Source: UNIVERSITY OF NEBRASKA submitted to
HIGH INTENSITY PHENOTYPING SITES.
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
EXTENDED
Funding Source
AFRI COMPETITIVE GRANT
Reporting Frequency
Annual
Accession No.
1022298
Grant No.
2020-68013-32371
Project No.
NEB-21-180
Proposal No.
2019-05467
Multistate No.
(N/A)
Program Code
A1141
Project Start Date
Sep 1, 2020
Project End Date
Aug 31, 2024
Grant Year
2020
Project Director
Ge, Y.
Recipient Organization
UNIVERSITY OF NEBRASKA
(N/A)
LINCOLN,NE 68583
Performing Department
Biological Systems Engineering
Non Technical Summary
The overall goal of this project is to establish two complementary high intensity phenotyping sites at University of Nebraska-Lincoln (UNL) and Texas A&M University at College Station, TX (TAMU), focusing on maize and wheat. Our four specific objectives are: (1) Support two ongoing community-based plant phenotyping efforts at both UNL and TAMU; namely, the maize Genomes to Fields Initiative (G2F) and the winter wheat breeding programs; (2) Advance and apply Unmanned Aerial Vehicles (UAVs) as our main platform for sensor deployment and data collection, focusing on the development of new sensors and tools, as well as precise and consistent systems and protocols, to enable high throughput analysis of plant physiological and biochemical traits at leaf and canopy levels; (3) Disseminate phenotyping data broadly and promptly via public data repositories, and create image processing and analytical tools that provide end-to-end solutions to extract target traits from UAV images; (4) Train a next generation workforce in high throughput plant phenotyping through graduate and undergraduate research and by developing in-class curriculum, online educational material, and summer short-term training workshops. Over the life of the project, we anticipate making leaps in the science and engineering of plant phenotyping, including new sensor/tool development, data modeling and analytics, and workforce development. In the long term, we seek to replicate our gains by establishing a network of high throughput phenotyping sites nationwide with what we and others have learned.
Animal Health Component
0%
Research Effort Categories
Basic
10%
Applied
50%
Developmental
40%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
2041510108150%
2031549202030%
4027210102020%
Knowledge Area
204 - Plant Product Quality and Utility (Preharvest); 203 - Plant Biological Efficiency and Abiotic Stresses Affecting Plants; 402 - Engineering Systems and Equipment;

Subject Of Investigation
1510 - Corn; 1549 - Wheat, general/other; 7210 - Remote sensing equipment and technology;

Field Of Science
2020 - Engineering; 1081 - Breeding; 1020 - Physiology;
Goals / Objectives
The goal of this project is to forge a bi-regional high intensity phenotyping alliance focusing on maize and wheat and involving complementary sites at University of Nebraska-Lincoln (UNL) and Texas A&M University (TAMU). The project aims toaccelerate the development of accurate and repeatable phenotyping sensors and tools, data models and analytics, and the science and engineering workforce required for plant phenotyping to make the next leap. In the long term, we expect to use the knowledge we generate so that our gains can be replicated through a network of high throughput phenotyping sites nationwide.The specific objectives of this project are to:Enhance and further integrate two ongoing community-based plant phenotyping efforts at both UNL and TAMU; namely, the maize Genomes to Fields Initiative (G2F) and the winter wheat breeding programs;Advance the application of unmanned aerial systems (UASs) as our main platform for sensor deployment and data collection, focusing on the development of sensors and tools, as well as precise and consistent systems and protocols, to enable high throughput analysis of plant physiological and biochemical traits at leaf and canopy levels;Disseminate phenotyping data broadly and promptly via public data repositories, and create image processing and analytical tools that provide end-to-end solutions to extract target traits from UAS images;Train a next generation workforce in high throughput plant phenotyping through graduate and undergraduate research and by developing in-class curriculum, online educational material, undergraduate summer experiential learning, and high throughput plant phenotyping workshops.
Project Methods
Three locations in NE (Lincoln, North Platte, and Alliance) and three locations in TX (College Station, McGregor, and Bushland) are selected for high-intensity plant phenotyping activities. These locations capture a gradient of climate and soil conditions in both States, where genotypes can be robustly tested to elucidate how measured phenotypes are controlled by the complex genotype-by-environment interactions. In addition, a fourth location in NE (Mead) will be included to use UNL's Spidercam field phenotyping facility for high-precision, advanced phenotyping with a number of high-end imaging, ranging, and spectroscopic sensors.The project will grow and test ~250 maize hybrid lines from the maize G2F (Genomes to Fields) Initiative. This collection of lines will be replicated twice in each location, and the total number of maize plots will thus be ~1500 in each State. The project will also coordinate to grow and phenotype three wheat trials: (1) the F1 hybrid wheat trial; (2) the Southern Regional Performance Nursery; and (3) The Nebraska Interstate Nursery and the Texas Elite trial. The total number of wheat plots will be ~780 in each State.Environmental variables regarding the soil and micrometeorological data will be collected. A weather station will be installed at each experimental site (Watchdog or Arable Mark weather stations) to record total solar radiation, photosynthetically active radiation, maximum and minimum air temperature, wind speed, relative humidity, vapor pressure deficit, and precipitation amount. Each field site will also be installed with soil moisture sensors to measure volumetric soil moisture at two depths for the entire growing season. At the first year before planting, a soil apparent electrical conductivity survey and elevation survey will be conducted using an EM38 sensor connected to an RTK-GPS. Then a stratified sampling design will be employed to collect ~40 soil samples from each site. Soil samples will be analyzed for an array of physicochemical properties including particle sizes, organic matter, pH, nitrate-N, and extractable phosphorus and potassium.High-intensity plant phenotypic data collection will be planned to cover several levels of spatial and temporal resolutions. The first category will be the basic traits that plant breeders traditionally record at the key developmental stages. G2F specifies a list of such traits for maize that we will collect from the maize trials in this project. A standard list of wheat traits will be collected from the wheat trials, including fall plant emergence and ground coverage, height , winter survival, days-to-heading and days-to-anthesis, plant height at full extension, lodging, grain yield, grain protein content, moisture, test weights and single kernel characteristics (seed weight and grain hardness).The second category of phenotypes will be made by handheld sensors that can be deployed in the field for rapid, in vivo measurements. These traits include leaf chlorophyll content, leaf area index, stomatal conductance, and leaf hyperspectral reflectance (from which an array of vegetation indices can be derived). Many vegetation indices are correlated with important physiological and biochemical aspects of plants. Because these sensor-based measurements are fast and non-destructive, they can be taken at multiple time points across a growing season. To validate these sensor readings, leaf disks will be sampled from the selected plots and analyzed for a number of biochemical traits including chlorophyll content, leaf water content, nitrogen, phosphorus, and potassium.Large multi-rotor UAS equipped with a RGB, multispectral, and thermal infrared camera will be our main platform for UAS-based phenotypic data collection. Flight height will be 25 or 50 m. UAS imaging will be conducted on a weekly basis at Lincoln and College Station, and biweekly at other four locations. UAS imaging will also occur at key crop developmental stages, and coincide with the ground-level data collection. Raw UAS images will be mosaicked into field maps using Pix4D or AgiSoft Photoscan and further analysis will be conducted at plot level for structural and morphological features and vegetation index calculations.About 150 maize lines and 60 wheat hybrids will be selected to phenotype with the NU-Spidercam facility, occupying roughly 0.5 ac. of the field. This activity represents the phenotyping effort with the highest intensity in terms of spatial and temporal resolution, data throughput, and precision. The sensors used on NU-Spidercam include a RGB camera, a 4-band multispectral camera, a thermal IR camera, a hyperspectral camera, a VNIR spectrometer, and a 3D LiDAR. Imaging will take place on a weekly basis in the season. One unique aspect is that the sensor platform will measure each plot five to six times during a day to capture the diurnal variations of the key traits such as leaf temperature and leaf drooping, which respond rapidly to the environment variation.Three novel plant phenotyping data analysis and modeling tools will be developed. Firstly, hyperspectral imagery data from NU-Spidercam will be analyzed to extract narrow spectral bands that exhibit high correlations with canopy chlorophyll content, water content, N, P, and K at the plot level. After sufficient validations, these bands will be implemented on commercially available multispectral cameras (by reconfiguring their optical channels) to be used on UAS for other sites in the project. Secondly, a leaf- and canopy-level spectral library will be compiled and developed from the hyperspectral reflectance data of maize and wheat. This spectral library, when put into practice, will allow rapid, nondestructive, and low-cost estimation of chlorophyll, water content, N, P, and K in plants. One emphasis will be to explore advanced machine learning algorithms (including convolutional neural network and deep learning) to better model these biochemical traits from the hyperspectral data, as well as sufficiently validate these models across the six testing sites. Thirdly, a new approach to estimate the crop evapotranspiration (ET) trait at the plot scale by combining the thermal infrared and RGB imaging, co-measured micrometeorological variables, and a two-source energy balance model will be investigated. This new approach will be developed and validated with the phenotyping data from the NU-Spidercam system first, and then will be tested on the imagery data from the UAS at other testing sites.Education and training activities of this project will be focused on the following five aspects. First, a course on high throughput plant phenotyping technologies and applications will be developed and taught as UNL and TAMU. This course will target for senior undergraduate and graduate students in plant breeding, agronomy, and agricultural engineering. Second, a series of Open Education Resources in the form of online text lessons and videos will be developed on high-throughput plant phenotyping technologies and applications, with the goal of contributing to improved agriculture and science literacy. The materials will be housed on the Plant and Soil Sciences eLibrary (PASSeL) website of UNL and accessible to the general public. Thirdly, the project will develop and host Undergraduate Summer Learning Experience at UNL for a two-week immersion experience in research. Efforts will be made to recruit underrepresented groups by advertising with colleagues at Historically Black Colleges and Universities. Fourthly, the project will train four PhD students, one postdoc, and three professionals by directly participating in the research activities of this project. Finally, a two-day workshop focused on high-throughput plant phenotyping technologies (such as imaging processing and UAS) will be developed and held annually at UNL, targeting for 50 participants from both academic institutions and industry.

