Search NASA⌕ Search

SEARCH · Search NASA

Results for “evaluation datasets”

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 37 records · Page 2

Toward an AI-Powered Software Pipeline for Real-Time Tracking and Analysis of Wildfire and Smoke

Real-time tracking of wildfires and smoke is crucial for effective response, minimizing damage, protecting lives, and efficiently managing resources during fire emergencies. We develop a web-based AI-powered pipeline that detects wildfires in aerial video and estimates deployment-relevant behavior metrics, including cumulative burned area, burned-area growth rate, fire spread direction, and smoke dispersion. The system combines a YOLO-based detector with YCbCr-based fire segmentation, HSV-based smoke segmentation, Farneback optical flow, and centroid-based spatiotemporal tracking. Using ground sampling distance (GSD), pixel-level fire masks are converted to physical burned-area measurements by correlating fire pixel counts with camera altitude and tilt angle. We benchmark YOLO variants and non-YOLO baselines (GoogLeNet, CNN, DBN, Autoencoder, U-Net, and AlexNet) on the IEEE FLAME dataset and a newly created aerial frame dataset, Wildfire-DB. Cross-dataset evaluation uses a strict threshold-transfer protocol: decision thresholds are selected on FLAME validation and transferred unchanged to Wildfire-DB to quantify generalization under domain shift. YOLOv6 achieves the strongest cross-dataset frame-level fire detection on Wildfire-DB (ROC-AUC 0.8200, PR-AUC 0.8044, and transferred-threshold F1 0.7596). For tracking-oriented deployment requiring oriented localization, YOLO11-OBB provides the most reliable cross-dataset behavior among OBB-capable models while remaining computationally feasible. To analyze the feasibility of UAV deployment, we further measure inference efficiency using synchronized GPU and CPU power logs on a fixed workload of 1569 frames. YOLO-family models process the video in 5.73–12.47 seconds with net energy of 1247.28–1775.39 J, substantially lower latency and energy than heavier classification and reconstruction baselines. Overall, model optimality depends on operational objectives: YOLOv6 is best for cross-dataset detection robustness, whereas YOL...

Color segmentation↗

A Pattern Based Algorithmic Autotuner for Graph Processing on GPUs

This paper presents Gswitch, a pattern-based algorithmic autotuning system that dynamically switches to the suitable optimization variants with negligible overhead. Its novelty is a small set of algorithmic patterns that enables configurable assembling of algorithm variants. The fast transition of Gswitch is based on a machine learning model trained from 644 real graphs from the network repository. In addition, Gswitch provides succinct programming interface which hides all low-level tuning details. We evaluate Gswitch for typical graph algorithms (BFS, CC, PR, SSSP, and BC) on Nvidia Kepler and Pascal GPUs. The results show that Gswitch runs up to 10× faster than the best configuration of the state-of-the-art programmable GPU-based graph processing libraries on ten representative graphs. Gswitch wins on 92.4% cases of 644 graph data which is the largest dataset evaluation reported to date.

HPC, graph algorithms, GPUs, auto tuning↗

Supporting ARPA-E Power Grid Optimization (Final Report)

