Experts in the field of image analysis and classification will present applications of single and fused data sets for mapping and monitoring vegetation, accuracy assessment considerations, and how these data are used in decision making.

2:00 PM – 2:15 PM – Automation in Producing High-Resolution Land Cover

The production of high-resolution land cover is a challenging topic. While humans can quickly review an image and determine a land cover call, using a computer to do the same can be a complicated task. High-spatial resolution imagery can result in a single object (building, tree, field) in the landscape being made up of 100’s of pixels, including different colors, textures, and shading with complex feature boundaries. Computers Traditional statistical modelling approaches, such as support vector machines (SVM), can struggle to understand the context of objects within a broader landscape image, thus resulting in confused, less useful classifications. More advanced approaches, including Artificial Intelligence and Deep Learning, are an attempt to overcome some of these issues, and rely heavily on accurate training data, and computer processing intensive algorithms. Object-based classification, using expert rules, or machine learning algorithms are other approaches which can be used alone or in tandem with Deep Learning. Dewberry is developing high-resolution NOAA Coastal Change Analysis Program land cover data for eight counties in coastal Mississippi and Alabama. These 1-meter data sets are being derived through a combined approach, leveraging both Deep Learning and Object-based, expert rules systems, taking advantages from both methodologies. The Deep Learning model has been trained on existing, older C-CAP data and refined using locally relevant data sets. The outputs are being incorporated into an object-based rule set to refine the output. Compared with past methods, this combined approach has allowed for the production of accurate land cover over large areas, resulting in a reduced timeframe for the production of accurate land cover over large areas, as compared to past methods.

Kyle Barnes, Dewberry

2:15 PM – 2:30 PM – The Damage Assessment of 2019–2020 Australia Bushfire to the Endangered Ecological Communities in Eurobodalla Shire Using Object-Based Classification

The 2019-2020 bushfire damaged many regions in New South Wales of Australia. The Eurobodalla shire is one of the local governments located in New South Wales. The Eurobodalla shire is important area in the state due to its high values in conservation, tourism, agriculture and timber production, and it includes many endangered ecological communities. However, 2019-2020 bush fire damaged Eurobodalla Shire including the endangered ecological communities of the area. Obviously, the quantification of the damaged area of endangered ecological communities is very important for the conservation efforts in this area. Therefore, the research objective of this study is to analyze and quantify the damaged areas of endangered ecological communities of Eurobodalla Shire. With this research objective, change detection was implemented using the object-based classification. Recent development of commercial GIS software (ArcGIS Pro) made object-based classification workflow very efficient. The Sentinel-2 imagery captured before and after the 2019-2020 bushfire was used as input data for the object-based classification. By comparing the classification results of pre- and post-bushfire images, it was found many land surface features were affected by the bushfire. For example, approximately 64 percent forest classes in Eurobodalla Shire were damaged. The burn scar areas identified by the object-based classification was used to delineate the damaged areas of the endangered ecological communities. In the result, many endangered ecological communities were found to be damaged. Among them, the Lowland Grassy Woodland in the South East Corner Bioregion was damaged most severely to the area of 47.881 square km.

Dylan Anderson, Sanborn Map Company

Jae Sung Kim, Michigan Technological University

2:30 PM – 2:45 PM – Unlocking the Past: Preserving and Digitizing Historical Aerial Film and Media

From the 1950s through the early 2000s, the primary method for collecting remotely sensed geospatial data was through large-format aerial film. These films—typically stored in roll format—are now housed in hundreds of private and local government collections across the United States. Additional media types, including mylar, small-format film, and photographic prints, are also preserved in libraries, archives, and other institutions.

Despite their high resolution (often surpassing that of contemporary satellite imagery or federal programs of the time) these materials remain underutilized due to the challenges of digitization. Their potential value spans a wide range of applications, including change detection, climate research, and legal investigations.

To address this gap, the ASPRS Data Preservation and Archive Committee (DPAC) has released comprehensive guidelines for the digitization of aerial film and related media. This presentation will:

Highlight key components of the new digitization guidelines
Discuss strategies for prioritizing collections for preservation
Explore methods for extracting value from digitized data
Share resources and best practices for long-term preservation
Provide guidance on connecting with agencies and organizations that offer digitization services or accept archival donations
Connections/Resources with existing digitized data ready for use

David Day, Vexcel Imaging, US

2:45 PM – 3:00 PM – Modernize Your Gis With Imagery and Remote Sensing

The Geographic Information System (GIS) of the future is expected to be more interconnected, intelligent, and user-friendly. It should empower individuals and organizations to make data-driven decisions in a rapidly changing world. At a minimum, organizations should be looking now to develop and incorporate cloud-based solutions, 3D visualization and modeling, and artificial intelligence technologies. A critical component of every modern GIS is a consolidated imagery repository that leverages high performance computing for scalability, accessibility, and collaboration. This presentation will provide key components to consider and present a roadmap for organizations looking to evolve their current GIS systems.

Stephanie Dockstader, Esri

3:00 PM – 3:15 PM – An Approach for Automated Change Detection From Satellite Imagery Spectra

Increased spectral content in satellite-based remote sensing necessitates new methods to produce wide area search capabilities. Some current linear efforts rely on spatial layer differencing, band ratios, and index combinations which do not utilize the full spectra collected. Such methods introduce errors from compounding detection estimates and bias change with simplified pixel counting.

This talk outlines a graph approach for agnostic change using passive spectral sensing to overcome (a) undirected graph input size, (b) time inefficiencies by analyzing “no change” areas, and (c) simplistic estimations in spectral detection.

Discussions will cover the approach, some satellite and airborne spectral examples, as well as challenges and ways forward for integration/deployment.

David Hughes, UT-Battelle

3:15 PM – 3:30 PM – High Resolution Historic and Commercial Low to No-Cost Satellite Imagery – Access and Applications

High-resolution historic and commercial satellite imagery are available through the U.S. Geological Survey’s Earth Resources Observation and Science (EROS) Center and the National Civil Applications Center (NCAC).

Historic high-resolution imagery derived from declassified Keyhole 1-9 (KH1-KH9) reconnaissance satellites covering the early 1960’s to the mid 1980’s are available at no-to low-cost to users through EROS’s Earth Explorer website. These data provide near global annual coverage and constitute a unique imagery dataset useful for land use-land cover change detection and feature detection during a period when satellite imagery was either unavailable or of low spatial resolution.

Commercially available high resolution imagery data is available at no-cost to Federally affiliated users through the CIDR (Commercial Remote Sensing Space Policy (CRSSP) Imagery Derived Requirements) tool hosted by EROS and adjudicated by NCAC. These data are made available through commercial contracts supported by the National Reconnaissance Office. The data available through these contracts includes electrooptical, multispectral, and radar imagery and non-imaging radio frequency data.

Peter Rinkleff, U.S. Geological Survey National Civil Applications Center