As high-resolution geospatiotemporal data sets from observatory networks, remote sensing platforms, and computational Earth systems increase in abundance, fidelity, and richness, machine learning approaches that can fully utilize increasingly powerful parallel computing resources are becoming essential for analysis and exploration of such data sets. We explore one such approach, applying a state-of-the-art distributed memory parallel implementation of Support Vector Machine (SVM) classification to large remote-sensing data sets. We have used MODIS 8-day surface reflectance (MOD09A1) and land surface temperature (MOD11A2) for classifying wildfires over Alaska and California. Monitoring Trends in Burn Severity (MTBS) burn perimeter data was used to set boundaries of burned and unburned areas for our two-class problem. MTBS covers years from 1984-2019, recording only fires over 1000 acres or greater in the western United States. We seek to find a parallel computing solution (using the PermonSVM solver, described below) to accurately classify wildfires and find smaller unrecorded wildfires. An initial assessment for wildfire classification over interior Alaska shows that PermonSVM has an accuracy of 96% and over 5000 false positives (i.e., fires unrecorded in MTBS). Next steps include mapping larger regions over Alaska and California and understanding the tradeoffs of scalability and accuracy. The parallel tool we employ is PermonSVM, which is built on top of the widely-used open source toolkit PETSc, the Portable, Extensible Toolkit for Scientific Computation. Recent developments in PETSc have focused on supporting cutting-edge GPU-based high-performance computing (HPC) architectures, and these can be easily leveraged in PermonSVM by using appropriate GPU-enabled matrix and vector types in PETSc. We achieve significant GPU speedup for the SVM calculations on the Summit supercomputer at Oak Ridge National Laboratory – currently one of the best available “at scale” proxies for upcoming exascale-class supercomputers – and are actively working to further improve computational efficiency on Summit as well as on prototype exascale node architectures.
From extracting nutrients to releasing energy, biological metabolism plays an integral role in determining evolutionary patterns of organisms through geologic time. A previous study depicted a positive relationship between metabolic rate and extinction probability for Mollusca within the Neogene period. We hypothesized that this relationship extends to other metazoan phyla during the Cenozoic Era. Using specific respiration rates measured from living organisms and body size data for fossil taxa, we estimated metabolic rates of animals across different phyla: Arthropoda, Brachiopoda, Echinodermata, and Mollusca. This analysis was performed at the class level by using the classes with the most data available to represent each phylum: Malacostraca, Ostracoda, Cirripedia, Rhynchonellata, Echinoidea, Bivalvia, Cephalopoda, and Gastropoda. We then used logistic regression to estimate the relationship between the calculated metabolic rates and extinction probability during each epoch of the Cenozoic Era. Results indicate that while each individual phylum has a different extinction probability across each epoch, the regression coefficients for the combination of all studied phyla illustrate no relationship since there is not enough evidence to reject the null hypothesis of no relationship between metabolic rate and extinction probability. Although this means that there is no significant correlation for most of the phyla, there are some exceptions where metabolism does affect extinction probability. During the Oligocene epoch, animals within the Mollusca phylum portray a clear negative correlation between metabolic rate and extinction probability. A negative relationship is also observed for Echinoderms during the Eocene epoch. Despite the crucial role that metabolism plays in species survival, our results indicate that more information is needed regarding specific environmental conditions in order to accurately predict the factors that ultimately affect species survival across marine animals within the Cenozoic Era.
