Source: OREGON STATE UNIVERSITY submitted to NRP
MODELING STAND AND LANDSCAPE PROCESSES USING UNMANNED AERIAL SYSTEMS
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
MCINTIRE-STENNIS
Reporting Frequency
Annual
Accession No.
1009739
Grant No.
(N/A)
Cumulative Award Amt.
(N/A)
Proposal No.
(N/A)
Multistate No.
(N/A)
Project Start Date
Jul 1, 2016
Project End Date
Jun 30, 2021
Grant Year
(N/A)
Program Code
[(N/A)]- (N/A)
Recipient Organization
OREGON STATE UNIVERSITY
(N/A)
CORVALLIS,OR 97331
Performing Department
Forest Engineering, Resources & Management
Non Technical Summary
A massive effort and significant resources have been placed in measuring and assessing terrestrial entities using remote sensing data supplied by flying platforms. Irrespective of the platform, the overhead viewing geometry enables only a nadir view of the surface. While valuable information can be extracted from overhead data, there are areas that can benefit from information derived from the addition of perspective (i.e., side-looking) views, particularly forests. Nadir views of a forest provides information about canopy, particularly spectral composition and vertical variation, but fail to reveal stem and branching details. Perspective views can supply information located under the main canopy and have become increasingly attractive, as they do not rely on the expensive, sensitive, and cumbersome equipment used by platforms flying above the canopy. Furthermore, advancements in very near infrared sensors and computer vision encouraged the acquisition of forestry data with remote sensing techniques that do not rely on stereopsis. Nevertheless, the implementation of these new technologies in forestry faces significant difficulties, as obstacles and similarities among trees reduce the performance of rendering algorithms. However, the reduced costs of the equipment and software needed to obtain valuable information from sensors moving under the canopy recommend pursuing the development of this novel technology for resources estimation and monitoring. The Association for Unmanned Vehicle Systems International, which is the main forum for small terrestrial and aerial platforms carrying sensors, predicts an exponential growth in the aerial field, with more than 100,000 new jobs and a market of almost $100 billion. The Washington Post supports the above figures, forecasting that unmanned aerial systems (UAS) skills will be in high demands in the near future, a significant number being dedicated to agricultural application, such as crop monitoring, resources estimation, and treatments application.The project aims aretodevelop inventory techniquesusing UAS, andto modelforest processes by merging remote sensed acquireddata. The project will use different algorithms, mostly based on computer vision, to convert remote sensing data in information usable by forest managers, researchers, educators, and public. The algorithms will help in fusing data from above forest canopy with measurements collected under the canopy. The project is expected to have a significant impact on forest inventory, as will change the standard procedures of acquiring and processing data. There are three main benefits of the project to society: 1) reduce inventory costs while increase accuracy, 2) time to deliver the information to public will be significantly reduced, and 3) decrease the risk ofinjury during field phase.
Animal Health Component
70%
Research Effort Categories
Basic
10%
Applied
70%
Developmental
20%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
1230612310030%
1230613310030%
1230622310020%
1360613107020%
Knowledge Area
136 - Conservation of Biological Diversity; 123 - Management and Sustainability of Forest Resources;

Subject Of Investigation
0612 - Conifer forests of the West; 0622 - Broadleaf forests of the West; 0613 - Mixed conifer-broadleaf forests;

Field Of Science
1070 - Ecology; 3100 - Management;
Goals / Objectives
The aim of the project is threefold: 1) identification of optimal inventory techniques and designs using UAS, 2) estimation of forest resources and 3) modeling of forest processes from fused data supplied by UAS, satellites, and airplanes. To achieve the goals the project will focus on five objectives:Optimize forest inventory using a combination of low-cost flying platforms and accurate tree segmentation algorithmsDevelop a method of forest inventory using "under the canopy moving sensors"Describe forest complexity using metrics derived from the data processed using 3D rendering algorithmsDevelop a method of fusing above and below canopy data supplied by UAS, satellites, and airplanesIdentify UAS configurations suitable for various inventory tasks, with focus on forest biodiversity, fractal measures of forest, and quantification of information and energy fluxes below canopy.
Project Methods
The proposed research builds on the airborne raster and vector data that represent vegetation since mid-1980s in 2D or 3D settings. The existing data will be supplemented with high precision orthophotos (resolution under 1 inch) and photogrammetric point clouds for above and below canopy obtained using UAS. The fusion will allow not only 3D representation of reality with sub-inch accuracy, but model landscape level processes by using individual trees as an elementary modeling unit. To ensure correct referencing of the photogrammetric point clouds two UASs will be flown, one above and one under the canopy. The UAS above the canopy will collect information on vegetation from the nadir perspective and ensure correct geopositioning through GPS. The UAS below the canopy will collect information on stem and tree branching, and communicate its relative position to the UAS flying above canopy. The present study will develop a method of collecting information based on two UASs flying in an above / below configuration. The method will be focused on improving flight synchronization algorithms, the considered method being a decentralized consensus algorithm8 and virtual structure and motion synchronization9. Once the flying method is completed, a forest inventory procedure will be developed, which will estimate common stand attributes, such as diameter, volume, basal area and trees /acre, and information characterizing the ecosystem (such as Shannon index), and the branching system (such as fractal dimension of the crown).The combination of information supplied by UASs and satellites / airplanes will serve in development of a new scalar metric describing species composition, and application of two measures, particularly information distribution and 3D fractal dimension, in predicting ecosystem dynamics, particularly growth and yield. This section of the study will be continuing the work of Magurran10, Peet11, and Tuomisto12 on biodiversity, Falconer13 and Zeide14 on fractals, and Jaynes15 and Zhang and Harte16 on information theory. The measures will be used to develop spatially explicit models for information and energy fluxes throughout the forest ecosystem. The models will be implemented in spatially explicit ecosystem dynamics models such as 3PG17or Sortie-ND18.

