National Academies Press: OpenBook

2019-2020 Assessment of the Army Research Laboratory (2021)

Chapter: 3 Computational and Atmospheric Sciences

« Previous: 2 Network and Information Sciences
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

3

Computational and Atmospheric Sciences

The Panel on Information Science at the Army Research Laboratory (ARL) also conducted its review of selected research and development (R&D) projects of the ARL computational sciences research core competency at Adelphi, Maryland, on June 18-20, 2019. The research areas reviewed were computational sciences and atmospheric sciences. Projects reviewed related to Battlefield Environment Division (BED) atmospheric observations and modeling and computational sciences research, considering both selected, in-depth research presentations and informal discussions surrounding research posters. Projects presented in depth spanned turbulence modeling in complex environments, using a lattice-Boltzmann computational model, understanding the influence of forest canopies on atmospheric dynamics on complex terrain, uncertainty quantification modeling for atomistic-scale modeling, and artificial intelligence/machine learning (AI/ML) at the edge implemented in field programmable gate arrays (FPGAs). In addition, posters spanned atmospheric model prediction via radar data assimilation, meteorological sensor arrays, aircraft vortex and rotor wake characterization, and ML characterization of particle shapes.

While daunting, the repurposing of the ARL mission also presents great opportunities to restate, revise, or create a new vision, which would allow scientists to reevaluate their individual programs in terms of how they fit into the new “big picture.” For some projects, the new multi-domain operations (MDO) viewpoint was an immediate and natural fit, while for others, the connections between project objectives and MDO were still being developed. Broadly, the work presented was of high quality, comparable to that conducted at major research universities or leading-edge federal R&D organizations.

Examples of this high-quality research include the hierarchical multiscale (HMS) project, which continues to break new ground, by maintaining and now extending its quality and utility from earlier reviews. Similarly, the AI project, using FPGAs, seeks to build capabilities into Army weaponry (beginning with gunsights), via a framework that involves software, customizable hardware, and reduced convolutional neural nets (CNNs) to meet space, weight, power, and time-to-solution constraints. In addition, the work on ML to characterize particle shapes using scattered light images has the potential for wide applicability throughout aerosol science and in the broad area of chemical or biological agent characterization.

There are a few areas of opportunity. As an enabling, broad-based capability, the computational sciences and BED need to maintain a critical mass of expertise, both for targeted projects and for collaborative engagement with other projects. Similarly, the BED atmospheric modeling research is foundational for many MDO activities, but it would benefit from stronger connections to specific projects.

While it is too early in the deployment of the new strategies to determine which of the BED projects presented during the review are disruptive, two projects have the potential, depending on the path and risks that the projects take over the next several years. The first is the application of ML to characterization of particle shape using scattered light images. As stated earlier, this project, as currently envisioned by the BED team, has the potential for wide applicability throughout aerosol science and in the broad area of chemical or biological agent characterization. A very narrow application of this technology—shape characterization of hydrometeors—could, depending on the research, influence areas such as numerical weather prediction (NWP), where cloud prediction remains a challenge even after 30+ years of microphysics research and testing.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

The second is a meteorological sensor array (MSA) application, where chemical detectors are deployed in addition to the meteorological sensors in the simulated urban areas located in the White Sands Missile Range (WSMR) complex. Besides attracting a number of interested atmospheric chemistry researchers, the installation of chemical detectors could allow detailed measurements of chemical dispersion and transport to be made within urban terrain under many different atmospheric conditions. The collection and analysis of such detailed data could be used in multiple AI development programs that require intelligent agents to operate, collaborate, and cooperate with soldiers in diverse environments, including those degraded by adverse weather or adversary concealment and deception strategies. The research teams could conduct a thorough investigation into the feasibility of pursuing these tracks before investing precious resources.

In turn, computational science is becoming a prominent element in training AI/ML systems, such as work using physics-based simulation engines from video games as the source of data to train autonomous vehicles. Further, many of the technical challenges in modern ML involve problems that have been well investigated in computational science work, and emerging research in academia and industry that involves substantive collaborations between computational science and AI/ML is increasingly seen. The Computational and Information Sciences Directorate (CISD) is well-poised to catalyze important new work that involves collaborations between computational and information sciences, an area that does not currently appear to be pursued.

In summary, there is continued research progress, particularly engagement with academic stakeholders via ARL’s distributed sites and collaborative academic projects. There is hiring and intellectual development of new postdoctoral associates and staff. ARL needs to continue to foster collaboration across its internal organizational structure. Last, all projects would benefit from clearer metrics for research project success and associated project exit strategies, including transitions that maximize Army benefits.

To provide continuity in review assessments, it would be helpful to identify which projects are part of broad-area efforts or important, single thrusts and which are new starts or continuation of previous projects, and to provide a brief status summary of all projects in progress.