Pacific Northwest National Laboratory (PNNL), Arizona State University (ASU), Georgia Institute of Technology (Georgia Tech), Los Alamos National Laboratory (LANL), National Renewable Energy Laboratory (NREL), Texas A&M University (TAMU), The University of Texas at Austin (UT), and the University of Wisconsin-Madison (UW-M) supported the ARPA-E Grid Optimization (GO) Competition by providing a common problem formulation, data format, datasets, evaluation mechanism, scoring, rules, and results that resulted in the awarding of $\$9.24$ million dollars to teams from academia, industry, and national labs for solving three sets of increasingly difficult non-linear, security- constrained AC Optimal Powerflow (AC-OPF) optimization problems in order to increase the efficiency of the US Electric Grid. It is estimated that a 1% increase in efficiency can save $\$1$ billion. Current industry practices typically use a linear DC model (DC-OPF) in order solve the OPF problem within the time constraints of the operation schedule. The GO Competition challenges the best power engineers, mathematicians, and computer scientists to make possible operational decisions based on accurate physical models. To accomplish this, the GO Competition created a series of Challenges and funded teams to produce the best solver. Challenge 1 was to solve the security constrained Alternating Current Optimal Power Flow (ACOPF) problem. Challenge 2 extended that to by adding adjustable transformer tap ratios, phase shifting transformers, switchable shunts, price-responsive demand, ramp rate constrained generators and loads, and fast-start unit commitment (UC). Furthermore, Challenge 2 was a maximization problem while Challenge 1 was a minimization problem. While Challenge 3 was being developed, the entrants were invited to find better solutions to the Challenge 2 synthetic datasets with no restrictions on time, hardware, or algorithms. The Challenge 2 solutions turned out to be very good. Challenge 3 expanded the Challenge 2 problem further by using multiperiod dynamic markets, including advisory models for extreme weather events, day-ahead markets, and the real-time markets with an extended look-ahead. These problems included active bid-in demand and topology optimization. Together the Challenges used nearly 30 million CPU hours. Since each team was working on the same problem, using the same data, and running on the same hardware, fair comparisons could be drawn as to the best solver. The datasets were varied enough, however, that the best solver for one dataset was not necessarily the best at another, so cumulative scores were used. The process was managed by the PNNL maintained website https://GOCompetition.energy.gov, where Entrants could find information about the problem, the data, the rules, submit their solver for evaluation, and see the scores of all the competing teams on a Leaderboard. Interest was world-wide but only American teams were eligible for prizes. The Competition has produced 34 journal articles 115 papers and been cited over 500 times in the literature, including 12 dissertations (4 from foreign countries; Columbia (2), Germany, and Italy) and 3 from the DOE ExaScale project. Software developed by Pearl Street Technologies for Challenges 1 and 2 is now deployed by Southwest Power Pool (SPP) and Midcontinent Independent Service Operator (MISO). Other teams have received inquiries from venture capitalists. Google DeepMind has thanked the Competition for making the datasets developed for the Competition public. They are using it to train machine learning models. The larger datasets have billions of unknowns to be solved for, but only a small percent matter in the final solution. Knowing what unknowns are important can dramatically speedup the solution.

24 POWER TRANSMISSION AND DISTRIBUTION↗

SPRUCE Representing Northern Peatland Microtopography and Hydrology within the Community Land Model: Modeling Archive

Here we provide model code, inputs, outputs and evaluation datasets for a new configuration of the Community Land Model (CLM) for SPRUCE, which includes a fully prognostic water table calculation for SPRUCE. Our structural and process changes to CLM focus on modifications needed to represent the hydrologic cycle of bogs environment with perched water tables, as well as distinct hydrologic dynamics and vegetation communities of the raised hummock and sunken hollow microtopography characteristic of SPRUCE and other peatland bogs. The modified model was parameterized and independently evaluated against observations from an ombrotrophic raised-dome bog in northern Minnesota (S1-Bog), the site for the Spruce and Peatland Responses Under Climatic and Environmental Change experiment (SPRUCE).

54 ENVIRONMENTAL SCIENCES↗

Comparison of temporal resolution selection approaches in energy systems models

Capacity expansion models for the power sector are used to project future decisions over the coming decades by simulating investment and operation decisions for the use of electricity. Due to model performance constraints, these models typically do not explicitly simulate every hour within a year, but instead simulate representative time segments (groups of hours). This paper evaluates different approaches for selecting time segments across three methods: sequential, categorical, and clustering, across a wide range of time-segment quantities, for a total of 204 temporal profiles. To measure the performance of each profile's ability to accurately represent data, the root-mean-square-error of each profile's time segments are compared to the data's original hourly data. The temporal alignment across regions is also measured (i.e., how often windy days align across regions). Different spatial resolutions were applied for a subset of the temporal selection methods to investigate the impact spatial resolution has on performance. This paper provides a framework for measuring the value of different temporal selection methods and of adding more granular data to energy system models. Overall, multi-criteria clustering yields the lowest root-mean-square-error across all datasets evaluated and provides a holistic view of the intertwined relationships between renewable generation and electricity demand.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Solubility of U(VI) in chloride solutions. III. The stable oxides/hydroxides in MgCl 2 systems: Pitzer activity model for the system UO 2 2+ –Na + –K + –Mg 2+ –H + –OH – –Cl – –H 2 O(l)