Progress 07/01/16 to 06/30/21

Outputs
Target Audience:The target group is represented by researchers acting in remote sensing and UAV areas and practitioners using information acquired with the sensors installed on UAVs. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?During the reporting period I have trained two groups of individual in using handheld lidar units, one from the forest extension and one from the Agricultural Research Service. How have the results been disseminated to communities of interest? Present the forest inventory method from terrestrial point clouds to the to the 2020 Western Mensurationists conference, Vancouver Canada What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? The project achieved most of its goals beyond expectations, as many of the products obtained were so advanced that made a significant impact on forestry. Among the five objectives of the project, namely optimization of a forest inventory using a combination of low-cost flying platforms and accurate tree segmentation algorithms, development a method of forest inventory using "under the canopy moving sensors", description of forest complexity using metrics derived from the data processed using 3D rendering algorithms, development of a method of fusing above and below canopy data supplied by UAS, satellites, and airplanes and identification of UAS configurations suitable for various inventory tasks, with focus on forest biodiversity, fractal measures of forest, and quantification of information and energy fluxes below canopy, two were fulfilled beyond expectations. The two objectives were associated with the second objective (method of forest inventory) and the third objective (description of forest complexity). From practical perspective, without doubt, the most important achievement of the project was the development of an algorithm that executes a forest inventory with high accuracy and precision using point clouds. The algorithm not only that estimates the dimensions of each tree, therefore a parameter not a statistic, but also allows computation of many attributes that were hard to quantify with the current inventory methods, such as lean and sweep, or the size of crown. From theoretical perspective, the project was able to develop a measure of forest complexity that integrates all the existing measures. The rest of the objectives were completed within expectation, as I proposed an inventory method that combines various platforms, including UAS, for forest inventories (objective #1 and 5) and I developed a new method of fusing above and below canopy data (Objective #4). All the objectives were attained with at least two publications, some even more.

Publications

  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Strimbu, B.M.; Mueller-Warrant, G.; Trippe, K. Agricultural Crop Change in the Willamette Valley, Oregon, from 2004 to 2017. Data, 6, 17. https://doi.org/10.3390/data6020017
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Strimbu, B.M., Amarioarei, A. & Paun, M. Nonlinear parsimonious forest modeling assuming normal distribution of residuals. Eur J Forest Res 140(3). https://doi.org/10.1007/s10342-021-01355-2
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Heffernan, S.; Strimbu, B.M. Estimation of Surface Canopy Water in Pacific Northwest Forests by Fusing Radar, Lidar, and Meteorological Data. Forests 2021, 12, 339. https://doi.org/10.3390/f12030339
  • Type: Books Status: Published Year Published: 2021 Citation: Strimbu BM. Unmanned aerial systems for land monitoring. 81 p. Oregon State University. ISBN 9781732664401


Progress 10/01/19 to 09/30/20

Outputs
Target Audience:The target group is represented by researchers acting in remote sensing and UAV areas and practitioners using information acquired with the sensors installed on UAVs. In the summer of 2020 I have trained forest extension personnel in using handheld lidar to acquire terrestrial point clouds aiming at forest inventory. Furthermore, I have trained the same people in merging terrestrial point clouds with the point clouds developed from the images acquired with UAV. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Mentor one undergraduate students in processing terrestrial point clouds Coordinate three students in processing UAV images aiming at seedling recognition Coordinate two students in assessment of herbicides impact on reed canary grass How have the results been disseminated to communities of interest? Present the forest inventory method from terrestrial point clouds to the IUFRO conference dedicated to the Sustainable Forest Management from Brasov Romania Present applications of UAV in assessment of the visual impact of forest harvests using the UAV to the 2020 Western Mensurationists conference What do you plan to do during the next reporting period to accomplish the goals? complete data fusion of point clouds acquired above and below canopy complete descsription ofcomplexity using metrics derived from the data processed using 3D rendering algorithms

Impacts
What was accomplished under these goals? In 2020 thre goals were achived: development of a forest inventory strategy that fuzes terestrial lidar with images acquired with UAV completed a forest inventory algorithm that uses mobile lidar under the canopy. identified UAS configurations suitable for various tasks, from forest inventory, to biodiversity assessment and fire resilience.