The conservation of natural ecosystems is an essential component for sustainable land use (LU). One of the challenges facing society worldwide is climate change, where reduce emissions, and sequestrate greenhouse gases from the atmosphere are fundamental to mitigate its effects. LU change plays a major role in the carbon (C) cycle, and understanding and quantifying its effects is one of the main challenges for effectively implementing climate change mitigation actions. Given this scenario, our objective was to calibrate the Daycent model to estimate the references equilibrium soil organic matter (SOM) for three important Brazilian biomes: Atlantic Forest (AF), Cerrado (CE), and Pampa (PA). Together, they represent 39% of native vegetation area, and over them are concentrated the majority of the agricultural production in Brazil. Estimating the equilibrium for major soil types in the three biomes is fundamental for evaluating C dynamics and the soil C loss regarding LU changes. Data from literature, including SOM, were collected for the three biomes: PA (29°30’S, 54°15’W; soil with sandy loam texture), CE (19°28’S, 44°15’W; very clayey texture) and AF (10°92’S, 37°19’W; sandy texture). Daycent parameters to represent the biomes biophysical properties were initially set up with values from local literature. Measured SOM was then employed during the calibration of the Daycent model. We ran the model for 6,000 years for the equilibrium simulations, obtaining the stabilization of the SOM compartments (active, slow, and passive). For the biomes’ biophysical properties the parameters for maximum potential production (PRDX) were adjusted for each biome, PA with 0.92 g C m-2 , AF with 1.5 g C m-2 and CE with 0.9 g C m-2 (default = 0.5 g C m-2). The relative error between measured and predicted total SOM was lower than 2% for all biomes, thus representing the equilibrium properly for the study conditions. The largest C compartment of the biomes (slow organic matter in the soil) had 71.7% for AF, 68.5% for PA, and 63.7% for CE of the total SOM. The highest SOM values were found in the CE, with 53 Mg C ha-1, followed by the PA with 37 Mg C ha-1, and in the AF with 35 Mg C ha-1. Eventual LU changes will impact the SOM equilibrium of these native vegetation, but sustainable practices must take place to avoid C losses as far as possible.
Wind-driven mixing and Ekman pumping from slow-moving tropical cyclones (TCs) can bring nutrients to the euphotic zone, promoting phytoplankton blooms (TC-PBs) observable by satellite remote sensing. We examine an exceptional (z-score = 18-48) TC-PB induced by category-1 Cyclone Oma near the South Pacific island of Vanuatu in February 2019, the most extreme event in the observed satellite record of South Pacific surface Chlorophyll-a (Chl-a). Examining 15 South Pacific TC-PBs since 1997, we identify a “hover” parameter derivable from storm track data correlated with post-TC surface Chl-a (r=0.83). Using a dataset of synthetic storm tracks, we show revisit times for South Pacific TC-PBs are O(250) years, and O(1,500) years for Oma-scale TC-PBs. The episodic, extreme, but consistent nature of such events means they may imprint on sediment records. If so, we show their signature could be used to reconstruct past TC variability.
To help store water, facilitate navigation, generate energy, mitigate floods, and support industrial and agricultural production, people have built and continue to build obstructions to natural flow in rivers. However, due to the long and complex history of constructing and removing such obstructions, we lack a globally consistent record of their locations and types. Here, we used a consistent method to visually locate and classify obstructions on 2.1 million km of large rivers (width ≥ 30m) globally. We based our mapping on Google Earth Engine’s high resolution images from 2018–2020, which for many places have meter-scale resolution. The resulting dataset, the Global River Obstruction Database (GROD), consists of 29,877 unique obstructions, covering six different obstruction types: dam, lock, low head dam, channel dam, and two types of partial dams. By classifying a subset of the obstructions multiple times, we are able to show high classification consistency (87% mean balanced accuracy) for the three types of obstructions that fully intersect rivers: dams, low head dams, and locks. The classification of the three types of partial obstructions are somewhat less consistent (61% mean balanced accuracy). Overall, by comparing GROD to similar datasets, we estimate GROD likely captured 90% of the obstructions on large rivers. We anticipate that GROD will be of wide interest to the hydrological modeling, aquatic ecology, geomorphology, and water resource management communities.
