Archival Imagery Research Articles

Archived Jewishness: material practices of diaspora in Danzy Senna’s Caucasia

ABSTRACT Both Black and Jewish studies theorize archival memory, as well as archival gaps and silences, in regard to slavery and the Holocaust, respectively. Yet, these modes of theorizing the archiving have not been put into conversation. This article examines Danzy Senna’s 1998 novel Caucasia as a staging ground for a dialogue between archival theory in Black and Jewish studies. In considering archival imagery and themes in the novel, I argue that collecting, rearranging, and reassembling physical objects is Birdie’s primary way of negotiating not only her Blackness but also her imagined Jewishness as a way beyond racial performance.

Spatiotemporal Analysis of Urban Forest in Chattanooga, Tennessee from 1984 to 2021 Using Landsat Satellite Imagery

Chattanooga, Tennessee is one of many cities in the Southeastern United States that is experiencing rapid urban growth. As these metropolitan areas continue to grow larger, more and more of Earth’s unique temperate forest, an ecosystem of enormous cultural, ecological, and recreational significance in the Southeastern United States, is destroyed to make way for new urban development. This research takes advantage of the extensive temporal archive of multispectral satellite imagery provided by the Landsat program to conduct a 37-year analysis of urban forest canopy cover across the City of Chattanooga. A time series of seven Landsat 5 scenes and three Landsat 8 scenes were acquired between 1984 and 2021 at an interval of five years or less. Each multispectral image was processed digitally and classified into a four-class thematic raster using a supervised hybrid classification scheme with a support vector machine (SVM) algorithm. The obtained results showed a loss of up to 43% of urban forest canopy and a gain of up to 134% urban land area in the city. Analyzing the multidecade spatiotemporal forest canopy in a rapidly expanding metropolitan center, such as Chattanooga, could help direct sustainable development efforts towards areas urbanizing at an above-average rate.

Climate drivers and human impacts shape 35‐year trends of coastal wetland health and composition in an urban region

AbstractThe future of coastal wetlands will depend on the combined effects of climate change and human impacts from urbanization and coastal management. Disentangling the effects of these factors is difficult, but satellite imagery archives provide a way to track biological and physical changes in wetlands over recent decades to reveal how coastal wetlands have been changing in response to climate and human drivers. In this study, we used Landsat to monitor the conditions of 32 coastal wetlands in southern California from 1984 to 2019 and identify environmental and human drivers of these trends. Wetland conditions were characterized by vegetation greenness, using the normalized difference vegetation index (NDVI), and by habitat composition, derived from areal estimates of wetland and subtidal habitats. Overall, wetlands displayed three types of long‐term response: greening and gaining wetland (10), greening and losing wetland (16), and browning and losing wetland (6). Regional environmental drivers with overall positive effects on wetland NDVI were sea level, wave height, and precipitation, whereas stream discharge, vapor pressure deficit, and air temperature had negative or nonlinear effects. Wetland area change was primarily correlated with sea level, but response was highly contextual among sites. Negative trends in wetland NDVI and area were more common in larger sites with low elevations and in sites with open inlets. Restoration had mixed effects, with only half of the restored sites showing positive changes in NDVI and wetland area post‐restoration. The important work of managing and restoring urban coastal wetlands is complicated by variability and context and requires us to account for the influence of humans and climate as we build a regional understanding of historic, present, and future wetland health.

