ADVANCING REMOTE SENSING APPLICATIONS FOR SUSTAINABLE FOREST MANAGEMENT IN INDIANA

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: COMPLETE
• Funding Source: MCINTIRE-STENNIS
• Reporting Frequency: Annual
• Accession No.: 1003456
• Project Start Date: Oct 20, 2014
• Project End Date: Sep 30, 2019
• Program Code: [(N/A)]- (N/A)

Recipient Organization
PURDUE UNIVERSITY
(N/A)
WEST LAFAYETTE,IN 47907

Performing Department
Forestry & Natural Resources

Non Technical Summary
Hardwood forests are important timber sources and provide critical wildlife habitat, and their overall quality is closely related to management activities (Annand and Thompson 1997, Jenkins and Parker 1998, Morrissey et al. 2010). Forest management relies on spatially-explicit information on tree species composition and size structure. The conventional remote sensing data analysis is not effective enough to derive forest cover maps with sufficient information for forest management. The existing forest cover data in Indiana contain limited species information and no tree-size/age information. For example, the 1992 and 2001 National Land Cover Data classified all the hardwood forests into one forest type and are not useful for designing tree species/size-dependent sylviculture approaches. The land cover data developed by the Indiana Gap Analysis Project did not consider subclasses of hardwood forests either, and the overall accuracy of the map product was only 70.98%. When such low-accuracy forest-cover maps are used for forest management planning, actions may be unexpectedly misled due to error propagation (Shao et al. 2001&2003, Shao and Wu 2008). Therefore, it is important to obtain accurate forest cover maps with adequate forest-type and site-structure information. Such map products are broadly needed for intensive management of hardwood forest ecosystems, both publically and privately owned, in Indiana. Various remote sensing techniques have been extensively used in boreal and tropical forests but their applications in the central hardwood forest region are still limited. Our remote sensing experiment in Indiana will have broader implications to forest mapping in the central hardwood forest region.

Animal Health Component

50%

Research Effort Categories

Basic

30%

Applied

50%

Developmental

20%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
123	0613	1060	60%
136	0613	1070	40%

Knowledge Area
136 - Conservation of Biological Diversity; 123 - Management and Sustainability of Forest Resources;

Subject Of Investigation
0613 - Mixed conifer-broadleaf forests;

Field Of Science
1060 - Biology (whole systems); 1070 - Ecology;

Keywords

remote sensing

lidar

obia

forest mapping

forest structure

forest management

hardwood forest

Goals / Objectives
The overall objective is to improve the applications of remote sensing data in sustainable management of the central hardwood forest ecosystems through institution capacity that involves both the producers and users of remote sensing algorithms. Specific objectives are threefold:1. To apply OBIA in characterizing forest landscapesWe will use state-of-the-art protocols and interfaces of OBIA techniques, which have been used and advanced in our previous land-use projects, to improve automated OBIA algorithms with multispectral remote sensing data acquired in Indiana. The goal is to map precise forest landscape characterizations that help forest sustainability in south-central Indiana.2. To apply LiDAR data in mapping forest structureWe will develop algorithms to estimate 3-D structure of hardwood forests with LiDAR data that have been acquired by the State of Indiana. The goal is to develop structure-oriented forest data layers, including forest canopy surface/gaps and forest tree heights for hardwood forests in south-central Indiana.3. Mapping forest structure and composition for sustainable forest managementWe will incorporate LiDAR-derived forest structure information into forest classification with OBIA. The goal is to develop improved forest maps with integrated information of forest structure and spectral characteristics and make the forest maps available for intensive forest management activities in south-central Indiana as an example for entire central hardwood forest regions.