COMPUTATIONAL SCIENCES

Computational science has a rich history at ARL, from pioneering work on ballistic calculations using some of the world’s first computers to the laboratory’s current role as an essential provider of computational resources for the entire Department of Defense (DoD) high-performance computing (HPC) program. Many ARL-resident research projects are at the leading edge of computational science applications for Army needs, including in areas such as simulation of energetic materials for explosives analysis and the design of advanced armor to protect soldiers and vehicles.

The recent changes in the laboratory and Army organizational structures present new challenges to the continuity of advanced computational research for the Army. These changes include the new emphasis on MDO, along with organizational and programmatic revisions at the laboratory. ARL management appears to be insulating the intellectual work of the technical staff from the administrative details of these widespread changes, and the ARL researchers continue to conduct excellent research despite the uncertainties that invariably arrive with organizational change.

Accomplishments and Advancements

The technical quality of the computational sciences is high, with lower variance in the technical quality than in past reviews. Presented projects were generally cast in the new reference frame of MDO. For some projects, this MDO viewpoint was a natural fit, while for others the connection needed further

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

development. Not surprisingly, the MDO framework will require time for each research effort to adapt, and future review panels could ensure alignment with this emerging set of higher-level DoD guidelines.

A few R&D projects in computational science proved especially worthy of mention, as summarized below.

Methodologies for Scale Bridging in Multiscale Simulation

The hierarchical multiscale (HMS) project continues to conduct relevant research, by maintaining and extending its quality and utility from earlier reviews. HMS has now been extended into the time domain via continuation of modeling of RDX1 reactivity as an evolutionary multiscale process. This extension to temporal scales, instead of a purely spatial focus, has generated new research work in stochastic modeling and probabilistic ML as applied to the loss and reconstruction of fine-scale microstructural information at each time step. This spatial-temporal extension has supported a number of research exemplars from practical applications (e.g., traction estimation for vehicle mobility in soils, essential to the effective function of the Army’s maneuver forces, and the design of next-generation armor enabled by advanced material processes such as additive manufacturing).

Current HMS work in capturing microscale structure of energetic materials is especially promising because this avenue of research increasingly opens new opportunities for simulating the onset of detonation response. In addition to these important advancements on existing ARL multiscale computational technology, the HMS project team is publishing results regularly in high-quality journals and has added a university collaborator to provide additional expertise on uncertainty quantification for the project. As such, this project includes a broad set of key elements of research quality, including high-quality publications, external collaboration, and applications to design, analysis, discovery, and uncertainty quantification. ARL could benefit from collaboration with the University of Illinois (D. Scott Stewart) in this area.

Artificial Intelligence/Machine Learning at the Edge: Inferencing Engines on Field Programmable Gate Array

This promising project concerns development of AI/ML inference engines that can be deployed as a digital chip (application-specific integrated circuit [ASIC]) in Army environments even when network connectivity is not available (e.g., for higher precision targeting). Further, when network connectivity is regained, learning online and training could resume to further enhance target solution quality. Such an ASIC-based online/off-line device could have multiple relevant applications across the cross-functional teams if the underlying systems design and engineering R&D were to be successful.

The technical approach involving a software framework, extensible instruction set architecture, and algorithm redesign/refactoring of convolutional neutral nets (CNNs) seeks to reduce computational costs by an order of magnitude and increase efficiencies to meet space, weight, power, and time-to-solution constraints. This project spans the software-hardware space to deliver performance, portability, and programmability across multiple applications, future CNN algorithms, and future FPGA architectures. Initial results seem encouraging, and there are many possible pathways to transition to the field while advancing the basic research. With improved direction, resources and leveraging of related research, there

___________________

1 RDX is an organic compound with the formula (O2NNCH2)3. Chemically, it is classified as a nitramide and is widely used as an explosive.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

is the potential for outstanding successes. If not already doing so, ARL needs to consider Google’s TensorFlow ML software, which is becoming the industry standard.2

Improving Numerical Weather Prediction in Convectively Unstable Environments by Assimilating Radar Observations

The work on improving numerical weather prediction over short time frames through assimilation of radar observations represents innovative research in support of an impactful real-world application.3 ARL (collaborating with the Combat Capabilities Development Command [CCDC] Aviation and Missile Center, formerly known as AMRDEC) is developing a new mesoscale modeling capability that will provide forecasts in data-sparse regions such as the test facility on Kwajalein Atoll in the Pacific Ocean. Because of the site’s unique remote location, current DoD weather capabilities cannot meet these needs.

The ARL approach assimilates radar observations already available at the location into the widely used weather research and forecasting (WRF) prediction model. Radar measurements of reflectivity are then used to infer rates of latent heating for input to WRF. Assessments of the new modeling capability demonstrate increases in probability of correct prediction of weather phenomena through assimilation of radar data by factors up to one order of magnitude.