The Pairing of Rapid Intensification Events and Eyewall Replacement Cycles in Tropical Cyclones in the Atlantic Basin from 2015 to 2020

Rapid intensification (RI) and eyewall replacement cycles (ERCs) frequently occur in intense tropical cyclones (TCs), often causing rapid, significant changes in intensity and structure. In some TCs, RI and ERCs can occur concurrently or within a short period of one another. This study investigates whether there is a link between RI and ERCs by conducting a statistical analysis of TCs that occurred in the North Atlantic basin from 2015 to 2020. The HURDAT2 dataset was used to detect RI events, while the Morphed Integrated Microwave Imagery archive by the Cooperative Institute for Meteorological Satellite Studies was used to detect ERC events. Three sets of data were constructed from this analysis: TCs with paired RI/ERC events that occurred within 24 h of each other, TCs with RI only, and TCs with ERCs only. Statistics selected for analysis within the constructed datasets were mean duration of phenomena, mean rate of intensification, and mean peak intensity. We performed t-tests to determine the statistical significance of results. The results of this study show that TCs with these paired RI/ERC events often intensified at a faster rate, intensified for longer, and ended up stronger than TCs that only experienced RI or ERCs in isolation.

Channel change during catastrophic flood: Example of Storm Alex in the Vésubie and Roya valleys

Documenting and interpreting channel responses to catastrophic floods help understanding how rapid fluvial metamorphosis can propagate in a catchment under sediment cascading effects. The recent example of the October 2020 Storm Alex in SE France (∼500 mm of rain in 24 h) provides a unique opportunity to investigate major geomorphic responses along 70 km of two confined alpine valleys (Vésubie and Roya) and to link them to sediment wave initiation and propagation. GIS-based analysis of remote sensing data (high-resolution ortho-imagery and airborne LiDAR data) acquired before and after the flood allowed combining channel changes with sediment erosion and deposition along a 35-km reach of the Vésubie, including the most impacted portions of the valley. In the Roya, the analysis was restricted to 2D morphological changes reconstructed with the sequence of ortho-imagery. Archives of aerial imagery were also used to integrate the storm impact in the historical trajectory of the rivers. The reconstruction of geomorphic responses shows a quasi-continuous fluvial metamorphosis along the investigated stream networks, with dramatic active channel widening and aggradation, having no antecedent analogs during the last 70 years in both valleys. The different glacial imprints between the two valleys are considered a key factor explaining the exacerbated channel response in the Vésubie, where a braided channel emerged along a 35-km river length. Many evidences strongly support that the fuelling effect of alluvial storage is a key element of the sediment cascade at the origin of the braided channel formation. This regional case study allows us to discuss the critical role of floodplain and terrace erosion in the formation of the post-flood braided channel, and to compare the geomorphic impact of the storm with similar reported cases in the literature.

Manifestation of internal waves in the Southern Caspian Sea using satellite imagery

Iran's South Caspian coast is an important agricultural hotspot and densely populated area that attracts about 20 million domestic tourists in summer. Therefore, not only the coastal area ecosystem of the South Caspian Sea (SCS) but also their marine ecosystem is exposed to different environmental pollution. The ecosystem of the SCS especially in the onshore area strongly affected by Internal Waves (IWs) dynamics. IWs can capture, concentrate and transport, floating plastic debris, larvae, contaminants, and toxic pollutants in this basin. Therefore, detecting and zoning the SCS for IWs is critical to the management and protection of this enclosed basin. Measured data obtained from CTD and satellite imagery archives from Landsat 7, 8, and Sentinel-2 have been used to analyze and manifest IWs. Analysis and comparison of different satellite images reveal that the maximum wavelength of the IWs packet was about 400 m with a crest length of 18,000 m. However, the minimum packet had a wavelength of about 20 m with a crest length of 360 m. Based on the occurrence of IWs extracted by edge-finding algorithms on satellite images, the SCS can be divided into six original zones. Four zones of them located near the coastal line where freshwater rivers discharge to the sea in the rainy season and two other zones are in the deep water, which form under the dominant sea surface wind regime and sharp changing of bathymetry.