Surface-atmosphere fluxes and their drivers vary across space and time. A growing area of interest is in downscaling, localizing, and/or resolving sub-grid scale energy, water, and carbon fluxes and drivers. Existing downscaling methods require inputs of land surface properties at relatively high spatial (e.g., sub-kilometer) and temporal (e.g., hourly) resolutions, but many observed land surface drivers are not available at these resolutions. We evaluate an approach to overcome this challenge for land surface temperature (LST), a World Meteorological Organization Essential Climate Variable and a key driver for surface heat fluxes. The Chequamegon Heterogenous Ecosystem Energy-balance Study Enabled by a High-density Extensive Array of Detectors (CHEESEHEAD19) field experiment provided a scalable testbed. We downscaled LST from satellites (GOES-16 and ECOSTRESS) with further refinement using airborne hyperspectral imagery. Temporally and spatially downscaled LST compared well to observations from a network of 20 micrometeorological towers and airborne in addition to Landsat-based LST retrieval and drone-based LST observed at one tower site. The downscaled 50-meter hourly LST showed good relationships with tower (r2=0.79, precision=3.5 K) and airborne (r2=0.75, precision=2.4 K) observations over space and time, with precision lower over wetlands and lakes, and some improvement for capturing spatio-temporal variation compared to geostationary satellite. Further downscaling to 10 m using hyperspectral imagery resolved hotspots and cool spots on the landscape detected in drone LST, with significant improvement in precision by 1.3 K. These results demonstrate a simple pathway for multi-sensor retrieval of high space and time resolution LST.
Land-use decisions, particularly in an agricultural setting, lie at the nexus of the colliding challenges of climate change and food insecurity. Understanding and guiding these decisions at the regional scale is a key strategy in the development of natural climate solutions and sustainable food production systems. These issues come together in a particularly high-stakes context in the Massachusetts cranberry industry, which occupies a position of significant economic and sociocultural importance in the region, but faces a number of challenges in the form of heightened competition, unstable prices, an aging farmer population, and changing ecological conditions. Many farmers are looking either for ways to become profitable, or to exit the industry in a financially sustainable way. One option is to sell their land to developers; another option, which is exciting to scientists and environmental advocates, is undergoing an active habitat rehabilitation to restore the beneficial ecosystem services of a functioning wetland environment. Integrating satellite data and in-situ sensor data collected over the past decade, we aim to conduct a systems analysis that unites the viewpoints of cranberry industry stakeholders and clarifies the trade-offs between environmental, economic, and social factors in the region. We propose to address this aim via three core research efforts: a contextual analysis of the industry; a valuation and mapping of key ecological, economic, and social factors; and an integrated modeling approach that models interactions and trade-offs between these factors. In particular, this presentation will focus on the progress we have made valuing and mapping key environmental and economic factors using publicly available satellite imagery and census data. This work demonstrates how these factors align with existing features of the natural and built environment, supports conservation organizations and municipalities in their restoration and conservation advocacy, and provides a foundation for future scenario mapping that will analyze trade-offs in different land use cases.
Climate change and deforestation influence the rainfall patterns in the tropics, thereby increasing the risk of drought-induced forest-to-savanna transitions. Forest ecosystems respond to these changing environmental conditions by adapting various drought coping strategies driven by different magnitudes of water-stress (i.e., defined here as a deficit in soil water availability inhibiting plant growth due to change in rainfall patterns). A better understanding of forest dynamics in response to the water-stress conditions is, therefore, crucial to determine the rainforest’s present ecohydrological conditions, as well as project a possible rainforest-savanna transition scenario. However, our present understanding of such transitions is entirely based on rainfall, which does not consider the adaptability of vegetation to droughts by utilizing subsoil moisture in a quantifiable metric. Using remote-sensing derived root zone storage capacity (Sr) and tree cover, we analyze the water-stress and drought coping strategies of the rainforest-savanna ecosystems in South America and Africa. The results from our empirical and statistical analysis allows us to classify the ecosystem’s adaptability to droughts into four key classes of drought coping strategies: lowly water-stressed forest (shallow roots, high tree cover), moderately water-stressed forest (investing in Sr, high tree cover), highly water-stressed forest (trade-off between investments in Sr and tree cover) and savanna-grassland regime (competitive rooting strategy, low tree cover). This study concludes that the ecosystems’ responses are primarily focused on allocating carbon in the most efficient way possible to maximize their hydrological benefits. The insights from this study suggest remote sensing-based Sr as an important indicator revealing important subsoil forest dynamics and opens new paths for understanding the ecohydrological state, resilience, and adaptation dynamics of the tropical ecosystems under a rapidly changing climate.