Project Methods
Remote sensing experiments will be conducted mainly in the Morgan-Monroe State Forest and Yellowwood State Forest of Indiana, where a long-term study of forest management is under way to evaluate even and uneven-aged harvesting practices on wildlife and vegetation communities in oak-dominated forest ecosystems. Extensive ground data are available for these forest stands with various conditions and they are useful for developing and testing remote sensing algorithms for this project.We will take advantage of the existing multispectral remote sensing imagery in this project, such as Landsat TM satellite data and digital orthophotography acquired by the National Agriculture Imagery Program (NAIP). We will use commercial spectral data, either spaceborne or airborne, when available.In 2011, the state of Indiana initiated a 3-year project (2011-2013) to produce airborne LiDAR data covering all 92 Indiana counties. The data are multiple returns discrete LiDAR point cloud at 1.5-meter average point spacing, and are used primarily for precise terrain mapping. This expensive LiDAR dataset will be utilized for forest-structure mapping in this project.The remote sensing methods will include three parts:1. Object-based forest classificationThis is an expansion of our past research that focused on OBIA applications in urban landscapes. Cooperators will include Drs. Jenkins, Wu, and Zhang as well as their graduate students and post-docs.The OBIA builds on older concepts of segmentation, edge-detection, feature extraction and classification that have been used in remote sensing image analysis for decades (Blaschke 2010). OBIA naturally emerged because the minimum objects on ground are sometimes beyond pixels. OBIA was technically realized through image segmentation that divides the image into relatively homogeneous groups of pixels or image objects (Pal and Pal 1993). Image objects being segmented can be classified by either simple algorithms such as nearest neighbor classifier or sophisticated algorithms such as combining shape and spatial information, ancillary data, as well as spectral response of image objects characteristics (Benz et al. 2004, Zhou and Troy 2008).This proposed project will focus on OBIA of high-resolution digital orthophotography and its derivatives (e.g. Digital Surface Model), which are readily available in Indiana. Landsat TM data will be integrated into the classification procedure to enhance spectral contrasts between different forest types. A commercial program called eCognition (Benz et al. 2001) will be used to carry out the classification experiment, which will target the discriminations of forest types in forested landscapes with different management backgrounds. OBIA can remove the 'salt and pepper effect' commonly found in classification results from traditional per pixel approaches (Yu et al. 2006, Li and Shao 2013), suitable for practicing patch-based forest management activities.2. Applying LiDAR data in forest structure mappingWe did preliminary experiments with the existing LiDAR data for a portion of our study area and the results of forest canopy structure are convincing. Cooperators will include Drs. Crawford, Lefsky, and Saunders as well as their graduate students and post-docs.The LiDAR is an active remote sensing technology that measures distance by illuminating a target with a laser and analyzing the reflected light. Airborne LiDAR is a useful technology to create a number of mapping products, such as topography and tree height. LiDAR either records the complete range of the energy pulse (intensity) reflected by surfaces in the vertical dimension or samples the returned energy from each outgoing laser pulse in the vertical plane. The former is referred to as waveform while the latter is called discrete-return systems. The spot size of laser pulse (point) is a footprint. The point spacing is spatial resolution. High-resolution, small-footprint LiDAR is useful to count trees and measure tree height, crown width, and crown depth, based on which, the standing volume of timber can be estimated on an individual tree basis, or at a stand level. We will use Matlab computer language to develop and test the discrete-LiDAR-data algorithms/models.3. Incorporating LiDAR data into OBIA proceduresThe fusion of optical and LiDAR data will be a new approach in our research. Broader cooperation will be particularly helpful and include all the cooperators and their graduate students and post-docs.The metrics involved in the OBIA procedure in the first method will be derived from multispectral imagery. We will combine these imagery-derived metrics with LiDAR-derived data layers to enhance information for more detailed forest classification. We will compare different classification approaches resulted from various combinations of selected LiDAR-data derivatives and selected imagery derivatives. It is our hope that optimum combinations of the two sources of derivatives will be determined for the development of management-oriented forest maps at the local and regional scales.We will use ground survey to develop reference datasets for accuracy assessment. A series of performance statistics, including producer's and user's accuracies, overall accuracy, and Kappa coefficient of agreement, will be used to compare different classification approaches.

Progress 10/20/14 to 09/30/19

Outputs
Target Audience:Forestry professionals, forest surveyers, forest researchers Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?We demonstrated tree diameter algorithm at the annual meeting of The USDA Hardwood Tree Improvement and Regeneration Center. How have the results been disseminated to communities of interest?Yes.The USDA Hardwood Tree Improvement and Regeneration Center What do you plan to do during the next reporting period to accomplish the goals?We will work with the organization of forest inventory and analysis of USDA Forest Service to improve the tree diameter measurement algorithm. We will publish an OBIAmethodin tree canopy measurements withumanned aerial systems.

Impacts
What was accomplished under these goals? Figured out a way to identify individual trees with low-density lidar data; summarized the inherent properties of the overallaccuracy metric in remote sensing classification;developed an algorithm for digitally measuring tree diameters; acquired newimagery for natural forests and plantations with umanned aerial systems.

