1. Geospatial Forest Inventory

We will employ k-NN imputation to generate tree-level forest inven-tory data (tree-list imputation) in a form that can be used to param-eterize two separate forest dynamics models: Climate-FVS (Stand-level) and Landis-II or Landis-pro (Landscape-level).

Objective

Preliminary Results - Tree-list Imputation

Enhancing Tools and Geospatial Data to Support Operational Forest Management and Regional Forest Planning in the Face of Climate Change

2. Modelling Forest Response (Cont.)

Forest managers have many potential adaptive management strate-gies at their disposal including.

Promoting resistance to change - Improve the ability of a forest to resist external stressors (e.g., fire fuels reduction, installation of fire breaks, removal of invasive species, among others).

Enhancing resilience to change - Improve the ability of a forest to return to a prior condition following disturbance (e.g., intensive planting, enhancing complexity such species diversity and struc-tural diversity).

Accommodating change - Assist transition to future conditions (e.g., assisted species migration, alter ecological or successional trajecto-ries, among others).

Potential Adaptive Management Strategies

3. Education

The educational plan is designed to meet the needs of multiple au-diences including forestry professionals (existing workforce train-ing) and students at a variety of levels (pre-college, undergraduate, and graduate education).

ObjectiveStudy Areas

Project Overview

A detailed understanding of how forest composition, structure, and function will be impacted by projected climate change and related adaptive forest management activities are particularly lacking at local scales, where on-the-ground management activities are imple-mented.

Climate sensitive forest dynamics models may prove to be effective tools for developing a more detailed understanding. However, to be applicable to both regional forest planning and operational forest management, modeling approaches must be capable of simulating forest dynamics across large spatial extents (required for regional planning) while maintaining a high-level of spatial detail (required for operational management). These data are difficult to generate with traditional remote sensing techniques.

Introduction

Approach - Tree-list Imputation

Example Application and Results

Effects of climate change on forest type in an example landscape in northeastern Minnesota using the LANDIS-II Biomass Reclassifica-tion extension. Results could be used to inform management deci-sion such as species assisted migration. Approach

1. Employ remote sensing data (airborne LiDAR and NASA satellite data and derived products) to generate detailed (tree-level), spatial-ly-explicit forest inventory data across large spatial extents.

2. Demonstrate how this data can be used in conjunction with climate-sensitive forest dynamics models to assess the potential im-pacts of changing climate, disturbance regimes, and adaptive forest management practices on future forest conditions at spatial scales relevant to forest management and planning.

Acknowledgements: This research is primarily funded by the NASA New Investigator Pro-gram (Grant number: NNX12AL53G S01). References: Millar, C.I., N.L. Stephenson, and S.L. Stephens. 2007. Climate Change and Forests of the Future: Managing in the Face of Uncer-tainty. Ecological Applications 17(8): 2145-2151.

The education plan is comprised of several linked research and edu-cation activities including: (i) meeting with land managers to dis-cuss practical silvicultural strategies for adapting to climate change (this will inform research tasks), and (ii) developing and delivering education modules to diverse audiences including forest manage-ment professionals, pre-college youth, community college students, as well as undergraduate and graduate students.

Approach - Educational Plan

3. Inform our science with management needs, share our results and decision-support tool with forest managers, and educate students at a variety of levels (pre-college, college, community college, under-graduate, and graduate students) about the impacts of climate on forest condition and function.

Since it is forest managers who will ultimately imple-ment climate considerations into their management strategies, there is a clear need to bridge the spatial disconnect between tradi-tional, regional-level remote sensing science and the scales at which forest man-agers operate.

Our primary objective is to develop a system to spatially parameter-ize and supply critical initial conditions for two separate climate-sensitive forest dynamics models across unique ecoregions (in terms of forest structure and composition) via an integration of sub-orbital LiDAR data with data and products derived from NASA remote sensing assets (e.g., MODIS, Landsat, GLAS canopy height).

2. Modelling Forest Response

r = 0.88RMSD = 4.76 m2 ha-1

r = 0.89RMSD = 13.95 m3

0 10 20 30 40

010

2030

40

Field Measured Basal Area (sq. m per ha.)

).ah rep m .qs( ae

A lasaB detup

mI

0 20 40 60 80 100 1200

2040

6080

100

120

Field Measured Volume (cu. m)

)m .uc( e

muloV detup

mI

The imputed tree-lists can be applied spatially to provide inputs to climate sensitive forest dynamics models. Each pixel in this dataset contains a tree-list (i.e., individual tree records).

ObjectiveAssess the potential impacts of changing climate, disturbance re-gimes, and forest management practices on future forest composi-tion, structure, and function, with a specific focus on assessing resil-ience to climate change.

ApproachClimate-FVS and Landis-II will be parameterized via the imputed tree-lists which in turn will be used to assess the efficacy of potential adaptive management strategies for enhancing forest resilience to climate change. Specifically we will run a separate simulation for various GCM and emission scenarios across a range of adaptive management strategies.

Study Landscapes across the PNW - Black areas represent LiDAR acquisition areas

The methodology leverages statistical similarity between remote sensing variables to predict field measured tree-lists to unsampled areas. This example leverage LiDAR data; however any spatial data related to forest structure could be employed as predictor variables.

Initial results demonstrate strong linear relationships between forest inventory metrics derived from the imputed tree-lists and indepen-dent validation data. Results are from the Malheur N.F. study land-scape in Oregon.

Collate Materials

Pre-CollegeModule

Research Activity

Education Activity

Manager Meetings Develop DecisionSupport Tool

Graduate Students

Undergraduate Students

Community College Student

Training Seminar National AdvancedSilviculture Module

Landsat

Landsat

LiDAR Canopy &

Topography

RF Spatial Imputation

Field Plot Database

Landscape-scaleSpatial Inventory

Landsat

GLAS

Ecoregion-scaleSpatial InventoryUSGS

Topography

Current Climate

RF Spatial Imputation

AggregatedLandscape-scaleSpatial Inventory

Future Climate,

Site Index,and Species

Viability Scores

Climate-FVS

Climate-FVSLANDIS-II

LANDIS-II

Management Options

Management Options

LANDIS Future Condition Maps

Future ConditionMaps

Spatial Aggregation

FIA Summaries

Established Processes

Novel ProcessesValidation or Uncertainty Assessment

Dependent Variable

Independent Variable

Model or Procedure

Products

Ecoregion-scaleLandscape-scale

Current Climate Future Climate

FVS Future Condition Maps

U

U

U

U

U

(adapted from Millar et al., 2007)

0

10

20

30

40

50 Ref. Plot 1

0

10

20

30

40

50 Trg. Plot 1

0

10

20

30

40

50 Ref. Plot 2

0

10

20

30

40

50 Ref. Plot 1 Trg. Plot 1

Unknown Inventory

Data

Tree-list 1

Tree-list 1

Tree-list 2

Reference Dataset

Target Dataset

Michael J. Falkowski1 (mjfalkow@mtu.edu), Andrew T. Hudak2, Nicholas L. Crookston2, Robert M. Scheller3, Matthew Duveneck3, Linda M. Nagel1, and Robert E. Froese1

1Michigan Technological University - School of Forest Resources and Environmental Science; 2US Forest Service - Rocky Mountain Research Station, 3Portland State University