Here we have developed new chemical, thermodynamic and activity models for the system UO 2 2+ –Na + –K + –Mg 2+ –H + –OH – –Cl – –H 2 O(l) within the Pitzer approach. The new thermodynamic model is based on previously reported data treated within the SIT approach for NaCl and KCl systems, as well as on new experimental data determined in this work for the MgCl 2 system. The solubility of uranium(VI) was studied in 0.01–5.15 mol·kg w –1 MgCl 2 solutions at pH m = 4.1–9.7 (with pH m = –log [H + ]). Experiments were performed under Ar atmosphere at T = (22 ± 2) °C. In all investigated systems, the solubility of U(VI) is controlled by metaschoepite, UO 3 ·2H 2 O(cr). In contrast to previously investigated NaCl and KCl systems, no ternary Mg–U(VI)–OH(s) solid phases formed in alkaline MgCl 2 solutions within the timeframe of this study (≤ 200 days). A very significant increase in the solubility (up to 3 log 10 -units) is observed in acidic to near-neutral pH m conditions when increasing MgCl 2 concentration from 0.01 to 5.15 mol·kg w –1 , which reflects the strong ion interaction processes taking place in concentrated MgCl 2 brines. The solubility of UO 3 ·2H 2 O(cr) in the investigated NaCl, KCl and MgCl 2 solutions is well described with the solubility and hydrolysis constants recommended by Altmaier et al., (2017) and NEA–TDB, and a Pitzer activity model derived in the present work. The latter model considers experimental data reported in the present study and available in the literature for NaCl, KCl and MgCl 2 systems (solubility, potentiometric and spectroscopic data), in combination with well-stablished estimation methods and correlations with SIT coefficients. Finally, chemical, thermodynamic and Pitzer activity models provided in this work for the system UO 2 2+ –Na + –K + –Mg 2+ –H + –OH – –Cl – –H 2 O(l) accurately describe all evaluated datasets, and represent an adequate tool for the calculation of U(VI) solubility and aqueous speciation in a variety of geochemical conditions including concentrated brine systems of relevance in salt-based repositories for nuclear waste disposal.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Validation of the Community Land Model Version 5 over the Contiguous United States (CONUS) using in situ and remote sensing data sets

The Community Land Model (CLM) is an effective tool to simulate the biophysical and biogeochemical processes and their interactions with the atmosphere. Although CLM Version 5 (CLM5) constitutes various updates in these processes, its performance in simulating energy, water and carbon cycles over the Contiguous United States (CONUS) at scales which land surface changes and hydrometeorological and hydroclimatological applications are more locally relevant is yet to be assessed. In this study, we conducted three simulations at 0.125? during 1979-2018 over the CONUS using different configurations of CLM, namely CLM5-biogeochemistry (CLM5BGC), CLM4.5BGC, and CLM5-satellite phenology (CLM5SP). We validated and compared their simulations against multiple remote-sensed and in-situ datasets. Overall, the parametric and structural updates (e.g., carbon cost for nitrogen uptake, variable soil thickness, dry surface layer) in CLM5 improve its ability in capturing terrestrial biogeochemical dynamics. The low evapotranspiration in CLM5BGC is associated with biases in simulating vegetation phenological characteristics rather than soil water limitations. The mismatch between CLM5BGC-simulated peak leaf area index and reference data can be attributed to CLM5BGC's inability in simulating phenology of trees and grasses. The differences between CLM-simulated irrigation and reference estimates can be attributed to differences between processes represented in models and in reality, and uncertainties in input and validation datasets. Evaluation against observations at small catchments suggest that hydrologic parameters needed to be calibrated to improve simulations of runoff, especially subsurface runoff. Additional efforts are needed to incorporate spatially-distributed plant phenology and physiology parameters and regional-specific agricultural management practices (e.g., planting, harvest).

Cheng, Yanyan↗

Unveiling four decades of intensifying precipitation from tropical cyclones using satellite measurements

Increases in precipitation rates and volumes from tropical cyclones (TCs) caused by anthropogenic warming are predicted by climate modeling studies and have been identified in several high intensity storms occurring over the last half decade. However, it has been difficult to detect historical trends in TC precipitation at time scales long enough to overcome natural climate variability because of limitations in existing precipitation observations. We introduce an experimental global high-resolution climate data record of precipitation produced using infrared satellite imagery and corrected at the monthly scale by a gauge-derived product that shows generally good performance during two hurricane case studies but estimates higher mean precipitation rates in the tropics than the evaluation datasets. General increases in mean and extreme rainfall rates during the study period of 1980–2019 are identified, culminating in a 12–18%/40-year increase in global rainfall rates. Overall, all basins have experienced intensification in precipitation rates. Increases in rainfall rates have boosted the mean precipitation volume of global TCs by 7–15% over 40 years, with the starkest rises seen in the North Atlantic, South Indian, and South Pacific basins (maximum 59–64% over 40 years). In terms of inland rainfall totals, year-by-year trends are generally positive due to increasing TC frequency, slower decay over land, and more intense rainfall, with an alarming increase of 81–85% seen from the strongest global TCs. As the global trend in precipitation rates follows expectations from warming sea surface temperatures (11.1%/°C), we hypothesize that the observed trends could be a result of anthropogenic warming creating greater concentrations of water vapor in the atmosphere, though retrospective studies of TC dynamics over the period are needed to confirm.