Progress 09/01/22 to 08/31/23

Outputs
Target Audience:Corn and wheat boards in Nebraska and Texas. Plant scientists includingmaize geneticists, breeders, plant physiologists, and pathologists. Scientists from other disciplines includingstatisticians, agronomists, soil scientists, computer scientists, entomologists, and ecologists. Graduate students, undergraduates, and research staff at the University of Nebraska-Lincoln, Texas A&M University, and Mississippi State University. Conference attendees of professional societies including ASABE, ASA-CSSA-SSSA, NAPPN, NAPB, and SPIE.The UNL team set up an exhibition booth at the ASABEand NAPPN meeting to promote plant phenotyping technologies. General public. The NU-Spidercam system hosted numerous visitors from all around the country in summer of 2023. Co-PD Yeyin Shi's group attended 2023 UNL's East Campus Discovery Day and demonstrated UAS technologies, different imaging technologies, and their application in crop production. The plant phenotyping data management system (as part of the more general agricultural data management system developed at UNL) was demonstrated to and tested by the graduate students and research staff. The project supported the formation of NC1212, a USDA multistate committee on collaborative plant phenotyping. Internationally, the project formed collaborations with scientists atRothamsted Research, National Institute of Agricultural Botany (NIAB), both in UK, and Wageningen University in Netherland. Changes/Problems:Mississippi State University: Kha Dan, the electrical engineer originally working on the project, first went to part-time employment and then left the project entirely. Thus, progress on collaborative communications between robot and UAS as well as sensor-based object avoidance on the robot slowed considerably. However, Collin McLeod, the mechanical engineer working on the project, ultimately solved those issues and has completed the construction, refinement, and testing of the system. What remains is field validation in Mississippi, potentially subsequent refinement, and then field testing in Nebraska and Texas. The Texas program had resources remaining and grew another season of the Genomes to Fields GxE trialunder three management conditions and collected substantial UAS data. Cyverse has changed it's business model and is no longer feasible as a repository for our data. Therefore, we are using Dr. Edgar Spalding's server at UW-Madison as our repository for G2F data for now and looking for other government repositories. What opportunities for training and professional development has the project provided?The project trained 4 undergraduate students in applied plant phenotyping in 2023. Graduate students attended professional society meetings(NAPPN, NAPB, ASABE, SPIE, ASA-CSSA-SSSA)and made technical presentations. Graduate students also attended various workshops such as the crop nitrogen use efficiency workshop in Lincoln, NE, and crop growth simulation modeling in Ames, IA. At UNL, four graduate students (3 MS and 1 PhD) supported by the grantsuccessfully defended their thesis and graduated.One graduate student secured an internship opportunity with industry (6 months). How have the results been disseminated to communities of interest?In this reporting period, the project resulted in 12 peer-reviewed journal articles, 1 PhD dissertation, and 3 MS theses. Co-PD Murray's group has worked with the G2F community to process the UAS images into the tabular data(additional funding through AG2PI). PDs and graduate students attended numerous professional society meetings, gave technical presentations to disseminate the results. What do you plan to do during the next reporting period to accomplish the goals?Objective 1: We will continue to support the corn G2F efforts in NE and TX.Leveraging these field trials, the team will continue to collect UAS images (primarily with RGB and multispectral cameras) on a weekly/bi-weekly basis throughout thegrowing season.Ground-truth measurements of key agronomic, physiological, and biochemical traits will be collected, but likely atreduced intensity. We plan to continue the data collection with the NU-Spidercam system for high intensity phenotyping data collection (~60 plots of corn as proposed). Objective 2: The analyses of the high-intensity phenotypingdata will continue to focus on (1) processing of UASRGB and multispectral images, and relating them to important agronomic andphysiological traits, (2) use leaf-level and canopy-level hyperspectral data to estimate biochemical traits such as chlorophyll, nitrogen, water content, and other nutrient contents, (3) plot-scale crop evapotranspiration modeling using thermal IR+multispectral imaging, and two-source energy balance modeling, and (4) modeling the diurnal variation of canopy spectral indices such as NDVI. Objective 3: We will continue to develop the Agricultural Data Management and Analysis Platform and use the experimental data generated from the project to test the platform. New models and algorithms will be written as analytics APIs and launched in the Platform. Algorithms and pipelines to process raw UAS images and plot-level trait extraction will be further refined and standardized. More data from our project will be made publically available. Objective 4: This objective is completed and no further activities are planned.