Challenges and Opportunities

Role of Computational Science at ARL

There is value in a careful definition of the relationship between work in traditional computational science (i.e., large-scale computational simulation based directly on applications implementing scientific principles) and the emerging applications domains of AI (in general) and ML (in particular). AI/ML is a form of computational science, owing to direct correspondence between their respective characteristics. Each is a form of computational simulation, each requires careful consideration of accuracy and uncertainty for an intended use, and each requires due attention to performance and scalability in practical settings.

Depth Versus Breadth

As an enabling, broad-based capability, the computational sciences need to maintain a critical mass of expertise, both for targeted projects and for collaborative engagement with other projects. Areas that have emerged in the past few years, such as AI and data analytics, are very large in scope. Concurrently, the roles of computational science and scientific computation are continuing to expand and evolve. At current staffing levels, ARL computational science may be forced to choose between focusing in depth on a few key areas of most value to the Army or trying to cover a broad range of topics at very limited depth.

___________________

2 See TensorFlow, “An End-to-End Open Source Machine Learning Platform,” https://www.tensorflow.org/, accessed May 11, 2020. TensorFlow is an end-to-end open source platform for machine learning. It has a comprehensive, flexible ecosystem of tools, libraries, and community resources that let researchers push the state of the art in ML and developers easily build and deploy ML-powered applications.

3 This is an atmospheric sciences project—details can be found in the “Atmospheric Sciences” section later in this chapter.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

Internal Collaborative Opportunities

Although many collaborations with academia and industry were reported during the review, the lack of significant computational science interactions with other ARL directorates was notable. With the increasing role of computational science and scientific computation in technical areas under the purview of ARL, this may represent an opportunity for the computational sciences group to become the “go to” division in ARL for advanced computational work. As an illustration, this is, to some extent, the model employed by the Air Force Research Laboratory with its Information Directorate.

Critical Personnel Needs

The computational science R&D portfolio is ambitious and well scoped to serve key needs, but the scope of this portfolio will likely require additional subject-matter expertise, via hiring of new staff, retraining of current staff, or collaboration with external and internal partners. Here are some of the potential areas that may need additional expertise:

  1. Expertise in real-time processing and discrete event simulation, when combined with computational science, HPC, and AI, can bring predictive power and decision-making support that otherwise may not be within reach.
  2. With an increased emphasis on autonomy, developing staff expertise in computational models for autonomy, complex systems, and emergent behaviors may be required for successful completion of research goals for projects involving autonomous capabilities.
  3. Additional expertise in uncertainty quantification (both in computational sciences and battlefield environments), sensitivity analysis, and credibility of computational models will be needed to assess how uncertainty can be propagated through decision chains and how it can be conveyed to battlefield decision-makers.

Collaborations Between Computational and Information Sciences

Simulation is becoming a prominent element in training AI/ML systems, such as work using physics-based simulation engines from video games as the source of data to train autonomous vehicles. Further, many of the technical challenges in modern machine learning involve problems that have been well investigated in computational science work, and emerging research in academia and industry that involves substantive collaborations between computational science and AI/ML is increasingly seen. CISD is poised to catalyze important new work that involves collaborations between computational and information sciences, an area that does not currently appear to be pursued.

ATMOSPHERIC SCIENCES

BED has three main objectives: (1) guide and concentrate foundational research to identify and address Army and DoD scientific and technological gaps; (2) couple fine-scale atmospheric models with novel sensing technologies to enable robust situational understanding; and (3) predict and express environmental impacts on military systems for rapid, human-in-the-loop and autonomous decision making. The 2019 BED review was conducted in the wake of unprecedented changes that have taken place in the Army and ARL within the past year. Because these changes are less than a year old, it will take time for BED scientists to integrate these concepts into their research portfolio and determine

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

whether changes in the direction of projects need to be undertaken. However, with such dramatic changes come equally dramatic opportunities, which are elaborated upon later in this report.

BED’s current staffing is 52, which includes 41 government personnel (including three ARL fellows) and 11 contractors. Contract personnel include three visiting senior researchers and three postdoctoral associates. Previous reviews reported that smart use of postdoctoral associates has benefited the quality of BED’s research, and that this strategy could be continued, especially if promising candidates can be converted to permanent staff. During fiscal year (FY) 2019, one former postdoctoral associate became a member of BED’s government staff.

Accomplishments and Advancements

The research projects presented covered the following topics: (1) turbulence modeling in urban, complex, and forested domains using the lattice-Boltzmann method; (2) influence of a forest canopy on flows in complex terrain; (3) improving numerical weather prediction in convectively unstable environments by assimilating radar observations; (4) simulating turbulent, natural convection using the vortex filament method; (5) meteorological sensor array overview/update; (6) impact of unmanned aerial system rotor-wake interactions on atmospheric sensing capability; (7) aircraft vortex detection and characterization: C-17 formation spacing research project; (8) application of the aerosol Raman spectrometer toward the understanding of ambient atmospheric aerosols; (9) multifidelity uncertainty quantification applied to nonlinear phenomena; and (10) application of machine learning to characterization of particle shape using scattered light images. Updates to projects first described to the ARLTAB in 2017 are indicated as such.

