Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “forest monitoring”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 757 records · Page 42

Analysis and Application of Airborne Thermal Data at the Local Level Salt Lake City, Utah

Expanding cities are transforming periurban environments such as agricultural land, natural grasslands, forests, wetlands, and and land, into urban surfaces, such as asphalt and concrete. This transformation is part of a process defined as "urban heat island". The urban surfaces get much hotter during the daylight hours in the summer than the natural or vegetated environment. The heat builds up creating a dome effect over the city making it many degrees hotter than it's surrounding area. The impacts from this, which include higher usage of air conditioners, water, etc., are numerous and costly. As cities expand, this problem is exacerbated. It is necessary to incorporate better quality data into urban analysis and for establishing methods that systematically and objectively monitor growth and change due to increased urbanization. NASA initiated Project Atlanta in 1997 "as an interdisciplinary remote sensing study to observe and measure the growth and development of the urban heat island effect over Atlanta, and its associated impacts". This project has recently included Salt Lake City, among others, in the study of the development and effects of "urban heat islands". NASA has made available to Salt Lake City, high resolution, 10 meter, multispectral thermal data collected in June 1998. The data collection was part of a special NASA over-flight, a mission supported by the U.S. EPA in conjunction with their Urban Heat Island (UHI) Mitigation Initiative. Salt Lake City is one of three pilot cities selected to participate in this unique initiative. Hence, this project constitutes a rare opportunity to capitalize upon state-of-the-art NASA technology and link it to an urban community very concerned about rapid growth and development. This data will enhance existing data and be used for improving technical tools used to plan for Utah's future.

Dudley-Murphy, Elizabeth A.↗

Hydrologic land use classification of the Patuxent River watershed using remotely sensed data

The Patuxent River Watershed is located in central Maryland between Baltimore and Washington, D.C. and is approximately 2330 sq km in area and 175 km long. This region is now at a critical point because of major concerns such as water management and quality, flooding and land use within the watershed. Data from the NASA-directed LANDSAT and Earth Resources Aircraft Programs were used to provide a new dimension in information collection and processing for the management of watersheds. Digital data from LANDSAT-1 were analyzed along with selected IR photography from U-2 flight number 74-060B taken 28 April 1974, which was digitized in three channels. Processing of the data was accomplished using a multispectral analysis system. Land use themes consisting of surface water, wetlands, forest, residential, cropland/pasture, urban, and extractive were developed and delineated through the watershed. Area measurements of watershed themes were obtained and will serve as a calibration input to a deterministic hydrologic model on a sub-watershed. Using the derived residential and urban theme areas from LANDSAT an estimated basin imperviousness was also calculated. Thematic maps were produced at 1:62,500 scale. Floodprone areas were also classified and delineated at a scale of 1:24,000. Comparison with standard floodprone area maps at the same scale have indicated a few areas of discrepancy. Such information can be used for updating or checking floodprone area boundaries as well as monitoring changes in floodplain areas.

Dallam, W. C.↗

Plant Responses to Increased UV-B Radiation: A Research Project

Ozone decrease implies more ultraviolet-B (UV-B) radiation reaching the surface of the Earth. Increased UV-B radiation triggers responses by living organisms. Despite the large potential impacts on vegetation, little is known about UV-B effects on terrestrial ecosystems. Long-term ecological studies are needed to quantify the effects of increased UV radiation on terrestrial ecosystems, asses the risks, and produce reliable data for prediction. Screening pigments are part of one of the protective mechanism in plants. Higher concentrations of screening pigments in leaves may be interpreted as a response to increased UV radiation. If the screening effect is not sufficient, important molecules will be disturbed by incoming radiation. Thus, genetics, photosynthesis, growth, plant and leaf shape and size, and pollen grains may be affected. This will have an impact on ecosystem dynamics, structure and productivity. It is necessary to monitor selected terrestrial ecosystems to permit detection and interpretation of changes attributable to global climate change and depleted ozone shield. The objectives of this project are: (1) To identify and measure indicators of the effects of increased solar UV-B radiation on terrestrial plants; (2) to select indicators with the greatest responses to UV-B exposure; (3) to test, adapt or create ecosystem models that use the information gathered by this project for prediction and to enhance our understanding of the effects of increased UV-B radiation on terrestrial ecosystems. As a first step to achieve these objectives we propose a three-year study of forest and steppe vegetation on the North slope of the Brooks Range (within the Arctic circle, in Alaska), in the Saguaro National Monument (near Tucson, Arizona) and in the forests and steppes of Patagonia (Argentina). We selected (1) vegetation north of the Polar Circle because at 70N there is 8% risk of plant damage due to increased UV-B radiation; (2) the foothills of Catalina Mountains because there is anecdotal evidence of plant damage on the saguaros that has been linked to increased UV radiation, and (3) the forests of Nothofagus spp. and the steppe of Patagonia where the risk of plant damage at 35S is 5% and increases to as much as 15% at 55S due to increased UV-B radiation. Measurements of UV-B radiation impinging on the surface at 55S largely exceed the predicted UV-B radiation values at 50 latitude and 0% ozone depletion. Preliminary HPLC analyses of UV-B absorbing compounds in Nothofagus antartica, N. pumilio, N. betuloides and Rumex sp. in natural conditions show species-specific patterns. The spectrum of N. antartica grown at 38S differs significantly from that of N. antartica in natural conditions in Ushuaia (55S). These results suggest that the selected main area (Patagonia) is appropriate for assessing the problem and its magnitude and that Nothofagus is appropriate for our study.