The relationship between dissolved solute concentration (C) and discharge (q) in streams, i.e., the C-q relationship, is a useful diagnostic tool for understanding biogeochemical processes in watersheds. In the ephemeral glacial meltwater streams of the McMurdo Dry Valleys [MDVs], Antarctica, studies show significant chemostatic relationships for weathering solutes and NO3-. Dissolved organic carbon (DOC) concentrations here are low compared to temperate streams, in the range of 0.1 to 2 mg C L-1, and their chemical signal clearly indicates derivation from microbial biomass. Many MDV streams support abundant microbial mats, which are also a source of organic matter to underlying hyporheic sediments. We investigated whether the DOC generation rate from these autochthonous organic matter pools was sufficient to maintain chemostasis for DOC despite these streams’ large diel and interannual fluctuations in discharge. To evaluate the DOC-q relationship, we fit the long-term DOC-q data to two models: a power law and an advection-reaction model. By using model outputs and other common metrics to characterize the DOC-q relationship, we found that this relationship is chemostatic in several MDV streams. We propose a conceptual model in which hyporheic carbon storage, hyporheic exchange rates, and net DOC generation rates are key interacting components that enable chemostatic DOC-q behavior in MDV streams. This model clarifies the role of autochthonous carbon stores in maintaining DOC-q chemostasis and may be useful for examining these relationships in temperate systems, where autochthonous organic carbon is readily bioavailable but where its signal is masked by a larger allochthonous signal.
In Alaska, pervasive irregularities of permafrost coverage and associated boreal forest heterogeneity within the North American Taiga-Tundra Ecological Transition Zone (TTE) are becoming more apparent as the climate warms. These anomalies correspond to extensive shifts in active layer thickness (ALT), carbon cycle disruption, and ecosystem response patterns. The feedback complexities associated with these climate-induced disturbances are evaluated with the integration of remote sensing, modeling, field observations, data assimilation and harmonization techniques, and artificial intelligence technology. In this study, to improve our understanding of shifting belowground dynamics and how they associate with aboveground vegetation patterns, we used the SIBBORK-TTE model to derive permafrost degradation and ecosystem transiency at high-resolution in this study. The intercomparison of model version output was first examined; then, multiple verification and validation methodologies revealed distinct historical and future implications resulting from ALT variability within four regions of the Alaska TTE domain (North Slope, Yukon Delta, Seward Peninsula, Interior). To quantify historical thaw variability and identify seasonality patterns across these regions of interest, in situ ALT point measurements were collected from two campaigns (CALM, SMALT) to cross-validate ALT-derived SAR data (AirMOSS, UAVSAR) and below-ground SIBBORK-TTE simulations between 1990-2020. Future conditions were then projected with a warming climate function and CMIP6 data from CNRM-CERFACS SSP126/585 scenarios. Initial results for derived and measured annual maximum ALT yield a mean-error performance metric of 0.2294. Paradoxically, future climate conditions advance the ubiquity of permafrost thaw and seasonality widening across the TTE. With this investigative approach, spatiotemporal variability in ALT provides a unique signal to enhance model precision and lower uncertainty through fine-tuning driver forcing and modular parameterization, forecast permafrost distribution, and identify the climatic and topographic mechanisms of earth system feedbacks and land cover change.