Unveiling the past: Utilizing satellite imagery archives to study archaeological landscapes in the northeastern Nile Delta, Egypt

This paper investigates the utilization of satellite imagery archives in the northeastern Nile Delta to examine and explore the evolution of archaeological landscapes. The study incorporates optical satellite imagery from multiple sensors, including Landsat, ASTER, and CORONA, in addition to historical topographic maps, and digital elevation data. The methodology involves employing Support Vector Machine (SVM) algorithm, spatial analysis, and analyzing vegetation indices. The findings reveal significant landscape changes during the period from the prehistoric to the Byzantine era, influenced by natural factors such as sea level changes and land subsidence. Moreover, the landscape changed after the construction of the Aswan High Dam, influenced by human activities such as agriculture and urbanization. The study was successful in detecting the remains of the Tanitic branch, an ancient branch of the Nile River. Additionally, the analysis of vegetation indices showed crop anomalies, which may be indicative of buried structures. However, further investigations, including ground surveys and excavations, are needed for validation. The research contributes to the understanding of the region's historical and cultural significance, highlights the impact of human activities on archaeological sites, and underscores the importance of preserving cultural heritage in the face of environmental changes and urban development.

Modeling forest canopy surface retrievals using very high-resolution spaceborne stereogrammetry: (I) methods and comparisons with actual data

The capability of spaceborne stereogrammetry using very high-resolution (VHR, <2 m) imagery with various environmental, experimental, and sensor configurations for characterizing forest canopy surfaces has not been completely explored. Existing archives of VHR imagery include a limited subset of potential stereo image acquisition configurations and may therefore exclude optimal configurations for capturing critical structural features of forest canopy surface. By contrast, simulated VHR imagery from 3-D radiative transfer models (RTM) can explore the full range of spatial, spectral, and sun-sensor configurations to identify factors that contribute to uncertainties in stereo-derived estimates of forest canopy structure. We developed a novel method to simulate VHR stereopairs using the discrete anisotropic radiative transfer (DART) model and then derive surface elevations from the simulated images. We reconstructed one open-canopy and one closed-canopy forest scene and created a reference digital surface model/digital terrain model (DSM/DTM) using airborne small-footprint lidar points over the study sites. The VHR simulations were configured to match three independent WorldView stereopairs. The results showed that, compared to the reference DSM, the surface elevations derived using simulated and WorldView image data were consistent, with differences of <1.6 m in vertical bias, < 1 m in root mean square error (RMSE), and < 0.07 in correlation coefficient (R). We demonstrated that realistic 3-D RTM simulations could be georeferenced with a camera model for DSM generation from simulated stereopairs. This work will support a follow-up investigation that examines stereo-derived DSM quality over a broad range of surface types and acquisition parameters to suggest optimal configurations for actual VHR stereo data acquisition of vegetation canopy surfaces.

Suitability of terrestrial archival imagery for SfM-MVS based surface reconstruction of steep rock walls for the detection of rockfalls

Suitability of terrestrial archival imagery for SfM-MVS based surface reconstruction of steep rock walls for the detection of rockfalls

Wetlands Insight Tool: Characterising the Surface Water and Vegetation Cover Dynamics of Individual Wetlands Using Multidecadal Landsat Satellite Data

Wetlands around the world provide crucial ecosystem services and are under increasing pressure from multiple sources including climate change, changing flow and flooding regimes, and encroaching human populations. The Landsat satellite imagery archive provides a unique observational record of how wetlands have responded to these impacts during the last three decades. Information stored within this archive has historically been difficult to access due to its petabyte-scale and the challenges in converting Earth observation data into biophysical measurements that can be interpreted by wetland ecologists and catchment managers. This paper introduces the Wetlands Insight Tool (WIT), a workflow that generates WIT plots that present a multidecadal view of the biophysical cover types contained within individual Australian wetlands. The WIT workflow summarises Earth observation data over 35 years at 30 m resolution within a user-defined wetland boundary to produce a time-series plot (WIT plot) of the percentage of the wetland covered by open water, areas of water mixed with vegetation (‘wet’), green vegetation, dry vegetation, and bare soil. We compare these WIT plots with documented changes that have occurred in floodplain shrublands, alpine peat wetlands, and lacustrine and palustrine wetlands, demonstrating the power of satellite observations to supplement ground-based data collection in a diverse range of wetland types. The use of WIT plots to observe and manage wetlands enables improved evidence-based decision making.

