Van Huong Le

Environmental observatory networks are fundamental in advancing scientific understanding of biogeochemical processes. FLUXNET is a global network of regional eddy covariance networks that measure ecosystem‐scale exchanges of greenhouse gases (e.g., CO ₂ , CH ₄ , H ₂ O) and energy between the biosphere and the atmosphere. MexFlux is the eddy covariance network of Mexico, a megadiverse country with many underrepresented ecosystems within FLUXNET. This study evaluates the spatial representativeness of MexFlux by assessing its ability to capture the statistical and spatial heterogeneity of annual gross primary productivity (GPP) and evapotranspiration (ET) within Mexico. We tested four network configurations: the historical distribution of MexFlux sites (MexFlux‐H, n = 33), an expanded network with 20 additional sites (MexFlux + 20, 53 sites), MexFlux sites with publicly available data (MexFlux‐P, n = 20), and a hypothetical optimized design with only 25 sites (MexFlux25, n = 25). Results show that MexFlux‐H and MexFlux‐P overrepresent regions with GPP values between 250 and 600 gC m ⁻² yr ⁻¹ and ET of 200–1,200 mm yr ⁻¹ . MexFlux + 20 demonstrates that adding 20 strategically located sites improves the representativeness of MexFlux while preserving the historical distribution of the network. The configuration of MexFlux25 highlights that a few but strategically distributed sites are an alternative way to enhance the representativeness of the network. Mountain regions, tropical forests, and urban sites may remain underrepresented in any network configuration, highlighting the challenges of monitoring efforts in this country. Our framework integrates distributions, copulas, semivariograms, and upscaling to highlight the value of a multidimensional assessment of spatial representativeness, which applies to other regional FLUXNET networks.

Implementation of the autocorrelated conditioned Latin Hypercube Sampling (acLHS) algorithm for 1D (time-series) and 2D (spatial) data. The acLHS algorithm is an extension of the conditioned Latin Hypercube Sampling (cLHS) algorithm that allows sampled data to have similar correlative and statistical features of the original data. Only a properly formatted dataframe needs to be provided to yield subsample indices from the primary function. For more details about the cLHS algorithm, see Minasny and McBratney (2006), . For acLHS, see Le and Vargas (2024) .

Long-term information is crucial for enhancing our ability to predict ecosystem trajectories under adaptive management and climate change scenarios. In this study, we assessed how the systematic incorporation of information affects our predictive capacity regarding the response of a target variable, offering insights into ecosystem dynamics and highlighting the importance of long-term data. We analyzed over 20 years of aboveground net primary production (ANPP) data across three highly dynamic dryland ecosystems in a grassland-to-shrubland transition zone. Our approach, widely applicable for testing long-term observations, involved modeling probability distributions, temporal semivariograms, and copula-based dependency functions between annual precipitation and ANPP. Our results indicate non-linear trends in prediction capacity as more data are included, demonstrating emergent unexpected responses not evident in short-term observations. These dynamic and non-stationary responses pose significant challenges for prediction, even with over 20 years of data, underscoring the need for ongoing measurements. Our findings emphasize the importance of long-term temporal variability for understanding trends and resilience of ecosystem processes. Quantitative methodologies for assessing predictive capacity and identifying trends are essential for making informed decisions regarding the continuation or termination of long-term monitoring initiatives. We strongly advocate for the sustained support of long-term ecological research, as it is crucial for deepening our understanding of ecosystem responses and for guiding effective management and policy decisions.

Accurate modeling of dependencies between variables of interest is imperative for understanding biophysical processes and mechanisms relevant to Earth Science research. This study presents a methodology to model the temporal or spatial distributions of variables using copula-based cosimulation (CopCoSim). This technique models and simulates the joint distribution of multiple variables by capturing their dependencies and correlations. We compared CopCoSim with the traditional Sequential Gaussian CoSimulation (SGCoSim) technique through two applications relevant to Earth Science representing one (i.e., time) and two dimensions (i.e., space). We present an application for soil CO2 efflux, which is a major flux in the global carbon budget, using two case studies: (1) temporal distribution of soil CO2 efflux and temperature, and (2) spatial distribution of soil CO2 efflux and temperature across the conterminous United States (CONUS). The methodology involves three steps: data input, application of stochastic simulation methods, and comparison of simulation outputs. The results indicate that CopCoSim provides a better model with higher precision and accuracy for representing variables of interest. CopCoSim better reproduces the univariate probability distribution, temporal or spatial autocorrelation, and dependency relationship between the predictor and response variables. We propose that CopCoSim is useful for research in Earth Science, where variables of interest (e.g., soil CO2 efflux and temperature) are often interdependent and exhibit complex temporal or spatial patterns.