Estimates of global carbon stocks in coastal wetlands reveal that these are some of the most efficient carbon-sequestering environments in the world, which has prompted a renewed interest in conservation and restoration programs as an opportunity for greenhouse gas abatement. Accumulation of carbon in coastal wetlands is linked to diverse factors such as the type of vegetation, geomorphic setting, and sediment supply. Feedbacks between these factors and the tidal flow conditions drive the dynamics of carbon accumulation rates. Climate change-induced sea-level rise has been shown to increase the vulnerability to submergence of saltmarsh and mangroves in coastal wetlands, even if accommodation and landward colonization are possible. These potential losses of wetland vegetation combined with the reduced productivity of newly colonized areas will directly affect the capacity of the wetlands to sequester carbon from sediments and root growth. Here, we implement an eco-geomorphic model to simulate vegetation dynamics, soil carbon accumulation, and changes in soil carbon stock for a restored mangrove-saltmarsh wetland experiencing accelerated sea-level rise. We evaluate model outcomes for existing conditions and two different management scenarios aimed at mitigating sea-level rise effects and conserve wetland vegetation. Even though some management measures can result in partial conservation of wetland vegetation, they do not necessarily result in the best option for soil carbon capture. Our results suggest that accelerated sea-level can trigger accelerated wetland colonization resulting in wetland areas with limited opportunities for soil carbon capture from sediment and root mineralization, an issue that has not been considered in previous studies.
In both natural and built environments, microbes on occasions manifest in spherical aggregates instead of solid-affixed biofilms. These microbial aggregates are conventionally referred to as granules. Cryoconites are mineral rich granules that appear on glacier surfaces and are linked with expanding surface darkening, thus decreasing albedo, and enhanced melt. The oxygenic photogranules (OPGs) are organic rich granules that grow in wastewater with photosynthetic aeration and present potential for net autotrophic wastewater treatment in a compact system. Despite obvious differences inherent in the two, cryoconite and OPG pose striking resemblance. In both, the order Oscillatoriales in Cyanobacteria envelope inner materials and develop dense spheroidal aggregates. We explore the mechanism of photogranulation on account of high similarity between cryoconites and OPGs. We contend that there is no universal external cause for photogranulation. However, cryoconites and OPGs, as well as their intra variations, which are all are under different stress fields, are the outcome of universal physiological processes of the Oscillatoriales interfacing goldilocks interactions of stresses, which select for their manifestation as granules. Finding the rules of photogranulation may enhance engineering of glacier and wastewater systems to manipulate their ecosystem impacts.
Soil carbon cycling and ecosystem functioning can strongly depend on how microbial communities regulate their metabolism and adapt to changing environmental conditions to improve their fitness. Investing in extracellular enzymes is an important strategy for the acquisition of resources, but the principle behind the trade-offs between enzyme production and growth is not entirely clear. Here we show that the enzyme production rate per unit biomass may be regulated in order to maximize the biomass specific growth rate. Based on this optimality hypothesis, we derive mathematical expressions for the biomass specific enzyme production rate and the microbial carbon use efficiency, and verify them with experimental observations. As a result of this analysis, we also find that the optimal enzyme production rate decays hyperbolically with the soil organic carbon content. We then show that integrating the optimal extracellular enzyme production into microbial models may change considerably soil carbon projections under global warming, underscoring the need to improve parameterization of microbial processes.
Model projections predict tropical forests will experience longer periods of drought and more intense precipitation cycles under a changing climate. Such transitions have implications for structure-function relationships within microbial communities. We examine how chronic drying might reshape prokaryotic and fungal communities across four lowland forests in Panama with a wide variation in mean annual precipitation and soil fertility. Four sites were established across a 1000 mm span in mean annual precipitation (2335 to 3300 mm). We expected microbial communities at sites with lower MAP to be less sensitive to chronic drying than sites with higher MAP; while fungal communities to be more resistant to disturbance than prokaryotes. At each location, partial throughfall exclusion structures were established over 10 x 10 m plots to reduce direct precipitation input. After the first nine months of throughfall exclusion, prokaryotic communities showed no change in composition. However, 18 months of throughfall exclusion resulted in markedly divergent prokaryotic community responses, reflecting MAP and soil fertility. We observed the emergence of a “drought microbiome” within infertile sites, whereby the community structure of the experimental drying plots at the lower MAP sites diverged from their respective control sites and converged towards overlapping assemblages. Furthermore, taxa increasing in relative abundance under throughfall exclusion at the highest MAP became more similar to taxa characteristic of the control plots at the lowest MAP site, suggesting a shift toward communities with life-history traits selected 1
