Geospatial Change Leadership: The Value and Challenge of Collaboration

Dr. Rachel MK Headley - Rose Group International

Moving at the speed of change has challenged the geospatial industry. An industry, which for decades operated with long-term projects requiring massive investment, now needs to operate more cooperatively to achieve ambitious new goals. From UAS to new information products, the industry has experienced incredible technical advances yet remains opaque to a wide variety of potential users.

For those trying to leverage these new opportunities, it can seem impossible to convince others to join them on the journey. Dr. Headley, former member of the USGS Landsat leadership team, will share critical secrets of how to see beyond our silos to develop collaborative efforts and expand the geospatial community.

Hyperspectral on the Space Station: Lessons Learned and Goals Achieved

Amanda O’Connor - Teledyne Brown Engineering

The International Space Station (ISS) provides a unique platform for imaging and earth looking data collection. While satellite launches depend on successful orbit and functioning, instruments on the ISS can be adjusted and returned to earth to analyze any issues, repair, and understand the impacts of launch and the space environment. DESIS, the DLR Earth Sensing Imaging Spectrometer, is located on the ISS on the MUSES, Multiple User System for Earth Sensing platform, which enables mounting an instrument outside of the ISS, provides downlink capabilities, compute capabilities and the option to return the instrument.

By providing return capability, the ISS provides a lower risk opportunity to test equipment and modify designs prior to expensive constellation launches. Hyperion, on the EO-1 Satellite and HICO on the ISS are the only previous publicly available space based hyperspectral system, both no longer collecting. DESIS represents a huge leap forward with regard to sensor performance (e.g. signal to noise ratio), spatial coverage, narrow spectral bands and ability to adjust geometry with ISS maneuvers if required. This talk will walk through the experience of DESIS on the ISS, the pros and cons of using the ISS for imaging, show results generated in ENVI from calibrated DESIS data, and talk about the applications of medium resolution hyperspectral data relative to aerial or drone hyperspectral imagery.

Landcon: A 15 Small-Landsat Constellation to Study the Planet Earth

Tom George – SaraniaSat Inc.

The Landsat program is the longest-running enterprise for acquisition of satellite imagery of Earth. Begun in the late 1960s, the acquisition of moderate-resolution, multi-spectral imagery was a revolutionary concept. Subsequently, over its nearly five-decade history, Landsat data has achieved international usage across a diverse set of public/private sectors, including geologic mapping and exploration, precision farming, forestry, climate change, environmental monitoring, droughts, floods, earthquakes, tsunamis, infrastructure, urban planning, social media, water resources, and even the discovery of new species.

SaraniaSat Inc. is proposing a public/private partnership to NASA’s Sustainable Land Imaging (SLI) program for a “paradigm-shift” leading to the next revolution, namely the transition of the Landsat program from a single or dual satellite configuration to a 15 small-Landsat constellation or “Landcon” program incorporating several, exciting technological advances with an emphasis on generating global timely and actionable information for a broad range of sectors. To be actionable, our proposed “H3” paradigm-shift will involve hyper-spatial (10m), hyper-temporal (daily revisit), and hyper-spectral (402 narrow wavebands) resolution imagery. Generation of H3 imagery will result in a revolutionary and fundamental paradigm-shift in earth-observation, remote-sensing science principles from acquiring individual, spectral data points of current generation of satellites to acquiring complete spectral “signatures” of every phenomena being observed on the Earth.

Empowering Autonomous Driving Through High-Definition Location Data

Matthew Preyss - HERE Technologies

Location data and services are critical components for self-driving cars. The autonomous vehicle’s ability to understand its precise position, plan beyond sensor visibility, possess contextual awareness of the environment and gain local knowledge of the road rules are critical pieces of the autonomous driving puzzle, which relates to having highly precise location data. A critical factor for providing this data is the ability to collect, analyze and produce data with value at scale, and to do this continuously and as close to real-time as possible. Learn how HERE Technologies is at the forefront of high-definition mapping to support the evolution towards self-driving vehicles.

