Source: UNIVERSITY OF FLORIDA submitted to NRP
USING REMOTE SENSING SENSORS AND GEOGRAPHIC INFORMATION ANALYSIS FOR NATURAL RESOURCE AND AGRICULTURAL APPLICATIONS
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
Reporting Frequency
Annual
Accession No.
1020250
Grant No.
(N/A)
Cumulative Award Amt.
(N/A)
Proposal No.
(N/A)
Multistate No.
(N/A)
Project Start Date
Jul 27, 2019
Project End Date
Jul 24, 2024
Grant Year
(N/A)
Program Code
[(N/A)]- (N/A)
Recipient Organization
UNIVERSITY OF FLORIDA
G022 MCCARTY HALL
GAINESVILLE,FL 32611
Performing Department
Gulf Coast Research and Education Center
Non Technical Summary
With the current technological advances in remote sensing imaging technologies, the need for methods to integrate such technologies and maximize their use is heightened. Multi-sensor technologies can provide different sources of data to account for the tradeoffs of each data type and target specific types of applications. Analysis of the spectral and geometrical information extracted from remote sensing data to, for example, compute plant traits, predict crop yield, assess disease damage, monitor restoration efforts has been the subject of significant research in the past decade (e.g., Abdulridha et al., 2019, Torres-Sánchez et al., 2013. Pande-Chettriet al., 2017). Recent efforts by Liu and Abd-Elrahman showed significant improvement in land cover classification of images taking by unmanned aircraft system using deep learning technologies and multi-view image analysis (Liu & Abd-Elrahman, 2018, Liu et al., 2019).The need to establish the relationship between ground objects and spectral measurements is of interest not only for the geospatial analysis scientific communitybut also to receiving stakeholders such as crop growers or natural resource managers. Being able to utilize spectral data for the early identification of invasive plant species or optimizing the use of fertilization in agricultural lands could save the economy billions of dollars, not to mention the positive impact on human health and the environment. Fortunately, recent advances in spatial data acquisition technologies including multispectral and hyperspectral imagery, lidar datasets and their acquisition platforms (e.g., space-borne or UAS-based systems) made the acquisition and use of this information more feasible and promising. However, in order to facilitate the use of such technologies to answer specific research questions, the research community needs to develop the methodologies required for data acquisition system integration, experimental design, and analysis algorithms.Improving remote sensing data acquisition tools, methodologies, and analysis algorithms for agricultural applications and land cover characterization continues to be the main focus of this research effort. In this context, this project aims todevelop and use integrated multi-sensor spatial data acquisition techniques that include hyperspectral, multispectral, lidar, positioning and attitude data acquisition sensors to analyze land cover (including vegetation) response. This research will help close the gap in understanding and quantifying spatial changes of, for example, vegetation and soils. Furthermore, successful development could lead to integrated multi-sensor platform that provide a suite of data and enables applied and basic research opportunities in the precision agriculture and natural resource monitoring fields.
Animal Health Component
70%
Research Effort Categories
Basic
(N/A)
Applied
70%
Developmental
30%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
2051122209050%
4027210208025%
1230199107025%
Goals / Objectives
The overall goal of this proposal is to advance scientific discoveries in the agriculture and natural resource monitoring/management field using different remote sensing systems and image analysis methods. To achieve this goal, the following objectives are identified:?1. Design experiments and develop methods to collect and analyze remote sensing datasets2. Use spatial data acquisition sensors and techniques (including multispectral and/or hyperspectral imagery) for agricultural applications 3. Use spatial data acquisition sensors and techniques (including multispectral and/or hyperspectral imagery) for land cover characterization
Project Methods
The following procedures are proposed to accomplish project objectives:Design experiments and develop methods to collect and analyze remote sensing datasetsSystems that combine hyperspectral, multispectral, and/or lidar sensors are subject to continuous improvement. Tighter integration of the system components for effective integration is proposed. This process involves acquiring the sensors, developing new integration techniques, implementing system calibration and testing steps.System integration involves integrating different remote sensing acquisition systems. For example, across platform integration involves utilizing ground-based system in combination with sensors onboard UAS. Such integration overcome many of the limitations associated with each individual system leading to diversified datasets that enable more applications and analysis methods.Use spatial data acquisition sensors and techniques (including multispectral and/or hyperspectral imagery) for agricultural applicationsThis objective aims at designing field experiments to collect spatial data using different imaging sensors. The data should be augmented by establishing ground control locations and spectral calibration techniques. The design includes identifying management procedures, frequency of observations and accompanied reference lab measurements. It is envisioned that the results these techniques will lead to improvement in several agricultural applications such as crop yield modeling, disease detection, and phenotyping.Use spatial data acquisition sensors and techniques (including multispectral and/or hyperspectral imagery) for land cover characterizationDifferent types of remote sensing imagery will be analyzed. Image correction algorithms will be implemented. Multispectral, hyperspectral, and/or lidar data analysis techniques such as dimensionality reduction using Minimum Noise Fraction Transform algorithm (Chen, et al., 2003; Pande-chhetri & Abd-Elrahman, 2013), statistical modeling, object based analysis, and machine learning algorithms are proposed. The results will be analyzed and compared to ground truth data for modeling and assessment. It is envisioned that this objective will lead not only to better methods for land cover (e.g. vegetation, soils, water quality, etc.) characterization, but also to the introduction of more efficient techniques to achieve these results.

Progress 07/27/19 to 09/30/19

Outputs
Target Audience:The audience of this phase of the project is the agricultural scientificcommunity interested in using ground-based and unmanned aircraft systems in phenotyping and biophysical parameter modeling. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The project trained two graduate students and one undergraduate student on methods to analyze high spatial resolution imagery. One graduate student received his part 107 remote pilot license to fly small unmanned aircraft systems. Several presentations were conducted as oral technical presentations or instructional workshops How have the results been disseminated to communities of interest?- A workshop on using deep learning in image classification was developed and presented to GIS specialists in the American Society of Photogrammetry and Remote Sensing Conference. - The developed image analysis methods were presented (oral presentations) in theAmerican Society of Photogrammetry and Remote Sensing Annual Conference (June 22-26, 2020), the American Society of Horticulture Science Annual Conference (August 5-6, 2020),and the GeoFlo regional meeting (November10,2020) What do you plan to do during the next reporting period to accomplish the goals?- Proceed with multispectral and hyperspectral image capturing for strawberry trials to estimate biophysical parameters - Test the methods developed this year on the drone images and develop new methods - Fly coastal natural areas and use machine learning in coastal vegetation image classification - Instruct technology transfer workshops

Impacts
What was accomplished under these goals? Goal 3: Only testing flights were conducted to test the multispectral and hyperspectral imaging payload?

Publications

  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Abd-Elrahman, A., Guan, Z., Dalid, C., Whitaker, V., Britt, K., Wilkinson, B., & Gonzalez, A. (2020). Automated Canopy Delineation and Size Metrics Extraction for Strawberry Dry Weight Modeling Using Raster Analysis of High-Resolution Imagery. Remote Sensing, 12, 3632. Guan, Z., Abd-Elrahman, A., Fan, Z., Whitaker, V. M., & Wilkinson, B. (2020). Modeling strawberry biomass and leaf area using object-based analysis of high-resolution images. ISPRS Journal of Photogrammetry and Remote Sensing, 163, 171-186.