Publications

  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Paun, M.; Gunaime, N.; Strimbu, B.M. Impact of Algorithm Selection on Modeling Ozone Pollution: A Perspective on Box and Tiao (1975). Forests 2020, 11, 1311.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Riddell, A.P., Fitzgerald. F.A.,Qi, C., and Strimbu, B.M. Classification Strategies for Unbalanced Binary Maps: Finding Ponderosa Pine (Pinus ponderosa) in the Willamette Valley. Remote Sensing 2020, 12(20), 3325
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Amarioarei, A., Paun, M., and Strimbu BM. Development of Nonlinear Parsimonious Forest Models Using Efficient Expansion of the Taylor Series: Applications to Site Productivity and Taper. Forests 11(4): 458-484
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Marin, Gh., Abrudan, I, and Strimbu BM. Regional Variability of the Romanian Main Tree Species Growth Using National Forest Inventory Increment Cores. Forests 11(4): 409-427


Progress 10/01/18 to 09/30/19

Outputs
Target Audience:The target group is represented by researchers acting in remote sensing and UAV areas and practitioners using information acquired with the sensors installed on UAVs. The PI described various approaches to extract information from images or lidar to industry and extension agency. A PhD students supervised by the PI developed an algorithm for stem measurements from point cloud data. A MS student supervised by the PI developed a method of assessing the surface canopy water using lidar and radar imagery. Both results were presented to two forest management companies: Green Diamond and Northwest Management. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The project did not provided any training opportunities yet, but now we are able to develop a professional development module that would train people involved in forestry activities in estimation of the forestry resources. How have the results been disseminated to communities of interest?During the reporting period, the PI showed the tree measurement algorithm to three forest extension people. Furthermore, the PI is writing a manual on how to use UAS in forest monitoring, that would be used by EPA and watershed councils in the west USA. What do you plan to do during the next reporting period to accomplish the goals?During the next reporting period, I plan to act on two directions, one focused on delivering information to be used by forestry professional and one on theoretical advancements on the usage of point loud data in forest inventory. From delivering information perspective, I aim at completion of two activities: Patent the tree measurement algorithm Complete the manual showing in details the steps needed for UAS based land monitoring From theoretical perspective, I aim at developing an algorithm that would measure multiple trees at the time from point cloud data.

Impacts
What was accomplished under these goals? The project affects the forest activities by increasing the accuracy andprecision of the forest inventory, as well as increases the speed of acquiring and delivering the relevant information. The projects acts on two directions, one is estimation of various parameters of interest, and one on acquisition of data that is suitable for analysis. The project succeeded in both directions, as we developed a new method of estimation the dimension of tree attributes from point clouds, as well as a method of synchronizing the data acquired from above canopy with the data from below canopy. During this reporting period, the PI coordinated the development of an algorithm that estimates the dimension of the trees, but most importantly classifies the points describing a tree in two classes: stem and crown, a major achievement in forest inventory and machine learning. The algorithm that fuses the data from above canopy with below canopy marks a major advancement in geo-referencing below canopy information, where no reliable GPOS signal is acquired. This part of the project will provide the platform of acquiring information with devices unable to locate the trees but suppling precise dimensional values that can be used to position the tree.

Publications

  • Type: Journal Articles Status: Published Year Published: 2019 Citation: Strimbu, B.M.; Qi, C.; Sessions, J. Accurate Geo-Referencing of Trees with No or Inaccurate Terrestrial Location Devices. Remote Sens. 2019, 11, 1877.
  • Type: Journal Articles Status: Published Year Published: 2019 Citation: Fang, R.; Strimbu, B.M. Comparison of Mature Douglas-Firs Crown Structures Developed with Two Quantitative Structural Models Using TLS Point Clouds for Neighboring Trees in a Natural Regime Stand. Remote Sens. 2019, 11, 1661.