Seasonal variability of characteristics of nonlinear internal waves in the Kuril-Kamchatka region by Sentinel 1 data

The seasonal variability of the surface manifestations of nonlinear internal waves (NIW’s) in the Kuril-Kamchatka region from December 2018 to November 2019 was investigated based on the analysis of the Sentinel 1 satellite imagery archive. On 827 satellite images, 1876 manifestations of NIW’s were registered with a wavelength from 80 to 1900 m, containing from 2 to 18 waves in a packet. It was identified that the number of NIW’s registered in the water area varied significantly during the year. The minimum number of manifestations was recorded from January to March, and the maximum number of manifestations was observed from July to September. Areas of the frequent occurrence of NIW’s manifestations were identified: near the Kunashir Island, over the Vityaz ridge, near Cape Shipunsky and in Kamchatka Bay. Manifestations of NIW’s in these areas were recorded in all seasons of the year. These areas accounted for about 53% of all registered manifestations of NIW’s, and their area was about 20% of the study region area. The calculation of the tidal body force criterion for the main diurnal and semidiurnal tidal harmonics made it possible to identify potential generation sites of internal tidal waves. About 70% of NIW’s manifestations in August were detected in areas close to potential generation sites of diurnal internal tidal wave K1. This is explained by closeness of the generation and the disintegration sites of internal tidal waves under conditions of being beyond the critical latitude, where the tidal frequency coincides with the inertial one. Using the example of the area above the Vityaz ridge, the connection between the seasonal variability in the number of recorded manifestations and the annual variations in the tidal body force for harmonics K1, O1 and P1, mixed layer depth, the surface wind speed and the Brunt-Vaisala frequency was shown. This is consistent with the results of studies carried out in the adjacent regions of the Pacific Ocean. The existing archives of the Sentinel-1 radar image and the continuation of this satellite mission in the future will make it possible to estimate the long-term variability of NIW’s characteristic and to assess the impact of climate change on them.

Post-accident dynamics of flora and vegetation in the Belarusian sector exclusion zone of the Chernobyl nuclear power plant

The issues of vegetation cover dynamics in the Chernobyl accident area considered in the article. The research carried out within the boundaries of the Polessky State Radiation-Ecological Reserve. The objects of research are the flora and vegetation of the region. The research is based on as classical methods for studying flora and vegetation, as modern ones, using a long-term archive of Landsat satellite imagery. The article interprets that the flora and vegetation transformation processes of the Chernobyl exclusion zone are mainly due to natural changes and anthropogenically stimulated processes associated with the economic activity cessation. The authors noted a progressive increase of the vascular plant species number within the Reserve, which explained by the restoration processes beginning of the vegetation cover after the population evacuation and the economic activity complete stopping. The spatial isolation of the Polessky State Radiation-Ecological Reserve territory led to the preservation of flora synanthropization level, similar to the end of the 1980s. Former settlements, wastelands, former arable lands, linear objects (roads) and other former anthropogenic habitats are the retention and penetration centers of synanthropic species within the Reserve. The forested territory increasing is a key moment of the vegetation demutational successions in the Chernobyl exclusion zone. A large increase of derived small-leaved forests participation characterizes the dynamics of the forest vegetation formational structure during the post-accident period. Indigenous coniferous and deciduous fen forests also tend to distribution increase, but their dynamics is not rapid. Forests formation, bushing, waterlogging due to the degradation of drainage systems are the processes of meadow dynamics. Fragments of unchanged meadows preserved only in the Pripyat River floodplain. Restoration of bog-forming processes observed on disturbed peatlands, part of unforested fens transformed into forested fens, sparse forests and shrub thickets in the post-accident period. The processes of forest formation, local swamping prevail on the former agricultural and residential lands.