Publications

• Type: Journal Articles Status: Published Year Published: 2019 Citation: Shao, G., G.F. Shao, and S.L. Fei. 2019. Delineation of individual deciduous trees in plantations 1 with low-density LiDAR data. International Journal of Remote Sensing 49(1): 346ÿ¢ÿ¿ÿ¿363.
• Type: Journal Articles Status: Published Year Published: 2019 Citation: Liao, J.F., G.F. Shao, C.P. Wang, L.N. Tang, Q.L. Huang, and Q.Y. Qiu. 2019. Urban sprawl scenario simulations based on cellular automata and ordered weighted averaging ecological constraints. Ecological Indicators 107, 105572
• Type: Journal Articles Status: Published Year Published: 2019 Citation: Shao, G.F., L.N. Tang, and J.F. Liao. 2019. Overselling overall map accuracy misinforms about research reliability. Landscape Ecology 34(11): 2487ÿ¢ÿ¿ÿ¿2492.
• Type: Book Chapters Status: Published Year Published: 2019 Citation: Shao, G.F. 2019. Optical remote sensing. In: International Encyclopedia of Geography: People, the Earth, Environment, and Technology. D. Richardson (ed.). Wiley & Sons, Inc., P2390ÿ¢ÿ¿ÿ¿2395.

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

Outputs
Target Audience:About 30 Foresters and forestry professionals associated with the Society of American Foresters in Indiana. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?We introduced our experience to foresters and forestry professionals associated with the Society of America Foresters (SAF) in Indiana. ? How have the results been disseminated to communities of interest?The USDA Hardwood Tree Improvement and Regeneration Center is planning to use the UAS to improve forest inventory. ? What do you plan to do during the next reporting period to accomplish the goals?We will explore automated, operational applications of UAS imagery in monitoring and managing hardwood forests in Indiana. We will start to develop ground-based photographic methods for tree-level forest survey. ?

Impacts
What was accomplished under these goals? We conduct an experiment on Unmanned Aerial Systems (UAS) imagery acquisition and analysis for a hardwood forest stand in Indiana. We obtained preliminary results but achieved important findings. Ordinary, inexpensive photographs taken with drone at a low altitude provide high spatial resolution imagery, with which we obtained true-ortho photos and digital surface models (DSM). Furthermore, we were able to obtain close correlations between photo-measured tree heights and ground measured tree heights, and between photo-measured crown width and ground measured diameter at breast height (DBH). This suggests that UAS technology is a reliable tool for stand-level forest survey. ?

Publications

• Type: Journal Articles Status: Published Year Published: 2018 Citation: 4. Dai, L.M., S.L. Li, W.M. Zhou, L. Qi, L. Zhou, Y.W. Wei, J.Q. Li, G.F. Shao, and D.P. Yu. 2018. Opportunities and challenges for the protection and ecological functions promotion of natural forests in China. Forest Ecology and Management 410:187192.
• Type: Journal Articles Status: Published Year Published: 2018 Citation: 3. Shao, G., G.F. Shao, J. Gallion, M.R. Saunders, J.R. Frankenberger, S.L. Fei. 2018. Improving lidar-based forest aboveground biomass estimation with the regard to site productivity in temperate hardwood forests. Remote Sensing of Environment 204: 872882.
• Type: Journal Articles Status: Published Year Published: 2018 Citation: 2. Hua, L.Z., J.F. Liao, H.X. Chen, D.K. Chen, and G.F. Shao. 2018. Assessment of ecological risks induced by land use and land cover changes in Xiamen City, China. International Journal of Sustainable Development & World Ecology 25(5): 439-447.
• Type: Journal Articles Status: Published Year Published: 2018 Citation: 1. Wang, Z.; Tang, L.; Qiu, Q.; Chen, H.; Wu, T.; Shao, G. 2018. Assessment of regional ecosystem health - A case study of the Golden Triangle of southern Fujian Province, China. Int. J. Environ. Res. Public Health 2018, 15, 802.

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

Outputs
Target Audience:Forestry professionals and forestland owners in Indiana Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?We made a demonstration at the event of Purdue GIS day. How have the results been disseminated to communities of interest?The lidar data application in forest biomass estimation is new in Indiana. There have been 129 downloads across three papers on lidar applications from Purdue E-Pubs. What do you plan to do during the next reporting period to accomplish the goals?We will analyze forest ecosystem services with the InVEST modeling on the aspect of landscape and ownership fragmentation in Indiana. We will explore operational applications of Unmanned Aerial Systems (UAS) imagery in monitoring and managing family-owned hardwoods in Indiana.

Impacts
What was accomplished under these goals? We incorporated site productivity information into lidar data analysis and obtained better estimations of aboveground biomass for hardwood forests in Indiana. We published a review paper on operational forest fire monitoring with infrared remote sensing. We finalized an ecosystem service model known as InVEST for analyzing spatially explicit return on investment to private forest conservation for water purification in Indiana and published a paper in the journal of Ecosystem Services.