Impacts
What was accomplished under these goals? Objective 1: In Nebraska and Texas, corn G2F (Genome to Fields) trials and winter wheat breeding trialswere grown. Mutiple UAS data collection usingRGB and multispectral cameras systems were conducted. For corn, the UAS image data are being processed into tabular forms and made available to the G2F community. Objective 2:At UNL's NU-Spidercam Field Phenotyping Facility, weplanted HIPS corn hybrid varieties (n=23) with two replicates (46 plots). A second corn experiment with 2 irrigation treatment levels and 3 nitrogen levels was also conducted (24 plots). In addition, we had the resources to conduct a third soybean experiment with 2 varieties and 2 irrigation treatments (24 plots). These 94 plots were subjected to intensivephenotyping at a frequency of at least 2 days per week, across the 2023 growing season. For many days, repeated phenotyping (e.g., 6-7 times) was conducted within a day to study the diurnal patterns of the plant physiological traits. We also used handheld sensors (leaf stomatal conductance sensors, chlorophyll meters, and leaf area index meters) to collect intensive ground-truth data to be compared with the automated phenotyping/imaging data later. In terms of phyisological and biochemical phenotyping, we focused on the following analyses of the collecteddata: (1) modeling leaf stomatal conductance (corn, wheat, and soybean) using the high-throughput imaging data and weather data; (2) modeling leaf area index using the height, spectral indices, and fractional vegetation cover derived from the high-throughput imaging data; (3) modeling of plot-scale evapotranspiration using the multispectral and thermal IR imaging, and a two-source energy balance modeling; (4) exploring the diurnal variation of canopy spectral indices (such as NDVI and PRI); and (5) modeling leafchemical properties (chlorophyll content, leaf thickness, water content, macronutrient contents) from the leaf-level hyperspectral reflectance data. We used a UASequipped with a hyperspectral camera system to acquire data from three field trials in 2023, each covering 4-5 growth stages. Hyperspectral images were processed to extract plot-level vegetation indices and estimate corn nitrogen and water status. Along with the data collected with the same system in the previous year (2022), we were able to show that (1) high-quality hyperspectral data could be collected from a UASplatform, and (2) the hyperspectral data could satisfactorily estimate nitrogen content of corn crops at both leaf and canopy level. In Mississippi, Dr. Thomasson's group has built and field-tested a robust ground-based electrically powered robot with autonomous navigation, containing reference objects for calibration of UAS image data. Last year we field-tested the robot to prove functionality and identified areas of needed refinement. This year we made the needed refinements and have initiated combined tests of the collaborative air-ground robotic system. Objective 3: We continued to develop the plant phenotyping data management system (as part of the more general agricultural data management system), with the goal to develop novel deep-learning based data storage and indexing algorithms to accelerate the processing of plant phenotyping data. We used plot images from UNL's Spidercam system as the developmental datasets.In addition, Co-PD Seth Murray worked with a group of scientists at various institutions to curate and process UAS RGB and multispectral images collected in the past three years for the maize G2F trials. Objective 4: A collaborative undergraduate mentoring effort among PIs, graduate students, and other research team members was accomplished in summer 2023. Four undergraduate research interns were hosted at University of Nebraska-Lincoln for 10 weeks. The interns were from institutions located in California, Missouri, Nebraska, and South Carolina. The interns were fully integrated in research teams and actively engaged in the research supporting the HIPS program. The students met once a week as a cohort to share reflections and questions they had about their research projects. Connections about the science and application of the research occurred for the interns during the cohort meetings. The capstone of the summer research experience was a poster symposium where each of the interns presented their summer research project work. Survey data (pre and post) was collected from learners who went through the High Throughput Phenotyping in Plant Breeding lesson during the fall 2022 and spring 2023. Learners were primarily in undergraduate courses with a small portion of respondents in a graduate level course or working in industry. This data is part of the research work completed by Catherine Mick (one graduate student supported by the grant). The results from the pre- and post-survey showed that learner self-reported confidence and objectively assessed knowledge increased regarding the application of UASs for data collection and evaluation in plant breeding.