El método de co-simulación estocástica espacial, basado en cópulas, es un método general que permite simular variables con cualquier tipo de dependencia y funciones de distribución de probabilidad. Esta flexibilidad proviene del uso de un modelo de cópula para la representación de la función de distribución de probabilidad conjunta. El método se ha implementado principalmente a través de un enfoque no paramétrico utilizando cópulas de Bernstein y se ha aplicado con éxito para la simulación de propiedades petrofísicas usando atributos sísmicos elásticos como variables secundarias. En el presente trabajo este método se implementa mediante otros dos enfoques: paramétrico y semi-paramétrico. Específicamente, para el enfoque paramétrico se usa la familia de cópulas Arquimedianas. Primero, el enfoque paramétrico se valida con un caso publicado y luego se realiza una comparación de los tres enfoques en términos de precisión y rendimiento. Los resultados mostraron que el enfoque paramétrico es el que peor reproduce las estadísticas de los datos y presenta mayor incertidumbre con un menor costo computacional, mientras que el enfoque no-paramétrico resultó el que mejor reproduce la dependencia de los datos a un alto costo computacional. El enfoque semi-paramétrico reduce un 10% el costo computacional respecto al no-paramétrico, pero se degrada significativamente su precisión.

A fundamental question for any sampling design is identifying where and when to measure. Technological advances allow us to measure multiple greenhouse gases (GHGs) simultaneously, and now it is possible to provide complete GHG budgets from soils (i.e., CO2, CH4, and N2O fluxes). We present a data-driven method for identifying optimized samples from times series (1D approach) or spatial arrays (2D approach) of information on soil gases. The autocorrelated conditioned Latin Hypercube Sampling (acLHS) combines a conditioned Latin Hypercube (cLHS) to obtain a representative sample of the joint probability distribution function and an autocorrelation model to ensure a reproducible spatial or temporal dependency function (i.e., temporal or spatial variability). The results show a conflict between the convenience of simultaneously measuring multiple soil GHG fluxes at fixed time intervals (e.g., once or twice per month) and the intrinsic temporal variability in and patterns of different GHG fluxes. Furthermore, the acLHS is more efficient than other sampling methods (i.e., fixed sampling, cLHS) as it can better reproduce the joint probability distribution and the temporal or spatial variability of the variables of interest. We test this approach using time series and spatial arrays to evaluate the relationship between soil CO2 efflux and temperature. These results have implications for assessing gas fluxes from soils and consequently reduce uncertainty in the role of soils in biogeochemical cycles.

Quantifying the role of soils in nature-based solutions requires accurate estimates of soil greenhouse gas (GHG) fluxes. Technological advances allow us to measure multiple GHGs simultaneously, and now it is possible to provide complete GHG budgets from soils (i.e., CO₂, CH₄, and N₂O fluxes). We propose that there is a conflict between the convenience of simultaneously measuring multiple soil GHG fluxes at fixed time intervals (e.g., once or twice per month) and the intrinsic temporal variability in and patterns of different GHG fluxes. Information derived from fixed time intervals – commonly done during manual field campaigns – had limitations to reproducing statistical properties, temporal dependence, annual budgets, and associated uncertainty when compared with information derived from continuous measurements (i.e., automated hourly measurements) for all soil GHG fluxes. We present a novel approach (i.e., temporal univariate Latin hypercube sampling) that can be applied to provide insights and optimize monitoring efforts of GHG fluxes across time. We suggest that multiple GHG fluxes should not be simultaneously measured at a few fixed time intervals (mainly when measurements are limited to once per month), but an optimized sampling approach can be used to reduce bias and uncertainty. These results have implications for assessing GHG fluxes from soils and consequently reduce uncertainty in the role of soils in nature-based solutions.

Quantifying the role of soils in nature-based solutions require accurate estimates of soil greenhouse gas (GHG) fluxes. Technological advances allow to simultaneously measure multiple GHGs and now is possible to provide complete GHG budgets from soils (i.e., CO₂, CH₄ and N₂O fluxes). We propose that there is a conflict between the convenience of simultaneously measuring multiple soil GHG fluxes at fixed time intervals (e.g., once, or twice per month) and the intrinsic temporal variability and patterns of different GHG fluxes. Information derived from fixed time intervals -as is commonly done during manual field campaigns- had limitations to reproduce statistical properties, temporal dependence, annual budgets, and associated uncertainty, when compared with information derived from continuous measurements (i.e., automated hourly measurements) for all soil GHG fluxes. We present a novel approach (i.e., temporal univariate Latin Hypercube sampling) that can be applied to optimize monitoring efforts of GHG fluxes across time. We suggest that multiple GHG fluxes should not be simultaneously measured at few fixed time intervals (especially once a month), but an optimized sampling approach can be used to reduce bias and uncertainty. These results have implications for assessing GHG fluxes from soils and consequently reduce uncertainty on the role of soils in nature-based solutions.