Tracking Nukes in the Open-Source Using Data Fusion

Melissa Hanham – Datayo

Datayo is an open-source collaborative data platform that collects data, cleans it, and translates it into a usable form. Looking at the international security sphere in general and the nuclear nonproliferation domain in particular, increased data access puts the power to monitor into the hands of all people. One of the challenges of nuclear risk reduction is marrying policy with data. Nuclear weapons are shrouded in secrecy. However, with increasing amounts of data available to the public, there are more opportunities for transparency -- if you can harness them.

Datayo seeks to close gaps between data scientists and those with domain knowledge by putting policy-relevant data into the hands of everyone in a simple, digestible format. We seek to break technical barriers by collecting and curating relevant publicly available datasets; providing a straightforward workspace for data visualization that doesn’t require coding; and a community-based approach to problem solving. We do the cleaning, prepping, and coding, so that civil society, international organizations, governments, journalists, and compliance officers can spend their time having more “Ah Ha!” moments. We build teams of expertise to tackle inter-disciplinary information analysis. We also partner with cutting edge tech and media companies, to foster inclusive, diverse access.

Geospatial Analytics for Anti-Money Laundering and Counter Threat Finance Investigations

Dan Steiner – Penn State University

Geospatial analytics improves the accuracy of risk assessments and reduces the time to identify suspicious activity. Combining descriptive, predictive, and prescriptive analytics identifies regions likely exposed to the flow of illicit funds, potential funding streams, and risks to nearby financial institutions (FIs). By leveraging the wealth of data available to FIs, and exploring additional physical variables, such as important rail and road networks, geospatial intelligence can be utilized to enhance the understanding of potential illicit funding corridors. Prescriptive analytics adds a previously unappreciated decision advantage to Anti-Money Laundering (AML) and Counter Threat Finance (CTF) investigations for the business, financial, and public safety sectors. Geospatial analysis connects the location and transaction attributes of customers and FIs to detect banking anomalies; potentially improving decision making and disrupting money laundering criminal activity. In geospatial analytics, coding and artificial intelligence (AI) automate routine tasks and elevate alerts, reports, and situational awareness. Advanced machine learning may be applied to identify high-risk procedural aberrations, false links, or patterns of fraud. Geospatial intelligence provides insights if transactions meet regulations or cumulative illicit activity skirts under thresholds. Proven geospatial analytic methods transform the efficiency of global commerce which is moving at high speed, undergoing great change.

Development of a Global Air Quality Satellite Constellation

Sheldon Drobot – L3Harris Corporation

Poor air quality affects 91% of the world’s population, leading to 4.2 million premature deaths every year. A key step in mitigating the impacts of poor air quality is to better understand local concentrations and dispersions of noxious gases. This presentation will discuss an innovative technology for providing sustainable, reliable, and consistent air quality observations via a global satellite constellation. Existing monitoring capabilities are typically ground-based stations funded by government and scientific sources for science-based missions. Further, most government ground stations are calibrated to report pollutant levels for each specific location, and estimates between stations can be wildly inaccurate. The new solution includes a 16-band sensor, operating in the visible and infrared portions of the electromagnetic spectrum. The bands are tuned to assess noxious gases, including methane, as well as track carbon dioxide

Applications of Airborne Imaging Spectrometer in Urban Environments - Opportunities and Challenges

Sunil Bhaskaran – CUNY

The presentation will go through the details of a project conducted in Sydney, Australia using airborne imaging spectrometer scanner. The study demonstrates the opportunities and challenges in using hyperspectral data to map features that are spectrally difficult to delineate.