Turbulence Modeling in Urban, Complex, and Forested Domains Using the Lattice-Boltzmann Method

This is an update to a project first described to the ARLTAB in 2017. ARL’s assessment is that there is no dominant community model for forecasting microscale flow in the atmospheric boundary layer (ABL). ARL developed the atmospheric boundary layer environment-lattice-Boltzmann method (ABLE-LBM) microscale model for horizontal grid spacing of 1-100 m, temporal resolution on the order of minutes, and efficient use of modern compute accelerators (such as GPGPU). There are computational speed advantages of LBM, and early simulation comparisons with idealized laboratory simulations are promising.

Influence of a Forest Canopy on Flows in Complex Terrain

This project made use of Perdigão, Portugal, field experiment data (May-June 2017), for which ARL scientists played a key role as a planning and principal investigating organization, and during which they emplaced and operated a substantial complement of meteorological towers and instrumentation at the test site. Results were shown for various configurations of WRF model with small horizontal grid spacing, the smallest of which operated on scales of 50 m. The high-resolution simulations produced the most accurate results owing to the impact of high-resolution, corrected land-surface data—not just improved terrain representation.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

Improving Numerical Weather Prediction in Convectively Unstable Environments by Assimilating Radar Observations

This project’s objective is to improve short-term precipitation forecasts to minimize damage to high-speed projectiles by hydrometeors and was highlighted as an exemplar earlier in this report. The initial location is the Kwajalein Atoll in support of a CCDC Aviation and Missile Center contractor’s launch activities. The method adapted a proven (National Oceanic and Atmospheric Administration Earth System Research Laboratory) technique for assimilating radar data into the preforecast period of an NWP model (WRF) for the purposes of improving latent heating profiles, which improves the forecast model representation of clouds, and thus the 0-6 hour forecast of precipitation. A particularly challenging aspect is the lack of surrounding data given the atoll’s location. Despite these challenges, fractional skill scores4 for the model are promising, with best scores coming in the 0-3 hour time frame, when using the longest assimilation window, a very important lead time for the user.

Simulating Turbulent, Natural Convection Using the Vortex Filament Method

This is an update to a project first described to the ARLTAB in 2017. An idealized Lagrangian vorticity model is used to explore new ways of representing turbulence in the presence of localized heating. The flow is represented as vortex filaments with constant circulation. Filaments are constructed piecewise by vortex tubes. Filaments interact, stretching and distorting each other. Additional vortex tubes are introduced in response to diabatic vorticity generation from the heating or to account for interactions between vortex filaments that can result in contorted filament geometries. The energy equation for Lagrangian transport of particles simulates heat transport.

Previous results with this technique showed the ability to reproduce the thermal bubble simulations of Shapiro and Kanak.5 The nondiffusive Large Eddy Simulation method has produced turbulent flows on a vertically heated wall, and there are plans to study the method over a vertical, viscous heated wall as well as with other geometries. This framework provides a promising method for representing turbulence and heat diffusion in the boundary layer. One application of atmospheric changes forced by localized heating is changes in acoustic signals, potentially affecting communications or “hiding” battlefield resources (see additional comments in the following “Challenges and Opportunities” section).

Meteorological Sensor Array (MSA) Overview/Update

This was an update from the 2016 and 2017 ARLTAB reviews. The MSA is a large, multiyear project designed for acquisition and understanding of atmospheric phenomena in complex terrain. Significant progress in the development of the MSA has been made over the past 2 years despite multiple logistical and operations/maintenance challenges. In 2019, all sensors along the foothills location (Jornada Experimental Range at WSMR) have been installed, and while instruments for the east side of the domain have been purchased and delivered, they have not all been installed. Some cooperative data collection by the U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory and Naval Research Laboratory (NRL) has already taken place, and the next meeting of

___________________

4 The fractions skill score was one of the measures that formed part of the intercomparison of spatial forecast verification methods. The fractions skill score was used to assess a common data set that consisted of real and perturbed WRF model precipitation forecasts, as well as geometric cases.

5 A. Shapiro and K.M. Kanak, 2002, Vortex formation in ellipsoidal thermal bubbles, Journal of the Atmospheric Sciences 59(14):2253-2269.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

micrometeorological research groups interested in the study of atmospheric flows over complex terrain at high horizontal resolution will take place at WSMR.6

Impact of Unmanned Aerial System Rotor-Wake Interactions on Atmospheric Sensing Capability