Publications

• Type: Journal Articles Status: Published Year Published: 2017 Citation: Hua, L.Z., G.F. Shao and J.Z. Zhao. 2017. A concise review of ecological risk assessment for urban ecosystem application associated with rapid urbanization processes. International Journal of Sustainable Development & World Ecology 24(3): 248261.
• Type: Journal Articles Status: Published Year Published: 2017 Citation: Wang Y.Y., S. Atallah, and G.F. Shao. 2017. Spatially explicit return on investment to private forest conservation for water purification in Indiana, USA. Ecosystem Services 26: 4557.
• Type: Journal Articles Status: Published Year Published: 2017 Citation: Chen, X., G. Zhou, Y. Chen, G. Shao, and Y. Gu. 2017. Supervised multiview feature selection exploring homogeneity and heterogeneity with ?1,2-norm and automatic view generation. IEEE Transactions on Geoscience and Remote Sensing 55(4): 20742088.
• Type: Journal Articles Status: Published Year Published: 2017 Citation: Hua, L.Z. and G.F. Shao. 2017. The progress of operational forest fire monitoring with infrared remote sensing. Journal of Forestry Research 28(2): 215229.
• Type: Journal Articles Status: Accepted Year Published: 2017 Citation: Shao, G., G.F. Shao, J. Gallion, M.R. Saunders, J.R. Frankenberger, S.L. Fei. 2017. Improving lidar-based forest aboveground biomass estimation with the regard to site productivity in temperate hardwood forests. Remote Sensing of Environment (accepted).

Progress 10/01/15 to 09/30/16

Outputs
Target Audience:Forest landowers, forestry professionals, forest researchers Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?We provided informal support to forestry professionals in Indiana for using various data sources of remote sensing and geographic information systems. How have the results been disseminated to communities of interest?Three book chapters on various remote sensing techniques were published for foresters and natural resource managers who have limited education in remote sensing. Three presentations in the InVEST modeling were made in professional society conferences. What do you plan to do during the next reporting period to accomplish the goals?We will publish Lidar remote sensing research outcome. We will improve the analysis of very-high-resolution aerial photography. We will report the InVEST modeling output to foresters and natural resource managers in Indiana. We include timber values into ecosystem service assessment with InVEST simulations.

Impacts
What was accomplished under these goals? We compared visual interpretation of very-high-resolution aerial photography with Lidar data analysis for quantifying stand density. The accuracy was assessed with ground measurements. We proposed and published semi-supervised class-specific feature selection methods to improve classification accuracy with very-high-resolution aerial photography. We parameterized an ecosystem service model known as InVEST for assessing the impacts of forest management on water quality in the White River Basin of Indiana.

Publications

• Type: Journal Articles Status: Published Year Published: 2016 Citation: Liao, J.F., L.N. Tang, G.F. Shao, X.D. Su, D.K. Chen, and T. Xu. 2016. Incorporation of extended neighborhood mechanisms and its impact on urban land-use cellular automata simulations. Environmental Modelling & Software 75: 163175.
• Type: Journal Articles Status: Published Year Published: 2016 Citation: Zhao, J. X. Liu, R.C. Dong, and G.F. Shao. 2016. Landsenses ecology and ecological planning toward sustainable development. International Journal of Sustainable Development & World Ecology 23(4): 319325.
• Type: Journal Articles Status: Published Year Published: 2016 Citation: Tang, L.N., L.M. Gui, G.F. Shao, L.Y. Wang, and L.Y. Shi. 2016. Practice and research progress on ecosystem conservation in trans-boundary areas. Chinese Geographical Science. 26(1): 109116.
• Type: Conference Papers and Presentations Status: Published Year Published: 2016 Citation: Wang, Y.Y., A. Shady, and G.F. Shao. 2016. Nutrient retention returns on investment in private forest conservation: the case of the Classified Forest and Wildlands Program in Indiana. Agricultural & Applied Economics Association (AAEA), Boston.
• Type: Conference Papers and Presentations Status: Published Year Published: 2016 Citation: Wang, Y.Y., A. Shady, and G.F. Shao. 2016. Maximizing Freshwater ecosystem service returns on investment in forest conservation. Society of American Foresters convention (SAF), Madison
• Type: Conference Papers and Presentations Status: Published Year Published: 2016 Citation: Wang, Y.Y., A. Shady, and G.F. Shao. 2016. Maximizing the Nutrient Retention Return on Investment in Forest Conservation: the case of the Classified Forest and Wildlands Program in Indiana. Natural Capital Symposium, Stanford University.
• Type: Journal Articles Status: Published Year Published: 2016 Citation: X. Chen, J. Qi, Y. Chen, L. Hua, and G. Shao. 2016. Adaptive semisupervised feature selection without graph construction for very-high-resolution remote sensing images. Journal of Applied Remote Sensing 10(2): 025002. doi: 10.1117/1.JRS.10.025002.
• Type: Journal Articles Status: Published Year Published: 2016 Citation: X. Chen, G. Zhou, H. Qi, G. Shao, and Y. Gu. 2016. Semi-supervised class-specific feature selection for VHR remote sensing images. Remote Sensing Letters 7(6): 601610.
• Type: Book Chapters Status: Published Year Published: 2016 Citation: Shao, G.F. 2016. Satellite Data. In: Wiley StatsRef: Statistics Reference Online, Online ISBN: 9781118445112. DOI: 10.1002/9781118445112.
• Type: Book Chapters Status: Published Year Published: 2016 Citation: Shao, G.F. and L.N. Tang. 2016. Remote Sensing. In: Wiley StatsRef: Statistics Reference Online, Online ISBN: 9781118445112. DOI: 10.1002/9781118445112.
• Type: Book Chapters Status: Published Year Published: 2016 Citation: Shao, G.F. 2016. Optical remote sensing. In: International Encyclopedia of Geography: People, the Earth, Environment, and Technology. D. Richardson (ed.). Wiley & Sons, Inc., P23902395.