Analytics and Machine Learning for Precision Agriculture and Green Infrastructure

Dan Staley – Arbor Drone LLC

Green Infrastructure is being recognized as an increasingly important component of societal resilience and key to livable cities efforts. Healthy green infrastructure provides food, makes cities more habitable, reduces urban heat, reduces building energy usage, and intercepts stormwater among many other benefits. How can decision-makers ensure healthy green infrastructure continues to provide benefits to the public? This presentation describes the process and outcomes of several innovative projects that used multispectral remote sensing and machine learning algorithms to successfully detect pest and disease outbreaks in urban forests, and determine orchard and vineyard health by utilizing analytics beyond those currently used in precision agriculture. Providing high-resolution analytics in near-time for green infrastructure means that decision-makers have the best information with which to make decisions about addressing pest and disease outbreaks, mitigating orchard or vineyard stress, planning for global change, or responding to severe weather damage. Attendees will learn about advanced machine learning analytics for precision agriculture and green infrastructure, challenges in urban data collection, and how these processes and techniques will change the face of orchardry, viticulture, and urban plant care. Data, outcomes, analytics and graphics from past and current projects will be shared.

Improving SmallSat Economics Creates More Analytics Opportunities

Jim Beckley – York Space Systems

As Investment dollars and scale continue to expand the space ecosystem; these act as an accelerant to the analytics market hence driving the demand for more and varied remote sensing data to feed hungry machine learning algorithms which automatically yield additional breadth and depth to your analytics products/services. This discussion provides current information on how and when remote sensing data users data can expect to benefit from said improvements in SmallSat economics as well as highlight a couple case studies where orbiting your own sensors/payloads potentially creates an unfair competitive advantage without needing to build, launch and operate your own constellation.

Data Acquisition through Geospatial Information Retrieval

John Farrell – L3Harris

Wouldn’t it be nice if you could say “Alexa, find labels and annotations at 39.7499° N, 104.9917° W, and show me the results on my map”. This may be possible one day using Geospatial Information Retrieval or GIR. GIR is a way to find and extract geographic information from documents, web pages and other sources of text. With GIR a geographer would no longer need to manually search the internet, separate out the information that describes the sought-after labels and annotations, convert it the proper format, and add it to a map. With GIR the request to Alexa could annotate a map with an event list at the Ritz-Carlton Denver and provide information about EAS 2019. This presentation will introduce the challenges of GIR (i.e. translating a textual description of a location into a point on a map) and will present a potential means to use Geospatial and Image processing to identify phrases associated with labels and annotations.

Driving Commercial Outcomes with Machine Learning and Global Daily Imagery

Annie Neligh – Planet

Previously, imagery coverage was incomplete and sporadic in cadence, making activity around the world impossible or difficult to observe. Planet's Dove constellation changed this, however traditional methods of imagery exploitation are extremely time-consuming and prone to human error, limiting the utility of this sizable dataset. Recent advances in machine learning and availability of cloud computing make rapid analysis of large geospatial datasets possible. Recent advances in change detection analytics identify changes to features and objects of interest as they occur making data-driven business decisions a reality that were impossible before. Energy companies are able to observe road and well-pad development of competitors. Financial organizations can monitor housing starts over large areas to inform trading activity. Insurance companies can optimize the deployment of adjusters in the aftermath of a natural disaster. This presentation examines how machine learning can answer these use cases, some of the challenges encountered and will dive into the techniques used to transform pixels into, commercially valuable information.

Applying AI/ML to Geospatial Engineering Tasks

Joseph Sawruk – US Army

The 64th Engineer Detachment supports US Army Special Operations Command and we are looking to use artificial intelligence and machine learning. Defining the problem statement for the machine to solve versus our human (geospatial engineer) is our most challenging obstacle. In this talk, I will discuss this as well as three other topics: 1) The ability to apply machine learning and artificial intelligence and its potential for significant growth over the next 3-5 years. 2) A Commander's perspective of the efficiencies gained for the workforce through employment of AI/ML. 3) The cloud migration over the next several years, our current roadblocks and their solutions.

Multispectral/Hyperspectral Imagery to Prevent Nuclear Proliferation in Rogue States

Grace Liu – James Martin Center for Nonproliferation Studies

We are arguably approaching an era where more and more states may be interested in pursuing nuclear weapons. While export controls limit the sharing of technology and knowledge related to nuclear weapons development, much of the science behind these weapons can be found in the open-source. Because nuclear weapons designs are easier to obtain, the most effective strategy to prevent new states from developing such weapons is to restrict the acquisition or production of fissile material. The production of fissile material is often a closely held state secret; even for states that report production levels to the IAEA.

