufficient calibration of model parameters. The development of remote sensing technology and the eddy covariance (EC) technique has made it possible to overcome this dilemma. In this study, we have incorporated remotely sensed phenology into the Biome-BGC model and calibrated its parameters to improve the modeling of GPP of alpine grasslands on the Tibetan Plateau (TP). Specifically, we first used the remotely sensed phenology to modify the original meteorological-based phenology module in the Biome-BGC to better prescribe the phenological states within the model. Then, based on the GPP derived from EC measurements, we combined the global sensitivity analysis method and the simulated annealing optimization algorithm to effectively calibrate the ecophysiological parameters of the Biome-BGC model. Finally, we simulated the GPP of alpine grasslands on the TP from 1982 to 2015 based on the Biome-BGC model after a phenology module modification and parameter calibration. The results indicate that the improved Biome-BGC model effectively overcomes the limitations of the original Biome-BGC model and is able to reproduce the seasonal dynamics and magnitude of GPP in alpine grasslands. Meanwhile, the simulated results also reveal that the GPP of alpine grasslands on the TP has increased significantly from 1982 to 2015 and shows a large spatial heterogeneity, with a mean of 289.8 gC/m2/yr or 305.8 TgC/yr. Our study demonstrates that the incorporation of remotely sensed phenology into the Biome-BGC model and the use of EC measurements to calibrate model parameters can effectively overcome the limitations of its application in alpine grassland ecosystems, which is important for detecting trends in vegetation productivity. This approach could also be upscaled to regional and global scales.

Download Full-text

Spatial, Phenological, and Inter-Annual Variations of Gross Primary Productivity in the Arctic from 2001 to 2019

Remote Sensing ◽

10.3390/rs13152875 ◽

2021 ◽

Vol 13 (15) ◽

pp. 2875

Author(s):

Dujuan Ma ◽

Xiaodan Wu ◽

Xuanlong Ma ◽

Jingping Wang ◽

Xingwen Lin ◽

...

Keyword(s):

Land Cover ◽

Carbon Cycle ◽

Primary Productivity ◽

Gross Primary Productivity ◽

The Arctic ◽

Spatiotemporal Variations ◽

Land Cover Types ◽

Annual Variations ◽

Terrestrial Carbon Cycle

Quantifying the spatial, seasonal (phenological), and inter-annual variations of gross primary productivity (GPP) in the Arctic is critical for comprehending the terrestrial carbon cycle and its feedback to climate warming in this region. Here, we evaluated the accuracy of the MOD17A2H GPP product using the FLUXNET 2015 dataset in the Arctic, then explored the spatial patterns, seasonal variations, and interannual trends of GPP, and investigated the dependence of the spatiotemporal variations in GPP on land cover types, latitude, and elevation from 2001 to 2019. The results showed that MOD17A2H was consistent with in situ measurements (R = 0.8, RMSE = 1.26 g C m−2 d−1). The functional phenology was also captured by the MOD17A2H product (R = 0.62, RMSE = 9 days) in the Arctic. The spatial variation of the seasonal magnitude of GPP and its interannual trends is partly related to land cover types, peaking in forests and lowest in grasslands. The interannual trend of GPP decreased as the latitude and elevation increased, except for the latitude between 62°~66° N and elevation below 700 m. Our study not only revealed the variation of GPP in the Arctic but also helped to understand the carbon cycle over this region.

Download Full-text

Simulation of Gross Primary Productivity Using Multiple Light Use Efficiency Models

Land ◽

10.3390/land10030329 ◽

2021 ◽

Vol 10 (3) ◽

pp. 329

Author(s):

Jun Zhang ◽

Xufeng Wang ◽

Jun Ren

Keyword(s):

Remote Sensing ◽

Primary Productivity ◽

Bayesian Model Averaging ◽

Model Averaging ◽

Gross Primary Productivity ◽

Accurate Estimation ◽

Model Parameters ◽

Light Use Efficiency ◽

Multiple Model ◽

Use Efficiency

Gross primary productivity (GPP) is the most basic variable in a carbon cycle study that determines the carbon that enters the ecosystem. The remote sensing-based light use efficiency (LUE) model is one of the primary tools that is currently used to estimate the GPP at the regional scale. Many remote sensing-based GPP models have been developed in the last several decades, and these models have been well evaluated at some sites. However, an accurate estimation of the GPP remains challenging work using LUE models because of uncertainties in the model caused by model parameters, model forcing, and vegetation spatial heterogeneity. In this study, five widely used LUE models, Glo-PEM, VPM, EC-LUE, the MODIS GPP algorithm, and C-fix, were selected to simulate the GPP of the Heihe River Basin forced using in situ measurements. A multiple-model averaging method, Bayesian model averaging (BMA), was used to combine the five models to obtain a more reliable GPP estimation. The BMA was trained using carbon flux data from five eddy covariance towers located at dominant vegetation types in the study area. Generally, the BMA method performed better than any single LUE model. From the case study in the study area, it is indicated that the trained BMA is an efficient method to combine multiple LUE models and can improve the GPP simulation accuracy.

Download Full-text