54 ENVIRONMENTAL SCIENCES↗

Methods for Determining Likelihood of Tweet Deletion

Few works exist that attempt to build predictive models for tweet deletion. Zhou et al. (2015) focus on a subset of deleted tweets – regrettable tweets. These are tweets that the authors believe to contain inappropriate content. Inappropriate can range from vulgar language to sharing private content such as a personal email address. The presence of inappropriate content doesn’t guarantee that a tweet will be deleted, however intuition dictates it can be in an important factor in the tweet being deleted. In their work, the authors create a predictive model for identifying regrettable tweets. It is important to note the authors focus on predicting regrettable tweets that are distinctly not spam and only written in English. Through manual investigation, the authors identify ten major topics including negative sentiment, cursing, and relationships that are prevalent in regrettable tweets. The authors then exploit WordNet and UrbanDictionary to create keyword lists related to the ten topics. Finally, using a combination of existing lexica and the topic keywords as features, the authors build classifiers to test the accuracy of their model. The authors complement 700 manually labeled regrettable tweets with 700 normal tweets to create their evaluation dataset. The authors’ best performance from 10-fold cross-validation was an f1 score of 0.85 using a J48 classifier on a balanced dataset of deleted and non-deleted tweets.

Source record↗

Validation of the Community Land Model Version 5 over the Contiguous United States (CONUS) using in-situ and remote sensing datasets

The Community Land Model (CLM) is an effective tool to simulate the biophysical and biogeochemical processes and their interactions with the atmosphere. Although CLM Version 5 (CLM5) constitutes various updates in these processes, its performance in simulating energy, water and carbon cycles over the Contiguous United States (CONUS) at scales which land surface changes and hydrometeorological and hydroclimatological applications are more locally relevant is yet to be assessed. In this study, we conducted three simulations at 0.125? during 1979-2018 over the CONUS using different configurations of CLM, namely CLM5-biogeochemistry (CLM5BGC), CLM4.5BGC, and CLM5-satellite phenology (CLM5SP). We validated and compared their simulations against multiple remote-sensed and in-situ datasets. Overall, the parametric and structural updates (e.g., carbon cost for nitrogen uptake, variable soil thickness, dry surface layer) in CLM5 improve its ability in capturing terrestrial biogeochemical dynamics. The low evapotranspiration in CLM5BGC is associated with biases in simulating vegetation phenological characteristics rather than soil water limitations. The mismatch between CLM5BGC-simulated peak leaf area index and reference data can be attributed to CLM5BGC's inability in simulating phenology of trees and grasses. The differences between CLM-simulated irrigation and reference estimates can be attributed to differences between processes represented in models and in reality, and uncertainties in input and validation datasets. Evaluation against observations at small catchments suggest that hydrologic parameters needed to be calibrated to improve simulations of runoff, especially subsurface runoff. Additional efforts are needed to incorporate spatially-distributed plant phenology and physiology parameters and regional-specific agricultural management practices (e.g., planting, harvest).

54 ENVIRONMENTAL SCIENCES↗

Livewire: A Model Platform for Data Quality Assessment and AI Readiness Across DOE Missions

High-quality, well-governed data is essential for accelerating discovery and achieving operational excellence across DOE and national laboratory missions. The Livewire Data Platform is a DOE-supported platform that offers automated assessments of data quality, standardization, provenance, and Artificial Intelligence (AI) readiness. It allows researchers and data practitioners to systematically and easily evaluate datasets against established governance criteria and prepare them for advanced analytics. Livewire addresses critical challenges in DOE's data ecosystem with integrated capabilities for metadata validation, provenance tracking, and schema alignment. This platform's automated workflows assist users in identifying data quality gaps, enhancing interoperability between datasets collected from various stakeholders, and ensuring compliance with DOE data standards, all while reducing manual curation efforts. Additionally, we will discuss its AI readiness framework, which is being developed to prepare datasets for training models, developing advanced analytic tools, and machine learning applications. Using some of the more than one hundred tabular datasets on Livewire, processed with this open-source methodology, we will demonstrate how Livewire can serve as a model for scalable, standards-driven data management. This approach provides a pathway to leverage existing and future datasets within the DOE, boosting innovation and efficiency across national laboratories.