Spotting A Tree From A Pixel (With Remote Sensing Researchers)

This exposition contemplates the collaboration between me, a photographer, and remote sensing researchers from the Department of Geoinformatics at Aalto University, in an ongoing artistic research collaboration called Ground Truth. Ground Truth is a photography project about ‘seeing something when there’s nothing there’. Based in the research group’s initial intent to overcome the spatial resolution limits of satellite imagery, the project now investigates new imaging techniques such as computational photography and hyperspectral imaging of forests, while also referencing photography’s love affair with natural landscapes. Such a comparative approach to natural photography has allowed me to offer a vision that typically escapes human sight perception. I open this auto-theoretical text with a personal experience, namely being on a field trip with remote sensing scientist Daniel Schraik. I use this moment to, among other things, articulate my thoughts on the construction of the ‘real’ in different disciplines. I then contemplate the body of interviews and field trips that became a two-year-long interdisciplinary dialogue, and also brought together two distinct ways of looking at the forest: one symbolized by the camera, another by the terrestrial laser scanner. Inspired by remote sensing concepts such as ground truth and the inverse problem, I have come to examine photography through a new analytical framework. In everyday language, the term ‘ground truth’ refers, in part, to a first-hand experience. In our project, it makes sense, then, that Ground Truth connotes the documentary tradition and the act of witnessing. In the language of science, however, and specifically in remote sensing as a field, ground truth means something different. It refers to data collected on-site, which can then be used to calibrate, to build models, to predict, to interpret, and to decipher information from images; in this case, satellite images. Similarly, our interdisciplinary collaboration functions on another operational layer of photography beneath the immediately visible, one that illustrates an expanded notion of photography across contemporary discourse. Ground Truth interweaves archival imagery, documentary photography, experiment dataset, 3D digital art and conceptual photography. The constellation of employed elements contrasts the representational approaches of drawing and photography with the data-oriented and algorithmic approaches of computer-aided seeing. The two modes together allow for a parallel reading of the forest — one that contextualizes different epistemological regimes that allow for new configurations of the relationship between image and reality.

Discovery of TOI-1260d and the characterization of the multiplanet system

ABSTRACT We report the discovery of a third planet transiting the star TOI-1260, previously known to host two transiting sub-Neptune planets with orbital periods of 3.127 and 7.493 d, respectively. The nature of the third transiting planet with a 16.6-d orbit is supported by ground-based follow-up observations, including time-series photometry, high-angular resolution images, spectroscopy, and archival imagery. Precise photometric monitoring with CHEOPS allows to improve the constraints on the parameters of the system, improving our knowledge on their composition. The improved radii of TOI-1260b and TOI-1260c are $2.36 \pm 0.06 \rm R_{\oplus }$, $2.82 \pm 0.08 \rm R_{\oplus }$, respectively while the newly discovered third planet has a radius of $3.09 \pm 0.09 \rm R_{\oplus }$. The radius uncertainties are in the range of 3 per cent, allowing a precise interpretation of the interior structure of the three planets. Our planet interior composition model suggests that all three planets in the TOI-1260 system contains some fraction of gas. The innermost planet TOI-1260b has most likely lost all of its primordial hydrogen-dominated envelope. Planets c and d were also likely to have experienced significant loss of atmospheric through escape, but to a lesser extent compared to planet b.

Historical Structure from Motion (HSfM): Automated processing of historical aerial photographs for long-term topographic change analysis