Progress 10/20/14 to 09/30/15

Outputs
Target Audience:Forestry professionals and researchers in public and private forestry organizations. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?We have worked closely with forestry professionals in Indiana for measuring forest canopies, tree heights and stand densities with Lidar methods. We have compared the Lidar approaches and ground measurements by working with the forestry professionals. At the same time, the forestry professionals improved their understanding and expectation about the concepts and methods of Lidar remote sensing technology. How have the results been disseminated to communities of interest?Our review article of "Drone remote sensing for forestry research and practices" was published for forestry professionals who have limited technical knowledge about unmanned aerial systems. What do you plan to do during the next reporting period to accomplish the goals?We will further test Lidar remote sensing methods for hardwood forest measurements. We will conduct experiments to introduce ground-based stereo imaging methods into forest inventory. We will also try to perform automated analyses of image data collected with networked cameras in real time.

Impacts
What was accomplished under these goals? We conducted a variety of data analysis experiments with data sources that are readily available for the general public. We found that the Landsat data and high-resolution digital aerial photographs are useful to divide hardwood forests into finer types for the purpose of forest-type specific management. We also found that the statewide Lidar data in Indiana are useful for quantifying forest height and stand density with reasonable accuracy. We improved classification methods associated with image objects for county-scale mapping in Indiana.

Publications

• Type: Journal Articles Status: Published Year Published: 2014 Citation: Li, X.X. and G.F. Shao. 2014. A county-scale object-based land-cover mapping in U.S. Midwest region with high resolution aerial photography. Remote Sensing 6(11): 11372-11390.
• Type: Journal Articles Status: Published Year Published: 2014 Citation: Shao, G., B.P. Pauli, G.S. Haulton, P.A. Zollner, and G.F. Shao. 2014. Mapping hardwood forests through a two-stage unsupervised classification by integrating Landsat Thematic Mapper and forest inventory data. Journal of Applied Remote Sensing 8(1): 083546 doi: 10.1117/1.JRS.8.083546.
• Type: Journal Articles Status: Published Year Published: 2015 Citation: Pope, D.B., J. Harbora, G. Shao, L. Zanottie and G. Burniske. 2015. Deforestation of montane cloud forest in the Central Highlands of Guatemala: contributing factors and implications for sustainability in Qeqchi communities. International Journal of Sustainable Development & World Ecology 22(3): 201212.
• Type: Journal Articles Status: Published Year Published: 2015 Citation: Pauli, B.P., P.A. Zollner, G.S. Haulton, G. Shao, and G. Shao. 2015. The simulated effects of timber harvest on suitable habitat for Indiana and northern long-eared bats. Ecosphere 6:art58
• Type: Journal Articles Status: Published Year Published: 2015 Citation: Tang, L.N. and G.F. Shao. 2015. Drone remote sensing for forestry research and practices: a review. Journal of Forestry Research. 26(4):791797.