This new project was started to support future Army operations in which numerous unmanned aerial systems (UASs) will be operating in the battlefield, with the recognition that environmental conditions will affect their flight operation. In the process of conducting a review of the available studies on the placement and use of environmental and characterization sensors on UASs, the research team found that the literature was surprisingly limited. This justified the need for studying the full parameter space of the sensor-placement problem. The team employed the wind tunnel facility at New Mexico State University to measure particle-size velocimetry for variously configured rotary UASs. Results for forward and vertical flight of a quadcopter-style UAS showed that a rotor-wake interaction region below the multirotor body and rotors precludes the use of in situ sensors. Following these initial results, the team resorted to three-dimensional (3D) printing of various versions of the vehicle in order to test small variations in specifications. Preliminary flow visualization results indicate that in situ sensor placement is limited to the top of the multirotor, and data collected during descent is contaminated by rotor wake. Thus, a correction for velocity measurements will need to be applied that is a function of the vehicle dynamics. Additional work using the faculty, students, and facilities at New Mexico State University is being planned, and the group is aware of other atmospheric UAS efforts.7

Aircraft Vortex Detection and Characterization: C-17 Formation Spacing Research Project

The C-17 is one of the “workhorses” involved in airdrop operations supporting the Army. Existing standards for wake turbulence mandate a standard separation between aircraft, and the spacing limits the speed with which multiple drops of equipment and personnel can be accomplished. This project was aimed at reducing the spacing between C-17s participating in airdrops. A firm understanding of the influences of wake vortices is needed to accomplish this task. This multiyear project (set to end in 2020) has collected coupled ground- and air-based light detection and ranging (lidar) measurements of wingtip vortices to characterize atmospheric wind and turbulence vertical profiles. Modeling the behavior of the vortex structure is extremely difficult.

New models, incorporating statistical uncertainty, are being developed and validated using lidar-derived vortex characterization and evolution. The near-term approach is to focus on ground and airborne vortex measurements and development of a new automated vortex detection algorithm for rapid analysis of Doppler lidar data. Data collection in several locations is planned for 2020 using multiple C-17 formation airdrops for model validation. ARL scientists have collaborated with colleagues from Draper Laboratory, Simpson Weather Associates, the Naval Postgraduate School, and ATEC on this project. The ultimate goal is the development of new doctrine that will allow tighter aircraft formation spacing to enhance the speed and efficiency of airdrops.

___________________

6 For more information about complex terrain observational research and ARL’s role in the community, please see H.J.S. Fernando, J. Mann, J.M.L.M. Palma, J.K. Lundquist, R.J. Barthelmie, M. Belo-Pereira, W.O.J. Brown, et al., 2019, The Perdigão: Peering into microscale details of mountain winds, Bulletin of the American Meteorological Society 100(5):799-820.

7 See, for example, the CLOUD-MAP project, http://www.cloud-map.org.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

Application of the Aerosol Raman Spectrometer to Understanding Ambient Atmospheric Aerosols

This is an update to a project first described to the ARLTAB in 2017. Raman spectroscopy is an established tool for measuring atmospheric aerosols. The work that was conducted in this area over the review period extends past work and identifies the chemical nature of aerosol species. A new collaboration with New Mexico State University was undertaken, and it is clear that the project is making progress in advancing basic aerosol science. Efforts were made to utilize the MSA at WMSR to correlate aerosol measurements and meteorological parameters. Unique aerosol compositions were measured under different ambient conditions, and attempts were made to correlate composition with sources. The potential future development opportunities for this project are high, particularly as source apportionment activities are continued and efforts to miniaturize the system and expand the available database of spectra that may be used to identify species are undertaken.

Multifidelity Uncertainty Quantification Applied to Nonlinear Phenomena

Cost-conscious use of computational resources in delivering detailed, targeted model guidance is the end goal of this basic research project. In the theoretical framework presented, results from a high-fidelity model (HFM) capture the level of flow detail required; these are regarded as “truth.” The task is to make optimal use of a low-fidelity model (LFM) to explore the critical parameter space (in this case, upstream wind speed and building height). Many simulations of the LFM are used to explore this parameter space, and a “greedy algorithm method” to subselect parameter combinations is based on the LFM solutions. The HFM is run on this reduced set of parameter combinations and a mapping between the LFM and HFM solutions provides a framework for deducing HFM solutions for any arbitrary parameter combination.

While it is in its initial stages, this method has the potential for real-time calculation of detailed environmental situations in the field, and by running the LFM for many parameter combinations near to the first chosen, the uncertainty of the forecast can be quantified, providing a measure of the confidence in the details of the forecast.

Application of Machine Learning to Characterization of Particle Shape Using Scattered Light Images

The characterization of particle shape helps to identify classes of particles that may be present in an environment. Imaging techniques that can be used to help to identify particle shapes are widely available (e.g., scanning electron microscopy analyses), but require sampling followed by analyses in a laboratory setting. The work presented novel research to utilize ML tools and light scattering experiments in order to identify particle shapes. Simulated particle shapes with controlled refractive indices were used to train the code. One-dimensional (1D) light scattering data were subsequently used to test the shape characterization algorithm but were insufficient to predict the particle shapes with high certainty. However, two-dimensional (2D) light scattering data resulted in significant improvements in the identification of shapes.