Precisely measuring the Earth’s changing surface on decadal to centennial time scales is critical for many science and engineering applications, yet long-term records of quantitative landscape change are often temporally and geographically sparse. Archives of scanned historical aerial photographs provide an opportunity to augment these records with accurate elevation measurements that capture the historical state of the Earth surface. Structure from Motion (SfM) photogrammetry workflows produce high-quality digital elevation models (DEMs) and orthoimage mosaics from these historical images, but time-intensive tasks like manual image preprocessing (e.g., fiducial marker identification) and ground control point (GCP) selection impede processing at scale. We developed an automated method to process historical images and generate self-consistent time series of high-resolution (0.5–2 m) DEMs and orthomosaics, without manual GCP selection. The method relies on SfM to correct camera interior and exterior orientation and a robust multi-stage co-registration approach using modern reference terrain datasets for geolocation refinement. We demonstrate the method using scanned images from the North American Glacier Aerial Photography (NAGAP) archive collected between 1967 and 1997. We present results for two sites with variable photo acquisition geometry and overlap — Mount Baker and South Cascade Glacier in Washington State, USA. The automated method corrects initial camera position errors of several kilometers and produces accurately georeferenced, high-resolution DEMs and orthoimages, regardless of camera configuration, acquisition geometry, terrain characteristics, and reference DEM properties. The average RMS reprojection error following bundle adjustment optimization was 0.67 px (0.15 m) for the 261 images contributing to 10 final DEM mosaics between 1970 and 1992 at Mount Baker, and 0.65 px (0.13 m) for the 243 images contributing to 18 individual DEMs between 1967 and 1997 at South Cascade Glacier. The relative accuracy of elevation values in the historical time series stacks was 0.68 m at Mount Baker and 0.37 m at South Cascade Glacier. Our products have reduced systematic error and improved accuracy compared to DEM products generated using SfM with manual GCP selection. Final elevation change measurement precision was ∼0.7–1.0 m over a 30-year period, enabling the study of processes with rates as low as ∼1-3 cm/yr. Our results demonstrate the potential of this scalable method to rapidly process archives of historical imagery and deliver new quantitative insights on long-term geodetic change and Earth surface processes.

The future of coastal monitoring through satellite remote sensing

AbstractSatellite remote sensing is transforming coastal science from a “data-poor” field into a “data-rich” field. Sandy beaches are dynamic landscapes that change in response to long-term pressures, short-term pulses, and anthropogenic interventions. Until recently, the rate and breadth of beach change have outpaced our ability to monitor those changes, due to the spatiotemporal limitations of our observational capacity. Over the past several decades, only a handful of beaches worldwide have been regularly monitored with accurate yet expensive in situ surveys. The long-term coastal-change data of these few well-monitored beaches have led to in-depth understanding of many site-specific coastal processes. However, because the best-monitored beaches are not representative of all beaches, much remains unknown about the processes and fate of the other &gt;99% of unmonitored beaches worldwide. The fleet of Earth-observing satellites has enabled multiscale monitoring of beaches, for the very first time, by providing imagery with global coverage and up to daily frequency. The long-standing and ever-expanding archive of satellite imagery will enable coastal scientists to investigate coastal change at sites vulnerable to future sea-level rise, that is, (almost) everywhere. In the past decade, our capability to observe coastal change from space has grown substantially with computing and algorithmic power. Yet, further advances are needed in automating monitoring using machine learning, deep learning, and computer vision to fully leverage this massive treasure trove of data. Extensive monitoring and investigation of the causes and effects of coastal change at the requisite spatiotemporal scales will provide coastal managers with additional, valuable information to evaluate problems and solutions, addressing the potential for widespread beach loss due to accelerated sea-level rise, development, and reduced sediment supply. Monitoring from Earth-observing satellites is currently the only means of providing seamless data with high spatiotemporal resolution at the global scale of the impending impacts of climate change on coastal systems.

A method for considering the evolution of the visible landscape

The visible landscape represents an important consideration within landscape management activities, forming an inhabitants’ perception of their overall surroundings and providing them with a sense of landscape connection, sustainability and identity. The historical satellite imagery archive can provide key knowledge of the overall change in land use and land cover (LULC), which can inform a range of important management decisions. However, the evolution of the visible landscape at a terrestrial level using this information source has rarely been investigated. In this study, the Landsat archive is leveraged to develop a method that depicts changes within the visible landscape. Our method utilises other freely available data sources to determine the visibility of the landscape, and LULC composition, visible from road networks when the imagery was captured. This method was used to describe change in the visible landscape of a rural area in Ñuble, Chile, in the period from 1986 to 2018. Whilst native forests on the slopes of the mountains within the study area provide a natural backdrop, because of the flat topography of most of the area, the foreground dominates the overall landscape view. This has resulted in a visible transition from a landscape visibly dominated by agricultural use in 1986 to one of equal agriculture and plantation forestry in 2018. It is hoped that the method outlined within this study can be applied easily to other regions or at larger scales to provide insight for land managers regarding the visibility of LULC.