DAntoni, H. L.↗

Validating Protection System Behavior with Machine Learning in a Master State Overseer

As power system protection devices continue the widespread transition from analog to digital, they become increasingly intricate. The internal functions and communication between critical grid components must now be significantly more complex to keep up with the demands of the modern smart grid. This brings increased difficulty in maintenance and monitoring, making it harder to identify potential misoperation, power anomalies, and cyber threats. Such issues are often only pinpointed after an exhaustive and costly post-mortem analysis, when a major outage or damage has already occurred. A solution is needed for validating protection systems as they operate, independently evaluating grid state and confirming whether the protection system is behaving accordingly. As opposed to incident response, this acts as a constant verification mechanism that raises a flag when subtler issues are noticed, catching them earlier and preventing larger incidents. This work presents the implementation of such a system, expanding on the prototype developed by the authors in a previous paper. This is accomplished with a machine learning (ML) system capable of validating the performance of protection systems by classifying anomalous events and characterizing protection system responses based solely on available current and voltage measurements. Additionally, this system is contextualized within a larger, modular Master State awareness Overseer (MSO) framework, responsible for monitoring, analyzing, and managing an electric grid.

24 - POWER TRANSMISSION AND DISTRIBUTION↗

Measurements of air contaminants during the Cerro Grande fire at Los Alamos National Laboratory

Ambient air sampling for radioactive air contaminants was continued throughout the Cerro Grande fire that burned part of Los Alamos National Laboratory. During the fire, samples were collected more frequently than normal because buildup of smoke particles on the filters was decreasing the air flow. Overall, actual sampling time was 96% of the total possible sampling time for the May 2000 samples. To evaluate potential human exposure to air contaminants, the samples were analyzed as soon as possible and for additional specific radionuclides. Analyses showed that the smoke from the fire included resuspended radon decay products that had been accumulating for many years on the vegetation and the forest floor that burned. Concentrations of plutonium, americium, and depleted uranium were also measurable, but at locations and concentrations comparable to non-fire periods. A continuous particulate matter sampler measured concentrations that exceeded the National Ambient Air Quality Standard for PM-10 (particles less than 10 micrometers in diameter). These high concentrations were caused by smoke from the fire when it was close to the sampler.

54 ENVIRONMENTAL SCIENCES↗

EO-1 Hyperion Reflectance Time Series at Calibration and Validation Sites: Stability and Sensitivity to Seasonal Dynamics

This study evaluated Earth Observing 1 (EO-1) Hyperion reflectance time series at established calibration sites to assess the instrument stability and suitability for monitoring vegetation functional parameters. Our analysis using three pseudo-invariant calibration sites in North America indicated that the reflectance time series are devoid of apparent spectral trends and their stability consistently is within 2.5-5 percent throughout most of the spectral range spanning the 12-plus year data record. Using three vegetated sites instrumented with eddy covariance towers, the Hyperion reflectance time series were evaluated for their ability to determine important variables of ecosystem function. A number of narrowband and derivative vegetation indices (VI) closely described the seasonal profiles in vegetation function and ecosystem carbon exchange (e.g., net and gross ecosystem productivity) in three very different ecosystems, including a hardwood forest and tallgrass prairie in North America, and a Miombo woodland in Africa. Our results demonstrate the potential for scaling the carbon flux tower measurements to local and regional landscape levels. The VIs with stronger relationships to the CO2 parameters were derived using continuous reflectance spectra and included wavelengths associated with chlorophyll content and/or chlorophyll fluorescence. Since these indices cannot be calculated from broadband multispectral instrument data, the opportunity to exploit these spectrometer-based VIs in the future will depend on the launch of satellites such as EnMAP and HyspIRI. This study highlights the practical utility of space-borne spectrometers for characterization of the spectral stability and uniformity of the calibration sites in support of sensor cross-comparisons, and demonstrates the potential of narrowband VIs to track and spatially extend ecosystem functional status as well as carbon processes measured at flux towers.