Challenges and Opportunities

The team believes that in the new MDO environment, project scientists need to have a clearer idea as to what the ultimate application of their project’s results could be. This longer-term view is necessary to articulate project relevance to fellow scientists outside the division as well as to senior leadership, where the latter’s worldview is typically at a high level. An example of one way that a long-term view could be

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

created is presented for the simulating turbulent, natural convection using the vortex filament method project. The project’s abstract states that its transition will be “a fundamental tool for studying atmospheric turbulence.” In years past, this goal would have been more than adequate, and many scientists familiar with the problem would have applauded it. However, the team suggests taking this several steps further, such as examining how the turbulence characterization could be applied to study effects on aerosols, smoke, and propagation. Taking this yet further, the turbulent effects information could be applied to help sensors better detect and identify threats. Ultimately, the application in an MDO environment could inform strategies to counteract deception and concealment strategies employed by a potential near-peer adversary. As shown in this example, the opportunity to reevaluate BED’s research portfolio in light of the significant external changes and priority shifts to the ARL research framework provides lead scientists a chance to articulate interdependent relationships (such as in this example) and to make adjustments in cases where they are not as clear as they could be. The team acknowledges that setting these project visions will be more difficult for some projects than for others.

BED researcher interactions with academia continue to be strong and vital, but challenges remain in the engagement with operational entities such as the Air Force, who provide operational weather support for all Army operations except artillery. To make full use of the impactful BED research, it will be necessary to continue dialogues with leadership in the Air Force Director of Weather, in order to keep situational awareness of evolving Air Force requirements and operations and identify possible paths for technology transition. There are also entities within the Air Force outside the weather community who would find BED’s environmental research programs very useful. As the MDO doctrine evolves and is integrated into joint operations, it is anticipated that such broadening interactions will bear fruit.

In some respects, the challenges here are not unlike those discussed in the 2017 review. While BED’s research programs focus on atmospheric science, programs throughout ARL with environmental sensitivities could use the expertise of BED’s scientists. This is especially true within the emphasis area of AI, where many robotic algorithms need to be able to behave properly in degraded environmental conditions in order to provide added value to the soldier in theater. MDO will require such sophistication to defeat countermeasures involving obscurants and other types of concealment that will almost certainly be deployed by a near-peer competitor. As stated earlier, BED as a whole has a great opportunity to reevaluate its position and establish a vision that will keep its portfolio focused on the new and evolving requirements of an MDO operating environment. However, as in 2017, resource allocation continues to be a challenge. BED could continue to leverage use of as many avenues for assistance for its projects as feasible. This will ensure that personnel can strike the right balance between the environment-centric projects within the division and the environmentally sensitive ones outside the division. It is worth restating that, while full BED collaboration with other laboratory projects has the potential for great impact on these other projects, it will put a significant strain on BED’s existing resources.

Turbulence Modeling in Urban, Complex, and Forested Domains Using the Lattice-Boltzmann Method

Although the work is sound and has clear implications for supporting future Army operations in challenging operational environments, some important issues can be raised about the future of this effort. For instance, it was not obvious from the presentation what the operational testing strategy and success criteria would be once model runs are conducted using detailed observational field data. Examples of the types of questions that could be asked are: How many different configurations need to be tested and for how many different types of flows? How could observations be used to guide the initialization of the model? How many different data sets need to be used for comparison? Without this type of structure, the project runs the risk of becoming open-ended.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

Influence of a Forest Canopy on Flows in Complex Terrain

Several limitations or future challenges were identified. First, a number of modifications to the model’s preprocessing data files were required (e.g., changes to land-use categories based on lidar data to account for complex vegetation along the hillsides). How could such modifications be made in areas that do not have access to lidar? Second, while the 50-meter resolution version of WRF produced the most accurate simulations of small-scale flows along the parallel ridges there, upslope case simulations had some accuracy issues with temperature and wind direction, and there were similar problems for downslope cases. Third, several sets of field observations were shown that did not fit previously accepted conceptual models for upslope and downslope flows. All of these require additional analysis and study.

Improving Numerical Weather Prediction in Convectively Unstable Environments by Assimilating Radar Observations

The principal investigator (PI) is investigating the broader adaptability of this project to other Army applications; use of satellite data in addition to the radar will allow the technique to be applied in locations without conventional surface observations and weather radar. The PI is aware of research into satellite-based convective initiation methods such as cloud-top cooling, and is looking to engage the Federal Aviation Administration and Massachusetts Institute of Technology’s Lincoln Laboratory about using the Corridor Integrated Weather System (CIWS), which applies satellite imagery for short-range (0-2 hour) projections supporting air traffic management operations in U.S. offshore regions, to explore enhancing the predictability of mesoscale phenomena in data-sparse regions. Because the Air Force is also being provided offshore CIWS data, the PI wants to investigate the possibility of tapping into that data feed to examine results and evaluate the technique. ARL needs to also tap into work of the National Aeronautics and Space Administration and Department of Energy (DOE) in the area of NWP.