A Python toolkit to monitor sandy shoreline change using high-resolution PlanetScope cubesats

This study evaluates an emerging capability to monitor high spatial (pixel size = 3.7 m) and temporal (daily to sub-daily imagery) resolution coastal change using PlanetScope cubesats. A new toolkit (CoastSat.PlanetScope) is presented that enables users to map shorelines from PlanetScope imagery, using a workflow of image co-registration, segmentation, thresholding, shoreline detection and elevation correction. The toolkit is subsequently tested at Narrabeen-Collaroy Beach (SE Australia), evaluating combinations of shoreline detection indices, thresholding approaches and elevation correction. An optimal shoreline accuracy of 3.5 m (RMSE) is found for this coastline using the difference between the near-infrared and blue bands and a weighted peaks thresholding approach. A generic elevation correction model that considers tidal variability and wave setup at the shoreline is then proposed. With a growing archive of high-resolution imagery, PlanetScope presents enormous potential for enhanced coastline mapping of the coast and complements existing approaches using Landsat/Sentinel-2 imagery.

Enhancing performance of multi-temporal tropical river landform classification through downscaling approaches

ABSTRACT Multi-temporal remote sensing imagery has the potential to classify river landforms to reconstruct the evolutionary trajectory of river morphologies. Whilst open-access archives of high spatial resolution imagery are increasingly available from satellite sensors, such as Sentinel-2, there remains a fundamental challenge of maximising the utility of information in each band whilst maintaining a sufficiently fine resolution to identify landforms. Although image fusion and downscaling methods on Sentinel-2 imagery have been investigated for many years, there is a need to assess their performance for multi-temporal object-based river landform classification. This investigation first compared three downscaling methods: area to point regression kriging (ATPRK), super-resolution based on Sen2Res, and nearest neighbour resampling. We assessed performance of the three downscaling methods by accuracy, precision, recall and F1-score. ATPRK was the optimal downscaling approach, achieving an overall accuracy of 0.861. We successively engaged a set of experiments to determine an optimal training model, exploring single and multi-date scenarios. We find that not only does remote sensing imagery with better quality improve river landform classification performance, but multi-date datasets for establishing machine learning models should be considered for contributing higher classification accuracy. This paper presents a workflow for automated river landform recognition that could be applied to other tropical rivers with similar hydro-geomorphological characteristics.

Undercover Algorithm: A Secret Chapter in the Early History of Artificial Intelligence and Satellite Imagery

In the second decade of the twenty-first century, computer algorithms accurately and rapidly identify features of some objects on digital satellite imagery. These feature-recognition algorithms are expected to transform geospatial intelligence: to enable rapid retrospective searching of imagery archives and focus prospective analytic attention. This Perspective article establishes the beginning of the U.S. Intelligence Community’s research and development on this capability. Archival research of declassified Central Intelligence Agency documents produced two discoveries: one identifies that the earliest feature-recognition initiative predated the creation of the National Photographic Interpretation Center (NPIC) in January 1961. The other discovery reveals that the earliest neural network software, Frank Rosenblatt’s Mark I Perceptron, from which current feature-detection software descends, had been part of a previously secret four-year NPIC effort from 1963 through 1966 to develop this algorithm into a useful tool for photo-interpreters. The manager of that research effort, John Cain, defined the prospective utility of this software in 1963, and Cain’s criteria, derived from NPIC’s experience during the Cuban Missile Crisis, continues to shape the prospective geospatial intelligence uses of feature-recognition software.