Having a firm grasp on a state's production levels, capabilities, and activities can further inform whether a state has or is planning to develop and test nuclear devices. Uranium production can be difficult to verify without on-the-ground inspections. Multispectral and hyperspectral data can help verify uranium mining and milling activities and provide insight into a state's nuclear weapons stockpile. We did a proof-of-concept study on how MS and HS data can be used to monitor front-end uranium production and the potential for this to be implemented as a verification mechanism.

Method to Identify a Uniform Spatial Resolution for Multiple Raster Datasets

Sarah Becker – US Army Corps of Engineers

Often times image analysts have to grapple with comparing images of differing pixel sizes. Resampling to a common pixel size, or spatial resolution, may be necessary to perform certain types of statistical modeling but can introduce statistical errors or information can be lost during the resampling process. With the wrong common resolution, it is difficult to support things like land cover mapping, environmental quality studies, habitat and route modeling, mosquito/malaria migration mapping, and other use cases. Solution: Calculate semivariograms to find the ideal pixel size. The semivariogram technique determines optimal pixel sizes for individual image sets. Then datasets can be resampled as a whole, and Mean Absolute Errors can be calculated which then, in turn, can establish the best uniform spatial resolution for all of the resampled data sets. Once that ideal, uniform spatial resolution is determined it serves to guide future raster-based multivariate analysis and modeling.

Automated SAR Processing

David Valdez – NGA

NGA Research processes vast amounts of Sentinel SAR data for change detection products and extractions. This presentation will cover our best approaches for processing large amounts of SAR data as well as the products that we create and their applications.

US Marine Corps Analysis of Global Littoral Zones

Lawrie Jordan & Jamon Johnson – Esri

Please Join MSgt Mahoney (US Marines), Jamon Johnson (Esri Defense/Intel Product Manager), and Lawrie Jordan (Esri Director of Imagery) to learn about the US Marine Corps current analysis, production, and categorization of expeditionary points of entries of the global Littoral zone using advanced analytical techniques from remotely sensed data. In order to map and chart the worlds beaches, the Marine Corps turned to ArcGIS and ENVI to automate and scale production and analysis that supports the full range of military operations.

Weather Decision Support Using Big Data : A Surface Transportation Example

Sue Ellen Haupt – National Center for Atmospheric Research

Weather decision support is critical for effective operations in many realms, including surface transportation, aviation, energy, and many more. Such decision support systems require integrating real-time weather observations, historical observations, weather forecast data, and specialized data from the end user. Thus, these systems are inherently Big Data systems and include data from the Internet of Things (IoT). The most effective systems integrate both dynamical models and artificial intelligence (AI) systems, leveraging the best of each.

The National Center for Atmospheric Research has developed, tested, and deployed such decision support systems for many end users. As an example, we will describe in more detail one such system, the Pikalert® Road Weather Decision Support System. The Pikalert system is designed to blend real-time IoT vehicle-based observations with more traditional weather observations (e.g, radar or Road Weather Information System (RWIS) stations). The open-source Pikalert system produces frequently updated, high resolution assessments and forecasts of hazardous weather conditions, providing real-time situational awareness for companies, state DOTs, and the driving public. The system leverages weather forecasts that are optimally blended using AI techniques and we’re currently exploring using Harris’ Helios® as weather validation data feed.

Fusing SAR and Optical Information to Solve Geospatial Problems

Paolo Pasquali – sarmap

SAR vs. Optical: competitors or allies? Selecting one of these Earth Observation imaging technologies to investigate geospatial phenomena is often seen as a binary choice. However, the most efficient and successful method is to make use of the strengths of both technologies in a synergetic way. To do this, we need to understand the information that can and cannot be extracted from each data type, the contributions each can provide based on the application, and appropriate ways to fuse information. This talk will cover examples from real applications to highlight how today’s abundance of SAR and Optical data, if exploited properly, can help solve a large number of geospatial problems.