Publications

  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Bai, G., Ge, Y., Leavitt, B., Gamon, J.A., Scoby, D., 2023. Goniometer in the air: Enabling BRDF measurement of crop canopies using a cable-suspended plant phenotyping platform. Biosystems Engineering 230, 344-360. https://doi.org/10.1016/j.biosystemseng.2023.04.017
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Wijewardane, N.K., Zhang, H., Yang, J., Schnable, J.C., Schachtman, D.P., Ge, Y., 2023. A leaf-level spectral library to support high-throughput plant phenotyping: Predictive accuracy and model transfer. Journal of Experimental Botany 74, 4050-4062. https://doi.org/10.1093/jxb/erad129
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Li, J., Wijewardane, N.K., Ge, Y., Shi, Y., 2023. Improved chlorophyll and water content estimations at leaf level with a hybrid radiative transfer and machine learning model. Computers and Electronics in Agriculture 206, 107669. https://doi.org/10.1016/j.compag.2023.107669
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Xie, X., Ge, Y., Walia, H., Yang, J., Yu, H., 2023. Leaf-counting in monocot plants using deep regression models. Sensors 23, 1890. https://doi.org/10.3390/s23041890
  • Type: Journal Articles Status: Published Year Published: 2022 Citation: Zhan, Y., Zhang, R., Zhou, Y., Stoerger, V., Hiller, J., Awada, T., Ge., Y., 2022. Rapid online plant leaf area change detection with high-throughput plant image data. Journal of Applied Statistics https://doi.org/10.1080/02664763.2022.2150753
  • Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Izere, P. 2023. Plant height estimation using RTK-GNSS enabled unmanned aerial vehicle (UAV) photogrammetry (Master's Thesis). Biological Systems Engineering - Dissertations, Theses, and Student Research. Department of Biological Systems Engineering, University of Nebraska, Lincoln.
  • Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Thapa, K. 2023.Characterization of physical and biochemical traits in wheat and corn plants using high throughput image analysis (Master's Thesis). Biological Systems Engineering - Dissertations, Theses, and Student Research. Department of Biological Systems Engineering, University of Nebraska - Lincoln.
  • Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Zhang, J. 2023. Estimating crop stomatal conductance through high-throughput plant phenotyping (Master's Thesis). Biological Systems Engineering--Dissertations, Theses, and Student Research. Department of Biological Systems Engineering, University of Nebraska - Lincoln.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: Zhang, J., Thapa, K., Chamara, N., Bai, G., Ge, Y., 2023. Estimating crop stomatal conductance from RGB, NIR, and thermal infrared images. Proc. SPIE 12539, Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VIII, 125390A (13 June 2023); https://doi.org/10.1117/12.2663888
  • Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: Li, J., Ge., Y., Puntel, L., Heeren, D., Balboa, G., Shi, Y., 2023. Combining machine learning with a mechanistic model to estimate maize nitrogen content from UAV-acquired hyperspectral imagery. Proc. SPIE 12539, Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VIII, 1253905 (13 June 2023); https://doi.org/10.1117/12.2663817
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Adak, Alper, Seth C. Murray, and Steven L Anderson. 2023. Temporal phenomic predictions from unoccupied aerial systems can outperform genomic predictions. G3 Genes | Genomes | Genetics. jkac294 https://doi.org/10.1093/g3journal/jkac294.
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Adak, Alper, Steven L. Anderson, and Seth C. Murray. 2023. Pedigree-Management-Flight Interaction for Temporal Phenotype Analysis and Temporal Phenomic Prediction. The Plant Phenome Journal 6, e20057. https://doi.org/10.1002/ppj2.20057
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Adak, A., Murray, S.C., Myeongjong, C., Wong, R., Katzfu�, M. 2023. Phenomic data-driven biological prediction of maize through field-based high throughput phenotyping integration with genomic data. Journal of Experimental Botany 74: 53075326. https://doi.org/10.1093/jxb/erad216
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Lima, Dayane Cristina Lima, Alejandro Castro Aviles, Ryan Timothy Alpers, Bridget A McFarland, Shawn Kaeppler, David Ertl, Maria Cinta Romay, Joseph L Gage, James Holland, Timothy Beissinger, Martin Bohn, Edward Buckler, Jode Edwards, Sherry Flint-Garcia, Candice N Hirsch, Elizabeth Hood, David C Hooker, Joseph E Knoll, Judith M Kolkman, Sanzhen Liu, John McKay, Richard Minyo, Danilo E Moreta, Seth C Murray, Rebecca Nelson, James C Schnable, Rajandeep S Sekhon, Maninder P Singh, Peter Thomison, Addie Thompson, Mitchell Tuinstra, Jason Wallace, Jacob D Washburn, Teclemariam Weldekidan, Randall J Wisser, Wenwei Xu, and Natalia de Leon. 2023. 20182019 field seasons of the Maize Genomes to Fields (G2F) G x E project. BMC Genomic Data 24: 29. https://doi.org/10.1186/s12863-023-01129-2.
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Andrew W Herr, Alper Adak, Matthew E Carroll, Dinakaran Elango, Soumyashree Kar, Changying Li, Sarah E Jones, Arron H Carter, Seth C Murray, Andrew Paterson, Sindhuja Sankaran, Arti Singh, Asheesh K Singh. 2023. Unoccupied aerial systems imagery for phenotyping in cotton, maize, soybean, and wheat breeding. Crop Science 63,1722-1749. https://doi.org/10.1002/csc2.21028
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Adak, Alper, Seth C. Murray, Claudia Irene Calder�n, Valentina Infante, Jennifer Wilker, Jos� I. Varela, Nithya Subramanian, Thomas Isakeit, Jean-Michel An�, Jason Wallace, Natalia de Leon, Matthew A Stull, Marcel Brun, Joshua Hill, and Charles D Johnson. 2023. Genetic mapping and prediction for novel lesion mimic in maize demonstrates quantitative effects from genetic background, environment and epistasis. Theoretical and Applied Genetics 136: 155-162. https://doi.org/10.1007/s00122-023-04394-y
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Bagnall, G. C., Thomasson, J. A., Yang, C., Wang, T., Han, X., Sima, C., & Chang, A. (2023). Uncrewed aerial vehicle radiometric calibration: A comparison of autoexposure and fixed exposure images. The Plant Phenome Journal, 6(1), e20082. https://doi.org/10.1002/ppj2.20082.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: McLeod, C. J., & Thomasson, J. A. 2023. Continued development of autonomous mobile ground control point for increasing the accuracy of unmanned aerial-vehicle-based phenotyping. In Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VIII (Vol. 12539, pp. 131-139). SPIE. https://doi.org/10.1117/12.2663661