Calibration↗

Fine particulate concentrations over East Asia derived from aerosols measured by the Advanced Himawari Imager using machine learning

Fine particulate matter with a diameter below 2.5 μm (PM 2.5 ) is deleterious to the cardiovascular and respiratory systems. It is often difficult to assess the effects of PM 2.5 on human health over regions with limited ground monitoring sites, especially in East Asia. As an alternative, we estimated near-surface PM 2.5 concentrations by analyzing Advanced Himawari Imager (AHI) Yonsei Aerosol Retrieval (YAER) products. This study incorporates daytime data for East Asia covering the Korean Peninsula, China, Japan, Southeast Asia, and southern Mongolia. We collocated AHI YAER product pixels with meteorological, land-cover, and other ancillary data for the period from March 2018 to February 2019. To estimate PM 2.5 concentrations over wide areas spanning many countries displaying various relationships between aerosol optical depth and PM 2.5 , monthly models were developed by considering both the spatial and temporal characteristics of ground-based PM 2.5 measurements. Random forest machine learning model estimated ground-level mass concentrations of PM 2.5 ; subsequent 10-fold cross validation (CV) yielded a CV R 2 value of 0.81 and a CV root mean squared error (RMSE) of 12.3 μg m -3 . We investigated the spatial pattern of PM 2.5 concentrations over multiple countries and seasonal variation in PM 2.5 concentrations. Diurnal variation of a severe PM 2.5 event in the Korean Peninsula was investigated as a case study. The model captured the extremely heterogeneous spatial distribution of PM 2.5 concentrations peaked around local noon. To measure the capability of the developed model to estimate PM 2.5 concentrations in areas with few in-situ data, its predictive performance was evaluated using a dataset independent of the training process with an R 2 of 0.60 and RMSE of 8.18 μg m −3 . This study demonstrates the potential for satellite-based PM 2.5 estimation for areas with insufficient measuring stations.

Pm2.5↗

Predicting Initial Trans-Membrane Pressure for Optimized Operations in UF Unit Using Random Forest

With the growing scarcity of freshwater, innovative process design mechanisms like Reverse Osmosis (RO) are increasingly gaining attention among water treatment utilities to address the rising demand. Ensuring reliable water production necessitates efficient resource utilization, minimizing downtime in (ultra-filtration) UF systems. Recent advancements in machine learning (ML) have enabled the development of accurate data-driven models for Model Predictive Control (MPC), often requiring minimal prior knowledge of underlying physical processes. In this study, we present predictive regression models based on Random Forest (RF) and Auto-Regressive (AR) approaches to forecast the initial Trans-Membrane Pressure (TMP) for each filtration cycle in data generated by Direct Potable Reuse (DPR) systems. The proposed RF-based model demonstrates superior performance compared to baseline methods, including historical mean, Last Observation Carried Forward (LOCF), and naïve AR models, across various forecasting horizons in terms of root mean square error (RMSE) metric. To evaluate how different classes of process variables contribute to TMP dynamics over time, we examine the feature importance of independent covariates across multiple forecast horizons. This analysis provides insight into the temporal relevance of operational and sensor-derived features, guiding control and monitoring strategies. Additionally, the impact of hyperparameter tuning on TMP prediction performance is studied for both direct and recursive RF modelling approaches across increasing forecast horizons. Accurate prediction of initial TMP is critical for optimizing RO operations, as it enables the development of robust modelling frameworks by accurately estimating membrane fouling trends, thereby enhancing process efficiency and long-term reliability. The demonstrated efficacy of the RF-based approach highlights its potential as a tool for real-time decision-making in water treatment systems, paving the way for advanced process optimization and sustainable water resource management.