33 - ADVANCED PROPULSION SYSTEMS↗

Application and Evaluation of the China Meteorological Assimilation Driving Datasets for the SWAT Model (CMADS) in Poorly Gauged Regions in Western China

The temporal and spatial differentiation of the underlying surface in East Asia is complex. Due to a lack of meteorological observation data, human cognition and understanding of the surface processes (runoff, snowmelt, soil moisture, water production, etc.) in the area have been greatly limited. With the Heihe River Basin, a poorly gauged region in the cold region of Western China, selected as the study area, three meteorological datasets are evaluated for their suitability to drive the Soil and Water Assessment Tool (SWAT): China Meteorological Assimilation Driving Datasets for the SWAT model (CMADS), Climate Forecast System Reanalysis (CFSR), and Traditional Weather Station (TWS). Resultingly, (1) the runoff output of CMADS + SWAT mode is generally better than that of the other two modes (CFSR + SWAT and TWS + SWAT) and the monthly and daily Nash–Sutcliffe efficiency ranges of the CMADS + SWAT mode are 0.75–0.95 and 0.58–0.77, respectively; (2) the CMADS + SWAT and TWS + SWAT results were fairly similar to the actual data (especially for precipitation and evaporation), with the results produced by CMADS + SWAT lower than those produced by TWS + SWAT; (3) the CMADS + SWAT mode has a greater ability to reproduce water balance than the other two modes. Overestimation of CFSR precipitation results in greater error impact on the uncertainty output of the model, whereas the performances of CMADS and TWS are more similar. This study addresses the gap in the study of surface processes by CMADS users in Western China and provides an important scientific basis for analyzing poorly gauged regions in East Asia.

CMADS, SWAT, poorly gauged regions↗

Evaluation of linear interpolation method for missing value on solar radiation dataset in Perlis

This paper intends to reveal the ability of the linear interpolation method to predict missing values in solar radiation time series. Reliable dataset is equally tends to complete time series observed dataset. The absence or presence of radiation data alters long-term variation of solar radiation measurement values. Based on that change, the opportunities to provide bias output result for modelling and the validation process is higher. The completeness of the observed variable dataset has significantly important for data analysis. Occurrence the lack of continual and unreliable time series solar radiation data widely spread and become the main problematic issue. However, the limited number of research quantity that has carried out to emphasize and gives full attention to estimate missing values in the solar radiation dataset.

42 ENGINEERING↗

EI_MS_ML

The unambiguous identification of compounds from their electron ionization mass (EI-MS) spectra remains a significant unsolved problem in the field of metabolomics and analytical chemistry as a whole. Typically EI-MS spectra are compared using various mathematical operations that convert the spectral similarity or differences into a distance-like metric that roughly approximates the similarity of any two spectra. A commonly used metric for this is the cosine similarity metric which has values close to one for very similar spectra and a value of zero for very dissimilar spectra; however, no metric is perfect. Due to the prevalence of structurally-similar compounds such as isomers and the prevalence of certain fragmentation patterns across structurally-dissimilar compounds, the unambiguous assignment of EI-MS spectra compounds remains difficult. Frequently, querying an observed EI-MS spectrum against a large database such as the NIST17 library yields multiple possible assignments requiring the end user to distinguish between multiple high scoring hits, or multiple low scoring hits while keeping in mind that the correct hit may not be in the database at all. Although techniques such as orthogonal information from techniques such as chromatography can greatly aid in unambiguous assignment, this also requires more complicated experimental designs and access to more complicated analytical instrumentation. Substructures can be trivially detected and represented as strings using a previously published technique called node coloring from a known chemical structure. However, for experimentally-derived EI-MS spectra this information must be derived from the spectra itself (i.e., because we do not know what compound it represents). To achieve this, the software uses techniques from the field of machine learning and a large training dataset of EI-MS spectra corresponding to known structures annotated with substructure strings, to build models that can predict the presence of a given chemical substructure from an EI-MS spectrum directly.If these predictions are of high-quality (i.e., are unlikely to be false positives), the presence of one or more predicted substructures can be used to constrain the number of possible hits for a query spectrum. Mathematically, this restriction could be expressed in many forms, but the most straight-forward implementation is to weight the cosine similarity of a query spectrum and a plausible database match with a Tanimoto-like coefficient based on the ratio of the number of substructures predicted to the number of substructures present in the potential database hit. Determining which combination of models best reduces assignment ambiguity will be achieved using a combination of manual curation and optimization techniques such as genetic algorithms. This software will perform all the steps necessary to construct said models from a training dataset and evaluate them using a holdout dataset. Various statistical analyses can be performed to determine if this approach does decrease assignment ambiguity. For example, if this approach works, on average, the rank-order of the correct assignment for the holdout set of EI-MS spectra should decrease and the weighted cosine similarities for most of the possible matches in the database should be better than the unweighted cosine similarities. Furthermore, this same pipeline can be used on real experimental data to generate less ambiguous assignments.