Meteorological Sensor Array Overview/Update

In addition to potential international research collaborations, the group has proposed instrumenting space within the MSA that has buildings to create a unique, urban-styled test-bed area within WSMR that can provide valuable meteorological data within dense urban environments. Such a test-bed area could potentially lead to collaborative opportunities, with researchers investigating the use of robotic vehicles in urban environments. The test-bed could provide them with detailed meteorological data that can be used to evaluate the effects of dust and other obscurants on unmanned vehicle operations (e.g., in the AI for maneuver and mobility project). These urban test-bed data could also be valuable for developmental testing of the ABLE-LBM model.

Application of Machine Learning to Characterization of Particle Shape Using Scattered Light Images

The potential that this work provides for enhancing Army research and broader impacts in the scientific community is exceptionally high. Of particular note is the potential to enable the identification of multicomponent materials. Further coupling of the particle-shape characterization technique (for initial screening) with a fast-chemical characterization technique would enable significant enhancements in particle analyses. As such, the project’s potential as a disruptive technology is exceptionally high.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

OVERALL QUALITY OF THE WORK

Computational Sciences

The technical quality of the computational sciences work represented in the presentations and posters was generally high. Research papers are being published in high-quality journals and being presented at the leading conferences. The ARL staff demonstrated deep knowledge in many of the areas in which they are working, through their descriptions of their research and in providing answers to questions. There were also several good examples of effective collaboration with external experts in areas in which ARL seeks to develop internal capability.

The research quality of the material presented was high—comparable to R&D conducted in leading universities and within the federal national security complex. The particular exemplars detailed in the “Accomplishments and Advancements” section above would serve as typical research projects in U.S. university, federal, or industrial venues.

The projects presented demonstrate an appropriately broad appreciation of the relevant scientific principles, as well as a clear understanding of external research results. The three exemplar projects cited above are excellent illustrations of this, given that they each demonstrate connections to external research results and to appropriate advanced research methods.

Equipment needs are mostly computational, and ARL provides excellent support for these research requirements. The numerical methods utilized in the projects that were presented are appropriate for the scope of work as proposed and are representative of the advanced techniques used outside the institution.

The research teams are composed of high-quality personnel with relevant subject matter expertise for the associated projects. The team personnel reflect a good mix of junior, mid-career, and senior staff, and the research outcomes appear to be both ambitious and feasible

The computational sciences activities are very well supported through the ARL Supercomputing Research Center, which includes both a DoD Supercomputing Research Center and a DoD Data Analysis and Assessment Center. However, ARL could further collaborate with supercomputing programs at other DoD and DOE laboratories.

The research projects presented do employ appropriate theoretical and computational methods, and those projects that involve experimental validation results, or that use existing sensors (e.g., the radar project cited as an exemplar above), generate data sets appropriate to the various project needs.

Each of the exemplar projects is a good illustration of especially promising projects. While none of these were obvious candidates for “disruptive” status, each showed particular promise in important characterizations such as “potentially game-changing” or “especially cost-effective for the Army.”

As noted elsewhere, future reviews would benefit from a brief summary of the state of all projects, not just those presented at the review. This would allow a more holistic assessment of ARL’s work and the potential for disruptions.

Atmospheric Sciences

The technical quality of the BED research represented in the presentations and posters was generally high. Researchers are very familiar with the underlying science and cognizant of research conducted elsewhere; in some cases, BED researchers are in active dialogue or collaborating with researchers outside the laboratory. In all cases, the researchers are aware of the potential challenges associated with their projects. As noted in previous reports, the incorporation of researchers from different disciplines into BED projects is notable. In project updates, a clearer delineation of “new” versus “old” results needs to be made, as results presented in several of these looked similar (although not identical) to those presented in 2017.

The overall scientific quality of the work is high, comparable to research conducted at successful university, government, and industry labs. The majority of the work described in the “Accomplishments

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

and Advancements” section above is comparable to research projects in U.S. university, federal, or industrial venues.

The projects demonstrated a broad appreciation of relevant scientific principles, as well as an awareness and clear understanding of external research results.

The numerical methods used in the projects that were presented are appropriate for the scope of work proposed and are representative of advanced techniques used outside ARL.

The research teams are high quality, with relevant subject matter expertise for the projects, either organic or through collaborative arrangements. The teams have an appropriate mix of junior, mid-career, and senior staff, and are suitably organized for the research challenges presented.

The BED research projects that are computationally intensive appear to have good resource support. The MSA project has made good progress in fielding a vast array of observational equipment, although there are some logistical issues associated with the continuing operation and maintenance of the fielded equipment. The research projects presented employ the appropriate theoretical and computational methods, and while several projects involving experimental validation are just now beginning to utilize observational data, the necessary plans are in place to move those forward. Those using existing sensors, such as the radar assimilation project described earlier, have generated data sets appropriate to the project’s needs.