Mukherjee, Subrata [ORNL] (ORCID:0000000309930338)↗

A Collaborative Decision Environment for UAV Operations

NASA is developing Intelligent Mission Management (IMM) technology for science missions employing long endurance unmanned aerial vehicles (UAV's). The IMM groundbased component is the Collaborative Decision Environment (CDE), a ground system that provides the Mission/Science team with situational awareness, collaboration, and decisionmaking tools. The CDE is used for pre-flight planning, mission monitoring, and visualization of acquired data. It integrates external data products used for planning and executing a mission, such as weather, satellite data products, and topographic maps by leveraging established and emerging Open Geospatial Consortium (OGC) standards to acquire external data products via the Internet, and an industry standard geographic information system (GIs) toolkit for visualization As a Science/Mission team may be geographically dispersed, the CDE is capable of providing access to remote users across wide area networks using Web Services technology. A prototype CDE is being developed for an instrument checkout flight on a manned aircraft in the fall of 2005, in preparation for a full deployment in support of the US Forest Service and NASA Ames Western States Fire Mission in 2006.

D'Ortenzio, Matthew V.↗

Gas Concentration Mapping of Arenal Volcano Using AVEMS

The Airborne Volcanic Emissions Mass Spectrometer (AVEMS) System developed by NASA-Kennedy Space Center and deployed in collaboration with the National Center for Advanced Technology (CENAT) and the University of Costa Rica was used for mapping the volcanic plume of Arenal Volcano, the most active volcano in Costa Rica. The measurements were conducted as part of the second CARTA (Costa Rica Airborne Research and Technology Application) mission conducted in March 2005. The CARTA 2005 mission, involving multiple sensors and agencies, consisted of three different planes collecting data over all of Costa Rica. The WB-57F from NASA collected ground data with a digital camera, an analog photogrametric camera (RC-30), a multispectral scanner (MASTER) and a hyperspectral sensor (HYMAP). The second aircraft, a King Air 200 from DoE, mounted with a LIDAR based instrument, targeted topography mapping and forest density measurements. A smaller third aircraft, a Navajo from Costa Rica, integrated with the AVEMS instrument and designed for real-time measurements of air pollutants from both natural and anthropogenic sources, was flown over the volcanoes. The improved AVEMS system is designed for deployment via aircraft, car or hand-transport. The 85 pound system employs a 200 Da quadrupole mass analyzer, has a volume of 92,000 cubic cm, requires 350 W of power at steady state, can operate up to an altitude of 41,000 feet above sea level (-65 C; 50 torr). The system uses on-board gas bottles on-site calibration and is capable of monitoring and quantifying up to 16 gases simultaneously. The in-situ gas data in this work, consisting of helium, carbon dioxide, sulfur dioxide and acetone, was acquired in conjunction of GPS data which was plotted with the ground imagery, topography and remote sensing data collected by the other instruments, allowing the 3 dimensional visualization of the volcanic plume at Arenal Volcano. The modeling of possible scenarios of Arenal s activity and its direct impact on the surrounding populated areas in now possible with the combined set of data, linking in-situ data with remote sensing data. The study also helps in the understanding of pyroclastic flow behavior in case of a major eruption.

Diaz, J. Andres↗

Mechanistic and predictive understanding of needle litter decay in semi-arid mountain ecosystems experiencing unprecedented vegetation mortality. Final report

This research compared needle litter decay rates and underlying mechanisms between litter released from insect-killed trees to that of naturally senesced litter from healthy trees across climate variables with elevation and snowmelt rates as variables. Our research was designed to address questions of whether there is a significant impact to C and N cycling at the terrestrial-atmospheric interface in association with pine needle decomposition and the role of changes in soil moisture and temperature as a function of these processes. Field activities were primarily conducted within the Rocky Mountain Biological Laboratory’s (RMBL) East River watershed across a spectrum of ecoregions that spanned approximately 700 vertical meters. Forested catchments at these elevations in southwestern Colorado are threatened by heat-induced drought, altered snowmelt patterns, and invasive insect infestation that have all been associated with climate change. Needle litter from two members of the Pinaceae family experiencing large-scale insect infestations (spruce and lodgepole pine) were collected, redeployed and then sampled over the course of three years. We experimentally tested three tractable parameters that are known to impact litter decay and biogeochemical cycling and are relevant to bark beetle-induced tree mortality: (1) moisture content, (2) temperature by using elevation as a surrogate, and (3) changes to litter compositional chemistry that result from beetle infestation. We examined decay processes and monitored the migration of soluble and gaseous nitrogen and carbon species as well as microbiological signatures of decay such as biodegradable fractions, metal availability and the temporal succession of associated microorganisms at the terrestrial interface.