Progress 09/01/21 to 08/31/22

Outputs
Target Audience:Corn and Wheat Boards in NE and TX. Corn and wheat breeders.Other plant scientists, specifically maize geneticists, breeders of other crops, plant physiologists and pathologists. Scientists from other disciplines including statisticians, agronomists, soil scientists, computer engineers, entomologists, ecologists, etc. Graduate students and undergraduates at UNL, TAMU, and MS State. Professional societies including ASABE, ASA-CSSA-SSSA, NAPB, and SPIE. Corn and wheat producers in NE and TX. General public.For example, Co-PI Yeyin Shi's group again attended 2022 UNL's East Campus Discovery Day and demonstrated UAV-based imaging systems to the event participants. This event was attended by thousands of residents in Lincoln, NE. The demonstration focused on the UAV hardware and software, and an introduction on how UAV technologies are used in agriculture to improve crop performance and field management. The Spidercam site hosted tours for FFAR fellows (28 visitors), a delegate from NASA Earth Science Division(10 visitors), DOE program officers (3 visitors), and aplanning team for NSF ERC program (30 visitors). International collaborations. Two international collaborations are ongoing with the scientists at Rothamsted Research and National Institute of Agricultural Botany (NIAB), United Kingdom. A joint research proposal (through BBSRC-NSF/BIO) with NIAB on wheat yellow rust phenotyping was submitted (but not funded). Changes/Problems:Post COVID situation, it has been difficult to hire new personnel to work in the field. We have been still facing challenges with hiring more people to accomplish a lot of goals we plan to. This year, Eastern Nebraska experienced an unusually harsh drought. It is suspected this led to stunting of plots in specific areas of the field, though additional tests would need to be performed to test this hypothesis. At MS State,Kha Dan, the electrical engineer working on the project, has gone to part-time employment, so progress on collaborative communications between robot and UAS as well as sensor-based object avoidance on the robot have slowed considerably. Collin McLeod, the mechanical engineer working on the project, remains full-time and is making good progress. What opportunities for training and professional development has the project provided?The project trained 2 postdoc (both at TAMU), 10 graduate students (8at UNL and2 at TAMU),4 undergraduate students (at TAMU), and 3 research technicians. One PhD student successfully defended her dissertation at UNL, and continued her academic career as a research scientist in another institution.In addition, we trained 5 undergarduate research interns during the summer and they were from institutions located inAlabama, Indiana, New York, Puerto Rico, and Virginia. How have the results been disseminated to communities of interest?The project has resulted in fourpeer-reviewed journal articles, onePhD dissertation, and two referred conference proceedings in SPIE. UAS images were handled and deposited into the UAS-Hub (https://uashub.tamucc.edu/) for the TX wheat program. UAS images and processed tabular data were made publicly available for the TX corn program. PIsand graduate students attended many international/national level scientific meetings and symposiums to disseminate the research and results from this project, as listed below. PI Ge's group attended the SPIE meeting (1 oral presentation, Apr 2022),ASABE annual meeting (1 oral presentation, July 2022), and the NAPB annual meeting (1 invited talk, Aug 2022). Co-PI Thomasson's group attended the SPIE meeting (2 oral presentation, Apr 2022), and ASABE annual meeting (1 oral presentation, Jul 2022). Co-PI Murray's group attended the BrIAS meeting in Belgium (2 presentations, Jan-Feb 2022), PAG XXIX Conference (1 presentation, Jan 2022), University of Guelph Plant Science Symposia Series (1 presentation, Nov 2021), ASA-CSSA-SSSA annual meeting (1 invited talk, Nov 2021), Iowa State University Phenomics Phridays (2 presentations, Nov 2021), and 1st International Symposium on Plant Breeding (1 presentation, Sep 2021). Co-PI Shi's group attended the ASABE annual meeting (1 oral presentation, July 2022). What do you plan to do during the next reporting period to accomplish the goals?Objective 1: We will continue to support the wheat breeding programs and the maize G2F efforts in Nebraska and Texas. Leveraging these field experiments, the team will continue to collect UAS images on a weekly/bi-weekly basis throughout the plant growing season.Ground-truth measurements of key agronomic,physiological, and biochemicaltraits will be collected, either using handheld sensors or by collecting leaf samples. The Spidercam field phenotyping system at UNL will be used to collect intensive phenotyping data (with sub-day temporal resolution) from ~60 plots of corn and wheat each. Objective 2: We will continue to develop new methods to extract digital traits from UAV images at the plot-scale and relatethem to important agronomic, physiological, and biochemicaltraits of wheat and corn. Specifically, we continue to investigate the use of hyperspectral data (both the hyperspectral imagery from UAS/Spidercamand leaf-level hyperspectral reflectance data) to estimate leaf/canopy traits including chlorophyll, nitrogen,water content, and possibly macro-nutrient concentrations. This effort will allow us to identify useful spectral bands to quantify these traits and guide the design of low cost multispectral sensors (either handheld or cameras on UAS) to measure these important leaftraits for corn and wheat breeding and selection. We will continue to investigate the use of thermal IR images (from Spidercam and UAS) combining the weather data and a two-source energy balance model for cropevapotranspiration (and thus water use) modeling. We will also investigate the time-series RGB and multispectral UASimages for yield prediction across multiple experimental sites to standardize the sensors and protocols for UAV flying and image acquisition. Objective 3: We will continue to develop the Agricultural Data Management and Analysis Platform and use the experimental data generated from the project to test the platform. New models and algorithms will be written as analytics APIs and launched in the Platform. Algorithms and pipelines to process raw UAV images and plot-level trait extraction will be further refined and standardized. More data from our project will be made publically available. Objective 4:During the next reporting period we plan to host another group of interns in summer 2023. We will also complete work on an educational lesson about the process of gathering and analyzinghigh throughput plant phenotyping data and the application of this data in plant breeding. The lesson will be available online (open access) on the Plant and Soil Sciences eLibrary (https://passel2.unl.edu/).

Impacts
What was accomplished under these goals? Objective1: In NE for corn, we planted 550 plots on 5/10/22. Of the 363 hybrids represented, 178 were replicated once; 161 were replicated twice, and 24 were replicated four or more times. We collected stand count on each plot and various trait data from each plot during the growing season. Various phenotypes such as Days to Anthesis, Days to Silking, Ear Height, Height to Flag leaf and Height to the top of tassel from the ground. We monitoredplant growth and any disease incidence to score them if the plants are infected during the period. Wecollected stalk and root lodging scores if we sawany lodging before harvest. UAV imageswere collected from the corn plots using a RGB camera and multispectral camera on a weekly basis. Images were processed and mosaicked using Pix4D for further plot-level trait analysis. Similarly in TX, the team led by co-PI Murrayconducted UAV flights and image data collection throughout the growing season on a weekly basis. In NE's Spidercam site, 23 hybrids (from G2F HIPS) were planted as 6-row plots with two replications. The Spidercam system was operated weekly from emergence to maturation to capture multispectral, thermal infrared, and hyperspectral images, as well as 3D LiDAR point clouds from each plot. Weekly ground level measurements were also madefrom each plot for height, leaf chlorophyll concentration, and leaf area index. Selected plots were also measured for leaf stomatal conductance on selected dates covering a number of key physiological stages. Leaf samples were collected from each plot at R3 stage. The leaf samples were measured using an ASD spectrometer for hyperspectralreflectance, and measured in the lab for an array of biochemical and physiological traits (leaf thickness, water content, macro- and micro-nutrient content). Yield is estimated from the middel two rows for each plot by hand harvesting. In Texas for wheat,UAV data was collectedin three locations, College Station, McGregor, Bushland, and Castroville. Twenty-six flights were captured at the College Station and McGregor locations. High-resolution red, green, and blue (RGB) Images were captured using a DJI phantom 4 pro equipped with 20 megapixels one-inch CMOS sensor. Multispectral (G, R, red edge, & NIR) images were captured using a DJI Matrice 100 equipped with a SlantRange 3P (SlantRange, Inc., California, USA) sensor. Both UAS platforms were flown at 30 m altitude with front and side overlap of 75 percent to obtain high resolution orthomosaic images. Pix4Dcapture (Pix4D S. A, Switzerland) was used as the planning software for the UAS flights. Raw images obtained from the RGB sensor were processed using Agisoft metashape softwareto generate 3D point cloud, DSM (Digital Surface Model), and orthomosaic images. In NE, data collection in wheat was mainly in Lincoln,Mead, and Clay Centersites. The field trials screened by UAV were the F3:F6 preliminary yield trials, as well as an agronomic study involving different water and nitrogen application rates.UAV images were captured on a weekly basis from April to June with RGB and multispectral cameras. At selected dates and plots, ground-truth measurements of height, LAI, chlorophyll content were collected matching the UAV images. Objective2: Leaf-level hyperspectral data were modeled to estimate corn leaf physiological and biochemical traits. Hyperspectral images from the Spidercam plots were processed to extract plot-level hyperspectral reflectance, which were then modeled to estimate the physiological and biochemical traits.Plot-level multispectral images,thermal IR images, and LiDAR point cloudswere processed to estimate vegetation biomass and leaf area index. These data were also used to estimatestomatal conductance and plot-level evapotranspiration, by leveraging the weather data (measured by an onsite weather station) and a two-source energy balance model. In Mississippi State University, Dr. Thomasson's group has built a robust ground-based electrically powered robot with autonomous navigation, containing reference objects for calibration of UAS image data. We have field-tested the robot to prove functionality and have identified areas of needed refinement. Objective3: An integrated data management and analysis platform for high throughput plant phenotyping data isunder development. This platform integrates data submission, search, visualization, and analytics APIs. UAV images and phenotyping data from Spidercam Sites in Objective 1 are used as example datasets for platform development. The analytics APIs are based on the models and algorithms developed in Objective 2. Some innovative features of this system include (1) deep learning fueled data retrieval,(2) in-situ/real-time data processing, (3) system managed data processing and visualization procedures, etc. The platform is designed to accommodate multi-faceted plant phenotyping data including point data, spectral data, images, hyperspectral image cubes, LiDAR point clouds, and complex spatiotemporal data. Objective 4: A collaborative undergraduate mentoring effort among PIs, graduate students, and other research team members was accomplished in summer 2022. Five undergraduate research interns were hosted at University of Nebraska-Lincoln for 10 weeks. The interns were from institutions located in Alabama, Indiana, New York, Puerto Rico, and Virginia. The interns were fully integrated in research teams and actively engaged in the research supporting our HIPS program. The five students met once a week as a cohort to share reflections and questions they had about their research projects. Connections about the science and application of the research occurred for the interns during the cohort meetings. The capstone of the summer research experience was a poster symposium where each of the interns presented their summer research project work. One of the five interns from the HIPS project won first-place in the poster competition out of over 100 total undergraduate students participating in summer research experiences at UNL.