Progress 10/01/17 to 09/30/18

Outputs
Target Audience:The target group is represented by researchers acting in remote sensing and UAV areas as well as practitioners using information acquired with the sensors installed on UAVs and terrestrial vehicles. The PI described various approaches to extract information from images or lidar to foresters from Oregon and Washington. TwoPhD students,whose research is coordinated by the PI, presented to researchers,academia, and industryparticipatingat the premier international conference on Natural Resource Modeling two outcomes based on this project: a stemmeasurements algorithm from point clouds and a crown model alsofrom point clouds. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?The main dissemination avenues of the results were thru publication in peer reviewd journals and to presentaions at conferences. To this end we have published 5 papers stemmed from this project and I have presnetd the results to three conferences: Strimbu, B. M., Paun, M., Montes, C., Popescu, S., 2018 Joint Southern And Northeastern Mensurationists and IUFRO 4.01 Conference, "A scalar measure tracing tree species composition in space or time," Blacksburg VA. (October 29, 2018). Strimbu, B. M., Fang, R, "Branch sampling of tree structural models fitted from lidar point clouds, a case study of an experimental Douglas-fir forest," Blacksburg VA. (October 28, 2018). Strimbu, B. M., Garms, C, 20184 SAF National Convention, "A Comparison of Three Point Cloud Platforms Used in Forest Inventory," Portland OR. (October 6, 2018). Strimbu, B. M., Pacific Logging Congress, "Road maintenance and stream bead estimation using lidar," Corvallis OR. (September 13, 2018). What do you plan to do during the next reporting period to accomplish the goals? Publish the model describing the tree crown architecture as arelationship to their neighbors Publish the algorithm measure the tree dimensions from point clouds Publish the algorithm precisely geo-referencing the trees usingimprecise device

Impacts
What was accomplished under these goals? Inthe last year of the project I have succeded in completing the theoretical part of the follwoing three objectives: Develop a method of forest inventory using "under the canopy moving sensors" Describe forest complexity using metrics derived from the data processed using 3D rendering algorithms Develop a method of fusing above and below canopy data supplied by UAS, satellites, and airplanes

Publications

  • Type: Theses/Dissertations Status: Submitted Year Published: 2018 Citation: Heffernan S. Estimation of surface canopy water in Pacific Northwest forests by fusing radar, lidar, and climatic data. MS Thesis. Oregon State University
  • Type: Journal Articles Status: Published Year Published: 2017 Citation: Rong F and Strimbu BM 2017. Photogrammetric point cloud trees. Mathematical and Computational Forestry and Natural-Resource Sciences 9(2) 30-33
  • Type: Journal Articles Status: Published Year Published: 2018 Citation: Strimbu BM, Paun M, Popescu SC, Montes C. 2018. A scalar measure tracing tree species composition in space or time.Physica A: Statistical Mechanics and its Applications 512: 682-692


Progress 10/01/16 to 09/30/17

Outputs
Target Audience:The target group is represented by researchers acting in remote sensing and UAV areas and practitioners using information acquired with the sensors installed on UAVs. The PI described various approaches to extract information from images or lidar to foresters from Wheeler County. One of the graduate students supervised by the PI presented to the several industry representative a segmentation algorithm that identify and measure junipers using NAIP images in combination with digital terrain models. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?Help Wheler County in the creation of a UAV based survey system How have the results been disseminated to communities of interest? Present at a new technique of estimating the water storage using radar and lidar data at the Western Mensurtionist annual conference Present a new method of segmenting trees by fusing multispectral NAIP imagery with lidar derived products at the Southern Mensurationist conference Present a new estimation method for nonlinear models at the IUFRO congress Present applications of UAV in surveying regeneration and damage induced by broken tops at the annual conference of the Western Region of the Council of Forest Engineering What do you plan to do during the next reporting period to accomplish the goals? enhance the stem segmentation algorithms from photogrammetric point clouds start a procedure to estimate the residual biomass using imagery assess the perfomaces of various DTM algorithms on point clouds developed from UAV

Impacts
What was accomplished under these goals? This year the following goals were achived: 1. develop a metric to descibe forest complexity to be sued in conjunction with the 3D rendering algorithms 2. develop a method to fuse the above and below canopy data

Publications

  • Type: Journal Articles Status: Published Year Published: 2017 Citation: Strimbu, B. M., A. Amarioarei, and M. Paun. 2017. A parsimonious approach for modeling uncertainty within complex nonlinear relationships. Ecosphere 8(9):e01945. 10.1002/ecs2.1945


Progress 07/01/16 to 09/30/16

Outputs
Target Audience: Nothing Reported Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?In the next year the following activities are planned: Objective 1: Develop a strategy that integrates data supplied by consumer grade cameras installed on flying platform with tree segmentation algorithms Objective 2: Complete the development of the software that measure stem dimensions from unreferenced photos, particularly taper, volume, and sweep. Objective 3 and 5: Execute fractal measurements of tree branchingand tree location within stand. Publish one paper on complexity of mixed species stands. Objective 4: Fuse radar and lidar data for water stored by a stand after rain.?

Impacts
What was accomplished under these goals? In the firstthree month of the project a method for estimation the tree taperfrom photogrammetricwas developed, pursuing the second objective of the project. The methodis currently tested for accuracy andprecision. To deploythe model for commercial applicationsa Matlab program waswritten.

Publications