54 ENVIRONMENTAL SCIENCES↗

Application of LANDSAT to the surveillance of lake eutrophication in the Great Lakes basin

The author has identified the following significant results. A step-by-step procedure for establishing and monitoring the trophic status of inland lakes with the use of LANDSAT data, surface sampling, laboratory analysis, and aerial observations were demonstrated. The biomass was related to chlorophyll-a concentrations, water clarity, and trophic state. A procedure was developed for using surface sampling, LANDSAT data, and linear regression equations to produce a color-coded image of large lakes showing the distribution and concentrations of water quality parameters, causing eutrophication as well as parameters which indicate its effects. Cover categories readily derived from LANDSAT were those for which loading rates were available and were known to have major effects on the quality and quantity of runoff and lake eutrophication. Urban, barren land, cropland, grassland, forest, wetlands, and water were included.

Rogers, R. H.↗

Feasibility of Estimating Relative Nutrient Contributions of Agriculture and Forests Using MODIS Time Series

Around the Gulf of Mexico, high-input crops in several regions make a significant contribution to nutrient loading of small to medium estuaries and to the near-shore Gulf. Some crops cultivated near the coast include sorghum in Texas, rice in Texas and Louisiana, sugarcane in Florida and Louisiana, citrus orchards in Florida, pecan orchards in Mississippi and Alabama, and heavy sod and ornamental production around Mobile and Tampa Bay. In addition to crops, management of timberlands in proximity to the coasts also plays a role in nutrient loading. In the summer of 2008, a feasibility project is planned to explore the use of NASA data to enhance the spatial and temporal resolution of near-coast nutrient source information available to the coastal community. The purpose of this project is to demonstrate the viability of nutrient source information products applicable to small to medium watersheds surrounding the Gulf of Mexico. Conceptually, these products are intended to complement estuarine nutrient monitoring.

Ross, Kenton W.↗

Multi-scale integration of satellite remote sensing improves characterization of dry-season green-up in an Amazon tropical evergreen forest

In tropical forests, leaf phenology-particularly the pronounced dry-season green-up-strongly regulates biogeochemical cycles of carbon and water fluxes. However, uncertainties remain in the understanding of tropical forest leaf phenology at different spatial scales. Phenocams accurately characterize leaf phenology at the crown and ecosystem scales but are limited to a few sites and time spans of a few years. Time-series satellite observations might fill this gap, but the commonly used satellites (e.g. MODIS, Landsat and Sentinel-2) have resolutions too coarse to characterize single crowns. To resolve this observational challenge, we used the PlanetScope constellation with a 3m resolution and near daily nadir-view coverage. We first developed a rigorous method to cross-calibrate PlanetScope surface reflectance using daily BRDF-adjusted MODIS as the reference. We then used linear spectral unmixing of calibrated PlanetScope to obtain dry-season change in the fractional cover of green vegetation (GV) and non-photosynthetic vegetation (NPV) at the PlanetScope pixel level. We used the Central Amazon Tapajos National Forest k67 site, as all necessary data (from field to phenocam and satellite observations) was available. For this proof of concept, we chose a set of 22 dates of PlanetScope measurements in 2018 and 16 in 2019, all from the six drier months of the year to provide the highest possible cloud-free temporal resolution. Our results show that MODIS-calibrated dry-season PlanetScope data (1) accurately assessed seasonal changes in ecosystem-scale and crown-scale spectral reflectance; (2) detected an increase in ecosystem-scale GV fraction (and a decrease in NPV fraction) from June to November of both years, consistent with local phenocam observations with R 2 around 0.8; and (3) monitored large seasonal trend variability in crown-scale NPV fraction. Finally, our results highlight the potential of integrating multi-scale satellite observations to extend fine-scale leaf phenology monitoring beyond the spatial limits of phenocams.

54 ENVIRONMENTAL SCIENCES↗

Increased salinity decreases annual gross primary productivity at a Northern California brackish tidal marsh

