USING REMOTE SENSING TO MAP THE SPATIAL-TEMPORAL PATTERNS OF FOREST DISTURBANCE AND RECOVERY HISTORY

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: COMPLETE
• Funding Source: MCINTIRE-STENNIS
• Reporting Frequency: Annual
• Accession No.: 1017425
• Project Start Date: Oct 1, 2018
• Project End Date: Jun 22, 2020
• Program Code: [(N/A)]- (N/A)

Recipient Organization
UNIVERSITY OF CALIFORNIA, BERKELEY
(N/A)
BERKELEY,CA 94720

Performing Department
Ecosystem Sciences

Non Technical Summary
This project produces methods and software that can be used to track forest changes with remote sensing. It improves over current technology and increases the accuracy of what we see from space via satellites. This will allow natural resource managers at the State and Federal level to better manage forest resources to ensure the health and resilence of our forests.

Animal Health Component

90%

Research Effort Categories

Basic

(N/A)

Applied

90%

Developmental

10%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
123	0612	1070	50%
123	0612	2090	50%

Knowledge Area
123 - Management and Sustainability of Forest Resources;

Subject Of Investigation
0612 - Conifer forests of the West;

Field Of Science
2090 - Statistics, econometrics, and biometrics; 1070 - Ecology;

Keywords

forest monitoring from space

forest disturbance, harvest and regeneration

improved forest classification

Goals / Objectives
(1) The overall goal of the project is to develop new data analysis techniques for tracking spatial-temporal patterns of forest disturbance using remotely sensed data including random forest classification.(2) We will additionally explore using new sources of remotely sensed data that will provide additional information such as height available with Lidar. We expect that this will improve our ability to track forest change at the landscape level.

Project Methods
A key element in our research is to identify pixels (30m x 30m) in remotely sensed imagery that have been disturbed by management actions such as thinning, individual tree removals, clearcutting and afforestation. To do that we will utilize data from mixed conifer, oak and brushland areas at the approximately 4000-acre Blodgett Forest Research Station. Blodgett has forest inventory data going back several decades that can be used in this project. Because their forest inventory records all forms of management such as thinning and tracks individual tree sizes as well as tree status (alive, dead, diseased) it is possible to know the forest classes of the inventory plots which are well dispersed throughout the entire property. These data will be used in model development (calibration) and also in model testing (accuracy assessment). State-of-the art analysis/classification algorithms will be investigated including random forests - a robust and nonlinear, non-parametric multiple regression machine learning technique. The RF algorithm will be used to predict forest age as a function of spectral and biophysical variables using data available from multiple inventory years.We will use Landsat time-series stacks (LTSS) at the landscape scale along with the Vegetation Change Tracker (VCT), Landsat-based Detection of Trends in Disturbance and Recovery (LandTrendr), and Breaks For Additive Season and Trend Monitor (BFAST) to predict areas with forest activities such as thinning, clearcutting and afforestation. Additional remote sensing datasets using Lidar will be used as a supplement to Landsat satellite data. The Lidar data will add important height information that we expect will help us achieve higher, more accurate, classification (prediction) of forest vegetation classes. Lastly the Blodgett data will be used to compare known forest classes to the those predicted with the VCT, LandTrendr and BFAST models to compute model prediction accuracy.Outreach to the USGS, US Forest Service and the public will be made in conjunction with Cooperative Extension specialists on the mapping the spatial-temporal patterns of forest disturbance and recovery history in California and the Western United States. We will conduct seminars and training sessions in the best methods for how to utilize the random forest classification, Lidar data and customizations to VCT, LandTrendr and BFAST resulting from this research.

Progress 10/01/18 to 06/22/20

Outputs
Target Audience:The tartget audienceof this research included graduate studentsin a graduate seminar in spatial-temporalanalysis of natural resources. In addition, I mentored approximately 10 graduate students in my lab with this research and its results. We read and discussed the research papers generated by this research project and related projects. This also included field site visitation to demonstatrate how various forest sites are represented in the Vegetation Change Tracker (VCT). Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?I provided training and mentorship to approximately 10 graduate students to learn the methods we developed. We usedfield site visitation to demonstatrate how various forest sites are represented in the Vegetation Change Tracker (VCT) and to discuss how accurate VCT is and how it can be improved. 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? Nothing Reported

Impacts
What was accomplished under these goals? The goal of this project was to further develop evelop new data analysis techniques for tracking spatial-temporal patterns of forest disturbance using remotely sensed data including random forest classification. Based on the history of forest management, artificial and natural disturbances, we integrated remote sensing-based models (VCT, a spatial analysis model, random forest algorithm, as well as sample data) to map the resetting patterns of plantation stand ages resulting from forest disturbance and restoration events. In contrast to previous studies the innovation of this study resided in the utilization of forest age products derived from the VCT algorithm. In this study, we applied VCT-SA in the post-disturbance regrowing plantation area to estimate forest ages, with the prerequisite of VCT-SA being that certain initial information on forest management should be available. Besides, the number of observations could also affect the precision of the model. If we have observation with forest age less than 10 years in the model of VCT-SA for pine plantation, the model should perform in consistent with other two plantations. Further, it should be noted that the initial information required may be site specific or latitude-dependent, thus, prior to transferring our VCT-SA to other regions, we needed to confirm this initial information with local authorities to ensure the accuracy of the age mapping. Based on an assembled Landsat Time Series Stacks (LTSS) consisting of annual Landsat time-series observations and through the use of the VCT algorithm, spatial analysis model, and random forest regression algorithm, we mapped plantation stand ages for a VCT post-disturbance regrowing forest and VCT undisturbed forest in our study area. The results clearly revealed plantation stand age dynamics, both spatially and temporally, as well as the effects of forest management policy changes on forest age dynamics spanning 1987 to 2017. Since deforestation is contingent on the type, age, and management purposes of forest stands, our results demonstrated that mapping the distribution of plantation stand ages, in terms of forest type, can lay the foundation for improved and prudent forest management. Moreover, temporally consistent multi-temporal plantation stand age products may be considered as valuable data sources for multiple applications, including forest trajectory prediction and carbon sequestration dynamics. We believe that the combination of VCT, spatial analysis model, and random forest regression is robust for synergistically inferring historical plantation stand age distributions. This provides a critical new tool for assessing plantation stand management and carbon accounting under various climate change scenarios.

Publications

• Type: Journal Articles Status: Published Year Published: 2020 Citation: Diao, J.,Feng, T.,Li, M.,Zhu,Z.,Liu, J.,Biging, G.,Zheng, G.,Shen,W.,Wang,H., Wang, J. and Ji,B. 2020.Use of vegetation change tracker, spatial analysis, and random forest regression to assess the evolution of plantation stand age in Southeast China. Annals of Forest Science 77, Article Number:27 (2020)
• Type: Journal Articles Status: Published Year Published: 2019 Citation: Yang, B., Lee, D.K., Heo, H.K. and Biging, G. 2019. The effect of tree characteristics on rainfall interception in urban areas. Landscape and Ecological Engineering. 15, pages 289296(2019)
• Type: Journal Articles Status: Published Year Published: 2018 Citation: Kim, H.G., Lee, D.K.,Park C., Ahn, Y., Kil, S.H.,Sung S. and G.S. Biging. 2018. Estimating landslide susceptibility areas considering the uncertainty inherent in modeling methods. Stochastic Environmental Research and Risk Assessment. 32, 2987-3019.