The radar assimilation project and the C-17 vortex detection and characterization project are likely the closest to transition, although neither is immediately ready. These projects are different from those identified earlier as being potentially “disruptive” in an R&D sense, as the disruptive projects are not as far along in their development. Future reviews would benefit from a brief summary of the state of all projects, not just those presented at the review. This would allow a more holistic assessment of ARL’s work and the identification of potentially disruptive technology.

RECOMMENDATIONS

Computational Sciences

The research portfolio presented is of high quality and is comparable to that expected from a high-quality university research program or from a leading-edge federal R&D organization. The teams are making effective use of available resources, and a good mix of diverse projects in computational science were presented, ranging from practical projects that generate cost-effective results by integrating various established technologies to projects where intellectual venues that have rarely been explored are being developed for particular Army needs arising from the MDO frame of reference.

Discussions with ARL managers indicated that while organizational and programmatic changes are occurring at nearly all levels of the Army, the goal of research program continuity is being realized despite these changes, but that challenges remain to be overcome. These challenges include the usual concerns of available personnel and financial resources, the relocation of research projects that do not fit the fundamental research exploratory nature of ARL’s mission, and uncertainties about the relationship between traditional computational science methods and emerging AI/ML techniques that both support and supplant established methods.

The presentations demonstrate that ARL is leveraging synergies between its various research core competencies (e.g., the AI-FPGA project mentioned above). This development is guided by Army needs and could lead to critical breakthroughs. While advances in technologies or the human factors of the battlefield/edge processes/logistics are needed, innovation at their nexus that are guided by MDO needs present new opportunities.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×

Recommendation: With the mission of combat capability development, the Army Research Laboratory (ARL) should leverage synergies within the computational sciences and the networks and information sciences research core competencies—for example, artificial intelligence (AI), data science, high-performance computing, networks, and battlefield environment effects. Further, such an approach should be applied by ARL across other research areas—for example, human sciences and AI—which could be a force multiplier that may not be possible otherwise.

Atmospheric Sciences

Atmospheric science, particularly meteorology and atmospheric chemistry, is critically important to creating disruptive technologies that have broad applicability to the Army needs. ARL research in these areas is strong overall. The impact of ARL atmospheric science research can be enhanced through compelling communication of its relevance within the umbrella of ARL, and within the context of the overall Army needs. A striking example of the need for integration of meteorology from the beginning of a project is in operationalizing MDO, in which all of the envisioned enabling technologies are potentially vulnerable in some way to degradation from environmental factors on the battlefield and environmentally based countermeasures employed by our adversaries.

Thus, strategic integration of BED research thrusts with complementary research in other areas is encouraged. One mechanism for accomplishing this is the use of the network and information sciences and computational sciences taxonomy to optimize the impact of the BED expertise and limited resources, while providing opportunities for critical research in environmental research and across disciplines. There are numerous opportunities to advance fundamental science in these disciplines while encouraging multidisciplinary collaboration. This trend is one that is consistent with leading edge, collaborative approaches that the broader scientific community is undertaking in addressing grand-challenge questions and promoting use-inspired research.

There are a few areas of opportunity. As an enabling, broad-based capability, the computational sciences and BED need to maintain a critical mass of expertise, both for targeted projects and for collaborative engagement with other projects. Similarly, the BED atmospheric modeling research is foundational for many MDO activities, but it would benefit from stronger connections to specific projects. Last, efforts to date to include atmospheric considerations in other venues is commendable. However, it is important to develop new collaborations given that environmental phenomena are crosscutting and affect nearly every Army operation.

Recommendation: The Army Research Laboratory (ARL) should evaluate the potential for environmental integration at the start of a project’s activities. ARL should consider such work that could lead to novel basic science approaches and be transformative in the broader scientific community, particularly in the use of computational techniques and large data sets to enhance decision making, technology use, or operational efficiencies.

Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 27
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 28
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 29
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 30
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 31
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 32
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 33
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 34
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 35
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 36
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 37
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 38
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 39
Suggested Citation:"3 Computational and Atmospheric Sciences." National Academies of Sciences, Engineering, and Medicine. 2021. 2019-2020 Assessment of the Army Research Laboratory. Washington, DC: The National Academies Press. doi: 10.17226/26325.
×
Page 40
Next: 4 Human Sciences »
2019-2020 Assessment of the Army Research Laboratory Get This Book
×
Buy Ebook | $34.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The National Academies of Sciences, Engineering, and Medicine's Army Research Laboratory Technical Assessment Board (ARLTAB) provides biennial assessments of the scientific and technical quality of the Army Research Laboratory (ARL). These assessments include the development of findings and recommendations related to the quality of ARL's research, development, and analysis programs. 2019-2020 Assessment of the Army Research Laboratory reviews the following research core competencies of ARL: human sciences, network and information sciences, computational sciences, materials and manufacturing sciences, and propulsion sciences. This biennial report summarizes the findings of the ARLTAB from reviews conducted in 2019 and 2020.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!