Spatial Dynamics and Analysis of Crops using Super Multispectral Image Resolution and Radar Fusion
Abstract
As Earth’s population escalates, global demand for food is increasing at an alarming rate. Major challenges await global leaders in their efforts to secure stable and sufficient food supplies critical to the health and stability of their nations. Research efforts to support the spatial analysis (type, variety, growth, yield, and health) of agricultural crops and cropping patterns throughout the world are critical for understanding underlying factors and functional effects of the agricultural landscape. NGA’s ability to conduct efficient and accurate remote monitoring of world food production is critical to the safety of our nation. Accurate up-to-date GEOINT on the incidence, nature and causes of chronic food insecurity and vulnerability allows policy makers to formulate ideas, implement policies, and design programs to mitigate world food shortages. Remotely sensed (RS) imagery has been successfully used to monitor world food production since the early 1970s. Now, almost fifty years later, agronomists have access to many different RS sensors with increased spectral, spatial, and temporal resolutions. However, several challenges with RS imagery and crop analysis remain including the biophysical similarity of structures of most crops requiring extremely fine-resolution (< 0.5 m/pixel) multispectral imagery (MSI) for accurate identification, the mixing of spectral signatures within different crop types and within other types of vegetation, the amount of high variability in spectral signatures that occur during crop growth cycles and, the lack of optical RS sensor availability during weather events during key crop growth periods. To offset some of these issues, synthetic aperture radar (SAR) data has been successfully used to estimate soil moisture and plant heights. Although SAR data is extremely useful, researchers have found that combining information from multiple sources provides many advantages over a single RS sensor. To address these challenges researchers at the University of Alabama in Huntsville’s Information Technology and Information Center (UAH/ITSC) propose to extend their current research in deep learning and super image resolution to develop an improved method of pan-sharpening MSI and fusing the results with SAR data. Results will provide an enriched Geospatial Intelligence (GEOINT) product for conducting spatial analysis of agricultural crops. Once developed, this technology will be applicable to many other NGA domains including foundational feature extraction and environmental monitoring applications. This technology will greatly impact NGA’s mission by providing state-of-theart methods for combining a wide variety of RS data to solve complex problems. This proposal brings together a strong team of academic and research experts with a strong track record in conducting fundamental research in the areas of GEOINT, agronomy, image science, image analysis, deep learning, machine learning, computer science, and earth science: PI Dr. John Beck, GEOINT, agronomy, earth science and image analysis; Mr. Brent Atkinson, GEOINT and image science; Mr. Charles Collins, Dr. Susan Bridges, and Dr. John Rushing, computer science, artificial intelligence (AI), machine learning, and deep learning researchers.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 29, 2020
- Source ID
- HM04762010005
Entities
People
- John Beck
Organizations
- National Geospatial-Intelligence Agency
- University of Alabama in Huntsville