Getting Micro Weather Detection and Prediction Right for UAS Industry

Donald Berchoff – TruWeather Solutions

Micro weather significantly impacts the UAS industry. The uncertainty weather introduces creates either overly conservative flight plans, which increases cost and/or greater risk taking leading to more accidents and mishaps. This presentation will provide a framework to better detect and track micro weather hazards in real-time, and hours in advance, and translate this data into actionable impact knowledge for UAS. The weather data crowd-sourcing approach discussed will improve safety, reduce liability, preserve capital assets, and increase profitability per asset by improving micro-weather detection and prediction. It allows end-users to build flight and route plans that are more productive, cost effective, and safe. The solution includes leveraging and crowd-sourcing relevant IOT data sets and analytics, including Harris' Helios platform, to better detect weather threats where UAS fly, far from well weather instrumented airfields.

Breaking Through the Business Analytics Barrier

Matt Wood – Capella Space

For more than two decades, commercial remote sensing has been heralded as the great new source for business analytics information. This dream has not become a reality for many reasons. This presentation outlines some of the difficulties faced in the past and present, and discusses what needs to be done to break through the barrier to truly serve business analytics customers.

Build the Foundation First: Advanced Geospatial Analytics in Agriculture

Amariah Fuller – TerrAvion

There has been a huge investment in agtech and analytics over the last decade. Large companies that are already key players in the space are poised to benefit the most, potentially at the expense of the farmers and ag-consultants. But there's also an opportunity to democratize analytics for agriculture and help growers and agronomists benefit from these advances in technology. This talk will outline the five steps that are required to democratize analytics and AI for agriculture.

Making Complicated Data Simple, Making Sizable Data Practical

Megan Gallagher – L3Harris Technologies

New and improved satellites and amazing progress in UAV’s and camera technologies have led to an abundance of geospatial data. Ironically, we now have too much data, and it is hard to comprehend at the scale that we need. Monitoring things like forests, land displacement, weather, urban development, and many other applications of geospatial data have grown in breadth and scope. In order to effectively use geospatial data for these and other applications we need to address two specific items: 1) How do I harness all this geospatial data? 2) How can I get the answers that I need?

Cloud-Scale Geospatial Analytics

Shayn Hawthorne – AWS

New space systems, information, and analytics are transforming how space is applied to every aspect of our lives. Ubiquitous connectivity and access to data is making space accessible to more businesses and entrepreneurs around the world than ever before. This presentation explores how cloud computing services and features, combined with worldwide networks and near-real-time access to global space data, can transform how customers scale and use IoT and geospatial data to meaningfully impact business and benefit the world.

Deep Learning, Remote Sensing, and Utility Transmission and Distribution Asset Intelligence

Eric Spellman – Harris

In less than a decade, deep learning has had an enormous impact on many problems in many areas of life. This talk will address some of the challenges and opportunities in applying deep learning to problems in the utilities space – particularly making use of remotely sensed imagery and lidar collections of transmission and distribution infrastructure. Machine learning-based techniques have great promise here for a number of reasons including the periodic inspection cycles of assets and the rapidly growing economies of collection driven by sensor and platform improvements. However, this area also faces some challenges not found in many of the earlier applications of machine learning – particularly data sensitivity and aggregation. We will close with potential pitfalls and future directions.

Remote Sensing for Enhancing Global Security

Shane Deichman – Teledyne Brown Engineering

The 21st Century has seen exponential growth in humanity’s access to information. All facets of our lives – from education to commerce to security – have been affected by this “information revolution”. But our era is not unique: a review of human history reveals that every major evolution of civilization was spurred by some kind of “information revolution”, from the first spoken languages to physically-stored information, to mass-produced information, to today’s virtual worlds. Yet many institutions still rely on industrial-age processes, particularly with respect to strategic decision-making, that stymie the positive effects that remote sensing technologies can have on global security. In this presentation, the speaker will draw on more than three decades of experience in the international security marketplace to show how warfare has changed in the midst of technological advancements – and provide recommendations for how our dynamic technological ecosystem can be leveraged to protect life and property against a wide variety of threats.