As climate and land use increase the variability of many ecosystems, forecasts of ecological variables are needed to inform management and use of ecosystem services. In particular, forecasts of phytoplankton would be especially useful for drinking water management, as phytoplankton populations are exhibiting greater fluctuations due to human activities. While phytoplankton forecasts are increasing in number, many questions remain regarding the optimal model time step (the temporal frequency of the forecast model output), time horizon (the length of time into the future a prediction is made) for maximizing forecast performance, as well as what factors contribute to uncertainty in forecasts and their scalability among sites. To answer these questions, we developed near-term, iterative forecasts of phytoplankton 1 to 14 days into the future using forecast models with three different time steps (daily, weekly, fortnightly), that included a full uncertainty partitioning analysis at two drinking water reservoirs. We found that forecast accuracy varies with model time step and forecast horizon, and that forecast models can outperform null estimates under most conditions. Weekly and fortnightly forecasts consistently outperformed daily forecasts at 7-day and 14-day horizons, a trend which increased up to the 14-day forecast horizon. Importantly, our work suggests that forecast accuracy can be increased by matching the forecast model time step to the forecast horizon for which predictions are needed. We found that model process uncertainty was the primary source of uncertainty in our phytoplankton forecasts over the forecast period, but parameter uncertainty increased during phytoplankton blooms and when scaling the forecast model to a new site. Overall, our scalability analysis shows promising results that simple models can be transferred to produce forecasts at additional sites. Altogether, our study advances our understanding of how forecast model time step and forecast horizon influence the forecastability of phytoplankton dynamics in aquatic systems, and adds to the growing body of work regarding the predictability of ecological systems broadly.
Shifting cultivation is an important driver of forest disturbance in the tropics. However, studies of shifting cultivation are limited and current area estimates of shifting cultivation are highly uncertain. Although Southeast Asia is a hotspot of shifting cultivation, there are no national maps of shifting cultivation in Southeast Asia at moderate or high resolution (less than or equal to 30 m). Monitoring shifting cultivation is challenging because the slash-and-burn events are highly dynamic and small in size. In this research, we present and test an approach to monitoring shifting cultivation using Landsat data on Google Earth Engine. CCDC-SMA (Continuous Change Detection and Classification - Spectral Mixture Analysis) is used to detect forest disturbances. Then, these disturbances are attributed by combining time series analysis, object-based image analysis (OBIA), and post-disturbance land-cover classification. Forest disturbances are assigned to shifting cultivation, new plantation, deforestation, severe drought, and subtle disturbance annually from 1991 to 2020 at a 30-meter resolution for the country of Laos. The major forest disturbances in 1991-2020 are mapped with an overall accuracy of 85%. Shifting cultivation is mapped with a producer’s accuracy of 88% and a user’s accuracy of 80%. The margin of error of the sampling-based area estimate of Shifting cultivation is 5.9%. The area estimates indicate that shifting cultivation is the main type of forest-disturbance in Laos, affecting 32.9% ± 1.9% of Laos over the past 30 years. To study the development of shifting cultivation over time, the area of slash-and-burn events is estimated at 5-year intervals of 1991-2020 with all margins of error less than 17%. Results show that the area of slash-and-burn activities in Laos increased in the most recent 5-year period. We believe that the methods developed and tested in Laos can be applied to other regions.
The presence of Xenobiotic Organic Compounds (XOCs) in municipal wastewater treatment plants’ effluent raises a global concern due to the easy consumption of these micropollutants by organisms. The fate of XOCs removal mechanisms of these compounds remains a challenge in recent scientific research. This study aimed to create an uncalibrated mathematical fate model within the professional wastewater modeling simulation software in a first step that was able to address the fate of Sulfamethoxazole (SMX), its metabolite, and Nonylphenol ethoxylates (NPEOs) along with conventional compounds during an activated sludge process. For the calibration process as a next step, two different case studies were created with assigning related removal mechanisms to each group of compounds. In the calibration process, model parameters are tuned such that the model can best simulate the experimental data using optimization methods. The validation results showed that the proposed model successfully simulates the removal of XOCs. Results of sensitivity analysis showed that the fate model is more sensitive to biodegradation rate constant than Solid Retention Time (SRT).