Tidal marshes sequester 11.4–87.0 Tg C yr –1 globally, but climate change impacts can threaten the carbon capture potential of these ecosystems. Tidal marshes occur across a wide range of salinity, with brackish marshes (0.5–18 ppt (parts per thousand)) dominating global tidal marsh extents. A diverse mix of freshwater- and saltwater-tolerant plant and microbial communities has led researchers to predict that carbon cycling in brackish wetlands may be less sensitive to changes in salinity than fresh- or saltwater wetlands. Rush Ranch, a well-monitored brackish tidal wetland of the San Francisco Bay National Estuarine Research Reserve, experiences highly variable annual salinity regimes. Within a five-year period (2014–2018), Rush Ranch experienced particularly extreme drought-induced salinization during the 2014 and 2015 growing seasons. During drought years, tidal channel salinity rose from a 15 year baseline of 4.7 ppt to growing season peaks of 10.3 ppt and 12.5 ppt. Continuous eddy covariance data from 2014 to 2018 demonstrate that during drought summers, gross primary productivity (GPP) decreased by 24%, whereas ecosystem respiration remained similar among all five years. Stepwise linear regression revealed that salinity, not air temperature or tidal height, was the dominant driver of annual GPP. A random forest model trained to predict GPP based on environmental data from low salinity years (i.e. naive to salinization) significantly over predicted GPP in drought years. When growing season salinities were doubled, annual estimates of net ecosystem exchange of CO 2 decreased by up to 30%. These results provide ecosystem-scale evidence that increased salinity influences CO 2 fluxes dominantly through reductions in GPP. This relationship provides a starting point for incorporating the effect of changes in salinity in wetland carbon models, which could improve wetland carbon forecasting and management for climate resilience.

54 ENVIRONMENTAL SCIENCES↗

Model Effects on GLAS-Based Regional Estimates of Forest Biomass and Carbon

ICESat/GLAS waveform data are used to estimate biomass and carbon on a 1.27 million sq km study area. the Province of Quebec, Canada, below treeline. The same input data sets and sampling design are used in conjunction with four different predictive models to estimate total aboveground dry forest biomass and forest carbon. The four models include nonstratified and stratified versions of a multiple linear model where either biomass or (square root of) biomass serves as the dependent variable. The use of different models in Quebec introduces differences in Provincial biomass estimates of up to 0.35 Gt (range 4.942+/-0.28 Gt to 5.29+/-0.36 Gt). The results suggest that if different predictive models are used to estimate regional carbon stocks in different epochs, e.g., y2005, y2015, one might mistakenly infer an apparent aboveground carbon "change" of, in this case, 0.18 Gt, or approximately 7% of the aboveground carbon in Quebec, due solely to the use of different predictive models. These findings argue for model consistency in future, LiDAR-based carbon monitoring programs. Regional biomass estimates from the four GLAS models are compared to ground estimates derived from an extensive network of 16,814 ground plots located in southern Quebec. Stratified models proved to be more accurate and precise than either of the two nonstratified models tested.

Nelson, Ross↗

Application of Buckmaster Electrolyte Ion Leakage Test to Woody Biofuel Feedstocks

In an earlier ASABE paper, Buckmaster reported that ion conductivity of biomass leachate in aqueous solution was directly correlated with activity access to plant nutrients within the biomass materials for subsequent biological or chemical processing. The Buckmaster test involves placing a sample of the particles in a beaker of constant-temperature deionized water and monitoring the change in electrical conductivity over time. We adapted the Buckmaster method to a range of woody biomass and other cellulosic bioenergy feedstocks. Our experimental results suggest differences of electrolyte leakage between differently processed woody biomass particles may be an indicator of their utility for conversion in bioenergy processes. This simple assay appears to be particularly useful to compare different biomass comminution techniques and particle sizes for biochemical preprocessing.

09 BIOMASS FUELS↗

Recurrent Neural Network Language Models for Open Vocabulary Event-Level Cyber Anomaly Detection

Automated analysis methods are crucial aids for monitoring and defending a network to protect the sensitive or confidential data it hosts. This work introduces a flexible, powerful, and unsupervised approach to detecting anomalous behavior in computer and network logs; one that largely eliminates domain-dependent feature engineering employed by existing methods. By treating system logs as threads of interleaved ``sentences'' (event log lines) to train online unsupervised neural network language models, our approach provides an adaptive model of normal network behavior. We compare the effectiveness of both standard and bidirectional recurrent neural network language models at detecting malicious activity within network log data. Extending these models, we introduce a tiered recurrent architecture, which provides context by modeling sequences of users' actions over time. Compared to isolation forests and Principal Components Analysis (PCA), two popular anomaly detection algorithms, we observe superior performance on the Los Alamos National Laboratory Cyber Security dataset. For log-line-level predictions, our best performing character-based model provides test set anomaly scores ranking red team events in the 98th percentile on average, and yields 0.97 area under the receiver operator characteristic curve, demonstrating the strong fine-grained anomaly detection performance of this approach on open vocabulary logging sources.

Cyber Analysis, Deep Learning, language analysis, ↗