Tropical Forest Tree Species Classification Using High-Resolution Imagery

Matthew Cross – University of Colorado Denver

Accurate identification of tropical tree species is critical for studies of forest habitat, composition, biomass, and ultimately a better understanding of the role trees play in climate variability through carbon uptake. The aim of this study was to derive an accurate tree species classification of a tropical forest study site in Costa Rica using high-resolution imagery. Data collected from a portable spectrometer provided the required ground truth for the study. A series of corrections for look and view angle, and the utilization of the DigitalGlobe atmospheric compensation procedure (AComp) provided the study with an accurate surface reflectivity dataset from WorldView-3 imagery. A rule-set object-oriented classification schema defined trees in the study area using ENVI®-defined tree canopies through a segmentation of the multispectral image. The results show that select WorldView-3 bands, and WorldView-3-specific vegetation indices, can produce an accurate species-level tree classification in a complex tropical forest.

Operating UAS Beyond Line of Sight - Fact or Fiction?

Jon Standley – L3Harris Technologies

As the UAS industry in the US continues to evolve, one aspect has remained as the ultimate prize to achieve – flying beyond visual line of sight. But is the UAS industry really ready to achieve this in the volume of operations that are forecasted, or will it remain reserved for the few who are able to invest the time and money necessary to develop the technology and safety case required for regulatory approvals. Hear examples of successful operations already being conducted under FAA Part 107 and what steps industry leaders are taking to increase access to this highlight sought after method of operation, while maintaining the safety and reliability that US aviation has come to expect from operators.

RDSC: A New Way to Move Forward in Data Analytics

Walter Johnson – Rochester Data Science Consortium

The Rochester Data Science Consortium is a New York state funded organization founded by the University of Rochester and the Harris Corporation to help NY companies to better use data science, machine learning, and artificial intelligence in their businesses. The model behind it is unique: We have a full-time staff of PhD data scientists with broad commercial experience who collaborate with regional institutions to understand and solve their most challenging big and little data science problems. Our scientists work full time on the analytical challenges faced by our members, from geospatial to retail to healthcare and more.

One Common Vision Determines Partnership Success

Sky Rubin - Airbus

Partnerships are at the heart of our business. To advance products and services in the geospatial industry, Airbus understands that we need to leverage the innovators and industry-leading experts to help drive action. These smart, strategic partnerships bring value to customers but can also be quite challenging. Us, along with many other companies have experimented with partnerships that have brought some success as well as some failure, but in the end, they have all provided value in some way. In this presentation, we will share a few examples of partnering success as well as lessons learned along the way, with the hope that learning from each other will help us all be better partners.

Business Challenges and Opportunities for Satellite Imagery Providers

EOS – Alex Sakal

Producers and users of satellite imagery face numerous challenges, from lack of fresh coverage or lack of imagery coverage in areas of need; to legal issues associated with the development of value-added products (VAP) and derivative products (DP); as well as challenges with the quality of the imagery during first operational years of satellites. In this session, EOS will discuss the creation of an open library of verified processing methods, and the establishment of an alliance between satellite operators and producers of VAP and DP to make the world a better place, and better serve the needs of all our customers.

The Evolution of Geospatial Analytics - A Pixel’s Tale

Jamie Goodman – HySpeed Computing

At its surface a pixel may appear simple, and yet behind this simplicity is an inherent complexity that needs to be unraveled to derive information and generate value. Whether through spectral analysis, data fusion or artificial intelligence, the evolution from raw data to actionable insights is a multifaceted process. Additionally, the ability to efficiently and accurately analyze data at scale is critically important for making informed decisions in a timely manner.

In this fast-paced market, data analysis needs are also becoming increasingly demanding - not only is the volume, variety and velocity of data intensifying, but so too is the resolution and temporal cadence. Addressing analysis needs across the many different business domains thus requires an array of data providers, algorithm developers, software platforms, analytic tools, business models and experienced analysts working in concert to derive decisive information from a complex data landscape. Recent years have seen an impressive surge in geospatial capabilities and commercial applications, but this is just the beginning. From the underlying scientific research to the development of operational algorithms, we use a collection of examples to explore the interconnected ecosystem that drives today’s geospatial marketplace.