Publications

  • Type: Journal Articles Status: Published Year Published: 2022 Citation: Chamara, N., Islam, M.D., Bai, G., Shi, Y., Ge, Y., 2022. Ag-IoT for crop and environment monitoring: Past, present, and future. Agricultural Systems 203, 103497. https://doi.org/10.1016/j.agsy.2022.103497.
  • Type: Journal Articles Status: Published Year Published: 2022 Citation: DeSalvio, A.J., Adak, A., Murray, S.C., Wilde, S.C., Isakeit, T., 2022. Phenomic data-facilitated rust and senescence prediction in maize using machine learning algorithms. Scientific Reports 12, 7571. https://doi.org/10.1038/s41598-022-11591-0.
  • Type: Journal Articles Status: Accepted Year Published: 2022 Citation: Zhang, Zhiwu, Chunpeng Chen, Jessica Rutkoski, James Schnable, Seth Murray, Lizhi Wang, Xiuliang Jin, Benjamin Stich, Jose Crossa, Ben Hayes. 2022. Harnessing agronomics through genomics and phenomics in plant breeding: A review. Plant Breeding Reviews. (in press) doi: 10.20944/preprints202103.0519.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Dan, K., McLeod, C., Thomasson, J.A., 2022. Improved autonomous mobile ground control point robot collaborates with UAV to improve accuracy of agriculture remote sensing. Proc. SPIE. Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VII, 1211408. https://doi.org/10.1117/12.2622615.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Izere, P., Zhao, B., Ge, Y., Shi, Y., 2022. Estimation of plant height using UAS with RTK GNSS technology. Proc. SPIE. Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VII, 121140P. https://doi.org/10.1117/12.2623033.
  • Type: Theses/Dissertations Status: Published Year Published: 2022 Citation: Wang, L., 2022. Improved Yield Prediction with UAS-based Leaf Area Index Estimation and a Hybrid Machine Learning- and Process-based Model. PhD Dissertation. University of Nebraska-Lincoln.
  • Type: Journal Articles Status: Published Year Published: 2022 Citation: Lee, H.-S., Shin, B.-S., Thomasson, J.A., Wang, T., Zhang, Z., Han, X., 2022. Development of multiple UAV collaborative driving systems for improving field phenotyping. Sensors 22(4), 1423. https://doi.org/10.3390/s22041423.