Mitchell, Joshua↗

Cold-Season Precipitation Sensitivity to Microphysical Parameterizations: Hydrologic Evaluations Leveraging Snow Lidar Datasets

Abstract Cloud microphysical processes are an important facet of atmospheric modeling, as they can control the initiation and rates of snowfall. Thus, parameterizations of these processes have important implications for modeling seasonal snow accumulation. We conduct experiments with the Weather Research and Forecasting (WRF V4.3.3) Model using three different microphysics parameterizations, including a sophisticated new scheme (ISHMAEL). Simulations are conducted for two cold seasons (2018 and 2019) centered on the Colorado Rockies’ ∼750-km 2 East River watershed. Precipitation efficiencies are quantified using a drying-ratio mass budget approach and point evaluations are performed against three NRCS SNOTEL stations. Precipitation and meteorological outputs from each are used to force a land surface model (Noah-MP) so that peak snow accumulation can be compared against airborne snow lidar products. We find that microphysical parameterization choice alone has a modest impact on total precipitation on the order of ±3% watershed-wide, and as high as 15% for certain regions, similar to other studies comparing the same parameterizations. Precipitation biases evaluated against SNOTEL are 15% ± 13%. WRF Noah-MP configurations produced snow water equivalents with good correlations with airborne lidar products at a 1-km spatial resolution: Pearson’s r values of 0.9, RMSEs between 8 and 17 cm, and percent biases of 3%–15%. Noah-MP with precipitation from the PRISM geostatistical precipitation product leads to a peak SWE underestimation of 32% in both years examined, and a weaker spatial correlation than the WRF configurations. We fall short of identifying a clearly superior microphysical parameterization but conclude that snow lidar is a valuable nontraditional indicator of model performance.

54 ENVIRONMENTAL SCIENCES↗

Dataset documenting rock core evaluation from Brady’s Hot Springs well BCH-03

Nineteen short sections of core were taken from the BCH-03 borehole at Brady’s Hot Springs, Churchill County, Nevada, between 2,945 ft and the total depth of 4,885 ft. This well is in a geothermal field. Data collected includes thin section photomicrographs, Medical CT tiff stacks, microCT tiff stacks, XRD, porosity (from thin sections, He porosimeter, microCT segmentation) and P/S wave velocities.

Brady geothermal field↗

Detection of Outliers in LiDAR Data Acquired by Multiple Platforms over Sorghum and Maize

High-resolution point cloud data acquired with a laser scanner from any platform contain random noise and outliers. Therefore, outlier detection in LiDAR data is often necessary prior to analysis. Applications in agriculture are particularly challenging, as there is typically no prior knowledge of the statistical distribution of points, plant complexity, and local point densities, which are crop-dependent. The goals of this study were first to investigate approaches to minimize the impact of outliers on LiDAR acquired over agricultural row crops, and specifically for sorghum and maize breeding experiments, by an unmanned aerial vehicle (UAV) and a wheel-based ground platform; second, to evaluate the impact of existing outliers in the datasets on leaf area index (LAI) prediction using LiDAR data. Two methods were investigated to detect and remove the outliers from the plant datasets. The first was based on surface fitting to noisy point cloud data via normal and curvature estimation in a local neighborhood. The second utilized the PointCleanNet deep learning framework. Both methods were applied to individual plants and field-based datasets. To evaluate the method, an F-score was calculated for synthetic data in the controlled conditions, and LAI, the variable being predicted, was computed both before and after outlier removal for both scenarios. Results indicate that the deep learning method for outlier detection is more robust than the geometric approach to changes in point densities, level of noise, and shapes. The prediction of LAI was also improved for the wheel-based vehicle data based on the coefficient of determination (R2) and the root mean squared error (RMSE) of the residuals before and after the removal of outliers.

36 MATERIALS SCIENCE↗