Progress 09/01/20 to 08/31/21

Outputs
Target Audience:The Nebraska teamworked with researchers from Purdue University, University of Minnesota, and Iowa State University and created a new Multistate Hatch Project NC1212: Exploring the Plant Phenome in Controlled and Field Environments. The teammade presentations (virtual zoom meeting)to Nebraska Congressional staff and USDA Farm Service Agency Nebraska leadership onhigh throughput plant phenotyping research at University of Nebraska-Lincoln and particularly highlighted this project. Ge gavetours to two US cabinet-level officialsfor theGreenhouse Innovation Centerat UNL and introduced this project to them. The project catalyzed research collaborations between UNL and Rothamsted Research at United Kingdom. Now two institutions have formal collaboration in crop phenotyping research using UAV and the field phenotyping systems (Spidercam at UNL and the LemnaTec field gantry system at Rothamsted). The team members presented the research and results fromthis projectin a number of professional society meetings. In both Texas and Nebraska, the research teamreached to corn and wheat producers, commodity boards (corn and wheat) andgeneral public. PI Yufeng Ge and Co-PI Yeyin Shi's groups attended University of Nebraska's East Campus Discovery Day on Jun/12/2021. This event was attended by thousands ofresidents in Lincoln. We introduced and demonstrated high throughput plant phenotyping technologies (such as remote sensing and UAVs), and explained how these technologies can help researchers to accelerateplant breeding cycles and improve food security in the future. Changes/Problems:Our project has a fewchanges in personnel. Firstly, co-PI Dr. StephenBaenziger retired from UNLin May 2021. His successor, Dr. Katherine Frels started at UNL in March 2021. As the new small grains breeder, she intends to fully utilize the advantage of high throughput phenotyping to modernize the UNL Small Grains (including wheat) Breeding program. As such, she is now actively participating in this grant and taking over Dr. Stephen Baenziger's roles and responsibilities. Secondly, co-PI Dr. Vikas Belmkar left UNLin Jun 2021. Vikas was responsible for Obj. 3 of our project. InSep 2021, Dr. Yu Pan was hired as a new data science specialist (at the rank of research assistant professor). Dr. Yu Pan is now participating in this grantand responsible for leading Obj. 3. Thirdly, co-PI Dr. Alex Thomasson left Texas A&M University and joined Mississippi State University. This change has minimum impact on the project. Dr. Alex Thomasson is still actively involved in the project and will continue to do so. His team is actively contributing to Obj. 2 of the project. A few aspects of the project were impacted by Covid-19. For most of the time in this reporting period, our universitieshadstrict policies for social distancing and travel restrictions. We were not allowed to share a vehicle when driving to the field sites, which hampered our progress in field work and data collection.Wereduced the speed of plant sample processing and analysis due to the need for social distancing among project personnel working in the lab. Travel restrictionssomewhat limited our capacity toattend professional society meetings and disseminate project results to the community. We were not able to implement the summer research experiential learningpart of Obj. 4 (which planned torecruitundergraduate students to come to UNL during the summer for research experiential learning program) because of the Covid.We hope as the trend with the Pandemic has been getting better, we will be able to return to normalcy and improve our productivity. We have started working with UNL's Office of Graduate Studies' Summer Research Program to plan for this activity in Summer 2022. What opportunities for training and professional development has the project provided?The entire team (including PIs, graduate students, and technical staff) have met virtually on a monthly basis. In each meeting,we had onestudent from alab to give a research seminar (45 minute long followed by a Q&A and discussion) on the topic of high throughput plant phenotyping. The topics covered so far included "UAV for wheat yield estimation", "autonomous ground control point design", "UAS data hub at Texas A&M University", and "seasonal UAV images and spectral index-based selection of corn". At UNL, three undergraduate researchers participatedin the project.The students were trained on how to use instruments (such as NDVI sensors, handheld leaf chlorophyll meters, and leaf area index meters) to measure plants in the field. They were also trained to use MATLAB and R for plant image processing and statistical analysis. One of the students was leveraging this research experience for formal internship experiential learning course credits. At UNL, Co-PI Yeyin Shi taught a 3-week mini course "Aerial Imagery Processing and Analysis Using Python" in Jan/2021.Fifteen students (both senior undergraduates and graduate students) from Biological Systems Engineering and Agronomy & Horticulture registered and attended this course. PI Ge gave a 2-hour webinar at Phenome Force (Nov-2020) and introduced hyperspectral technologies for leaf chemical and physiological trait analysis. This webinar was attended by ~100 researchers from all over the world. How have the results been disseminated to communities of interest?The project has resultedthree peer-reviewed journal articles and two referred conference proceedings. The team members presented the research and results fromthis projectin a number of professional meetings as listed below: Co-PI Seth Murray's group attended Corn Breeding Research Meeting (Virtual, Feb/20/2021), Soybean Breeder's Workshop (Virtual, 02/24/2021), andThe National Turfgrass Federation (NTF) and The Foundation for Food and Agriculture Research (FFAR) - Turf Stakeholder Summit II(Virtual,10/20-22/2020). Co-PI Alex Thomasson's group attendedNorth American Plant Phenotyping Network conference (Virtual, Feb/19/2021),Quad Cities ASABE meeting (Feb/23/2021),IEEE Stratus conference (May/17/2021), andInternational Symposium on Global Trends and Challenges of Smart Farming Technologies and Applications atKangwon National University, Korea (Jun/15/2021). PI Yufeng Ge's group attendedNorth American Plant Phenotyping Network conference (Virtual, Feb/20/2021) and SPIE Conference (Virtual, Apr/5/2021). Co-PI Yeyin Shi's group attended SPIE Conference (Virtual, Apr/5/2021) and the ASABE Annual International Meeting (Virtual, Jul/2021). What do you plan to do during the next reporting period to accomplish the goals?Obj. 1:We will continueto support the wheat breeding programs and the maize G2F initiatives in Nebraska and Texas. In Nebraska, the locations will be Lincoln, Mead, North Platte, and Scottsbluff (or Sidney). In Texas, the locations will be College Station,McGregor, Castroville, and Bushland. Leveraging these experiments, the team will continue to collect UAS images (with RGB andmultispectral cameras, and for some selected trials with thermal infraredand hyperspectral cameras) on a weekly or bi-weekly basis. Ground-truth measurements of key chemical and physiological traits will be collected, either using handheld sensors or by collecting leaf samples. The Spidercam field phenotyping system at UNL willbe used to collect intensive phenotyping data (with sub-day temporal resolution)from ~60 plots of corn and wheat each. Obj. 2: we will continue to develop new methods to extract digital traits from UAV images and related them to important chemical and physiological traits of wheat and corn. Specifically, we continue to investigate the use of hyperspectral data (both the hyperspectral imagery from UAV and leaf-level hyperspectral reflectance data) to estimate leaf/canopy traits including chlorophyll, nitrogen, and water content. This effort will allow us to identify useful spectral bands to quantify these traits and guide the design of low cost multispectral sensors (either handheld or cameras on UAV) to measure these important biochemical traits for corn and wheat breeding. We will continue to investigate the use of UAV thermal IR images combining the weather data for crop canopy evapotranspiration modeling. We will also investigate the time-series RGB and multispectral UAV images for yield prediction across multiple experimental sites to standardize the sensors and protocols for UAV flying and image acquisition. Obj.3: At Texas, the team will continue to leverage the UAS data hub to store, analyze, and disseminate the UAS image data collected at Texas. At Nebraska, we will continue to develop the integrated data management and analysis platform for high-throughput phenotyping data. Obj. 4: Regarding the online training material, we will complete the two modules on high throughput plant phenotyping of crop height and leaf area index, upload them to the Plant and Soil Science eLibrary, and test them thoroughly. We will begin to work on the third and four modules on crop evapotranspiration and hyperspectral reflectance. In the next reporting period, we expect to complete 50% of the latter two modules. Regarding the classroom teaching, Co-PI Yeyin Shi will continue to teach the course "Aerial Imagery Processing and Analysis Using Python" in spring 2022. One or two new class modules related to UAV thermal infrared and hyperspectral image processing (using data from this project) will be created and added. Another mini-course: "Computer Vision and Artificial Intelligence Applications in Plant Phenotyping", developed by Dr. Sruti Choudhury, will also be taught in spring 2022. The summer research experiential learning for undergraduates will be implemented Jun-Aug of 2022. A preliminary recruiting webpage has already been built with the project and hosting mentor information. In the next reporting period, we will focus on (1) improving the information webpage, (2) recruiting participants (with emphasis on minorities and HBCU), and (3) implementing the summer research experiential learning program at UNL.

Impacts
What was accomplished under these goals? Obj. 1.For wheat in Nebraska, the focus of data collection was in the Lincoln and Mead sites and the main breeding trials screened by UAV were the hybrid trials and the F3:F6 preliminary yield trial (~1000 plots in total). UAV images were captured on a weekly basis from the beginning of April to end of June with RGB, multispectral, and thermal infrared cameras. At selected plots ground-truth measurements of plot height, stomatal conductance, and leaf area index (LAI) were also collected matching the UAV images. For wheat in Texas, the team led by co-PI Amir Ibrahim conductedUAV flights at four locations in Texas, namely College Station, McGregor, Castroville, and Bushland during the 2020 - 2021 wheat growing season. All flights were conducted with drones equipped with RGB and Multispectral cameras. The team tookground-truth data as well as agronomic and physiological data on chlorophyll content, canopy temperature, plant height,heading dates, and grain yield.Fourteen flights were conducted at College Station covering 723 plots. Nine flights were conducted at McGregor covering 1,340 plots. Three flights were conducted at the Multi-State Wheat Rust Evaluation Nursery at Castroville to capture response of plants to leaf rust, covering 60 plots and 14,000 observation rows.Twenty-five flights were conducted at Bushland covering 2000 plots per flight. G2F HIPS corn hybrid plots were planted both in Texas and Nebraska.In Texas, the team led by co-PI Seth Murray conducted UAV flights and image data collection throughout the growing season on a weekly basis. In Nebraska (the Lincoln and Mead sites), seven UAV flightswereconducted at each site starting early June till early September. Both RGB and multispectral images were acquired. At the time of tasseling, corn leaf samples (n = 1600) were collected from each genotype. The leaf samples were first subjected to ASD scanning to acquireVIS-NIR-SWIR hyperspectral readings and then processed to acquire chlorophyll concentration, leaf thickness, and leaf water content. A portion of the leaf samples (n=250) were sent to a commercial lab for chemical analysis including total nitrogen, phosphorus, and potassium. Another subset of the leaf samples (n= 96) were measured for photosynthesis and gas exchange traits using LICOR instruments. At UNL's Spidercam field phenotyping site, winter wheat plots and G2F HIPS corn hybrid plots were establishedand measured throughout their respective growing season. We maintained an average measurement frequency of twice each week from crop emergence to maturation. All imaging modules (including a hyperspectral camera) on the Spidercam were used for the measurement. Weekly ground-truth measurements were taken at the plot scale for plant height, leaf chlorophyll concentration, stomatal conductance, and leaf area index. Conventional phenotyping data such as flowering date and final yield/quality were also collected. Obj. 2: Using the data collected from Obj. 1, we have been developing both conventional and machine-learning models to predictleaf physiological and chemical traits of corn leavesfrom the ASD hyperspectral reflectance data. The conventional method included the calculation of vegetation indices from the hyperspectral data, and the machine-learning methods included Partial Least Squares Regression, Random Forest, Support Vector Regression, and Neural Networks. Our preliminary results showthat several leaf traits (chlorophyll, nitrogen, leaf thickness, etc.) can be predicted from hyperspectral data satisfactorily. Weinvestigated how the plot-level traits such as LAI and crop water use (evapotranspiration, or ET) can be modeled from the UAV multispectral and thermal infrared images. For LAI, we found that usingboth the coverage of green pixels and plot height can predict plot-level LAI quite accurately. Including the canopy spectral information (such as red-edge NDVI) and canopy temperature can further improve LAI modeling. For ET, we found that (1) accurate segmentation of plot-level thermal IR images to separate the vegetationtemperature from the soil temperature can potentially improve ET estimation; (2) the combination of instantaneous weather data (solar radiation, air temperature, vapor pressure deficit, wind speed) and the UAV multispectral/thermal IR images allows the use of a two-source surface energy balance model to estimate instantaneous ET from the crop; and (3) flying UAV multiple times during a day enable theestimation ofinstantaneous ETmultiple times during a day, therefore provides an opportunity to improve the daily and seasonal estimation ET from remote sensing. Co-PI Thomasson's group has designed a new air-ground cooperative robotic system for UAV image calibration. This automated system has the potential to greatly reduce the amount of work needed for the geometric and radiometric calibration of multispectral, hyperspectral, and thermal IR images from UAV platforms. Obj. 3. In Texas, the UAV images from both wheat and maize (G2F) programs aretransferred to and stored at UAS Data Hub (Texas A&M University at Corpus Christi). In Nebraska, the wheat teamhas beenworking on developing a pipeline for analyzing and utilizing the UAV imagedata in the breeding program. In addition, the new data scientist (Dr. Yu Pan) is designing and developing a new agricultural data management and analysis platform. All the high throughput plant phenotyping data generated from this project will be curated in this new platform for more efficient analysis, modeling and dissemination. Obj. 4. Co-PI Leah Sandall and graduate student Catherine Mick are working to developonline educational materials for high-throughput plant phenotyping. We have finalized the plan to develop four modules for morphological and physiological traits of plants from the UAV platforms. These four modules are plant height, LAI, canopy evapotranspiration, and leaf/canopy hyperspectral reflectance data. The first two modules (plant height and LAI) are now under development. Each module contains a webpage (with texts and illustrations), a 20-minute video, and quiz-type questions for evaluation. After completion, these modules will be placed in UNL's Plant and Soil Science eLibrary for dissemination. Co-PI Yeyin Shi taught a 3-week mini course "Aerial Imagery Processing and Analysis Using Python" in Jan/2021. The course focused on the collection, processing, and crop trait extraction of UAV images, as well as python programming. Fifteen students (both senior undergraduates and graduate students) from Biological Systems Engineering and Agronomy & Horticulture registered and attended this course. This course fills an emerging need at UNL for courses in data-driven plant science, phenotyping,and digital agriculture. We have worked with UNL's Office of Graduate Studies to plan for theresearch experiential learningprogram in summer 2022. We have identified four researchers from our project team who are willing to host undergraduate researchers. We are developing a website to introduce the HIPS project at UNL in general, as well as the research focus of each participating lab. This website will be used for recruiting in the next reporting period.

Publications

  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Adak, A., Murray, S. C., Anderson, S. L., Popescu, S. C., Malambo, L., Romay, M. C., & de Leon, N. (2021). Unoccupied aerial systems discovered overlooked loci capturing the variation of entire growing period in maize. The Plant Genome, e20102.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Adak, A.; Murray, S.C.; Bo~inovi?, S.; Lindsey, R.; Nakasagga, S.; Chatterjee, S.; Anderson, S.L., II; Wilde, S. Temporal Vegetation Indices and Plant Height from Remotely Sensed Imagery Can Predict Grain Yield and Flowering Time Breeding Value in Maize via Machine Learning Regression. Remote Sens. 2021, 13, 2141. https://doi.org/10.3390/rs13112141
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Han, X., J. A. Thomasson, V. Swaminathan, T. Wang, J. Siegfried, R. Raman, N. Rajan, and H. Neely. 2020. Field-based calibration of unmanned aerial vehicle thermal infrared imagery with temperature-controlled references. Sensors 20:7098; doi: 10.3390/s20247098.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Wang, L., Jiating Li, Lin Zhao, Biquan Zhao, Geng Bai, Yufeng Ge, and Yeyin Shi. 2021. Investigate the potential of UAS-based thermal infrared imagery for maize leaf area index estimation. Proceedings Volume 11747, Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VI; 1174703. https://doi.org/10.1117/12.2586694
  • Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Zhao, L., Lin Wang, Jiating Li, Geng Bai, Yeyin Shi, and Yufeng Ge. 2021. Toward accurate estimating of crop leaf stomatal conductance combining thermal IR imaging, weather variables, and machine learning. Proceedings Volume 11747, Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping VI; 117470L. https://doi.org/10.1117/12.2587577