Forest Service BC
Forest Renewal BC

Lesson 4: Assessing Windthrow at the Landscape Level

Learning Objectives

Home Page
Learning Objectives

In this lesson, participants will:

  • know how aerial photographs, forest cover and topographic maps can be used within a GIS environment to produce landscape-level maps of windthrow risk.

This lesson contains 10 pages. It introduces the methods and results of a GIS-based process for building empirical models and hazard maps of windthrow risk. The major steps and information sources used are summarized below. As well, read the linked reference articles for additional background information.



Background Page

There are three basic approaches to windthrow risk assessment: observational, mechanical and empirical. The diagnostic framework used in the FS 712 Field Cards and introduced in Lesson 2 of this course extends the observational approach by adding calibration and feedback steps. This makes it semi-empirical and semi-quantitative.

Using the capabilities of GIS for assembling large datasets, empirical models of windthrow risk can be built using stand-level information contained in forest cover maps and topographic maps.

Once models are built, they can be entered into the GIS for production of landscape-level maps. The following screens describe the process and results of a study on northern Vancouver Island. Read the paper by Mitchell, Hailemariam and Kulis for background on this study.

The purpose of presenting this method in this course is to make participants aware of the potential of using GIS to build empirical models and maps, and to set the stage for a development planning exercise. Most forest companies are now using GIS for drafting, but few are using them for analysis. Even without building empirical models, maps of windthrow patterns are very useful for observational evaluation of windthrow risk, and can contribute to improved planning by identifying orientation and recurrence of damaging winds, and locations subject to recurring damage.

Procedural Summaries

Page 2 - Risk Models Procedure Summary

Summary of procedures used to build empirical windthrow risk models with stand-level data in a GIS environment (ArcView GIS).




Information assembly


Creation of sample units


Windthrow detection and mapping


Determination of topographic variables


Construction of segment database


Model fitting and testing


Page 3 - Risk Models Procedure – Step 1: Information assembly
  • Obtain GIS layers for ecosystem, stand, and logging history.
  • Obtain TRIM contours.
20 m Contours


Forest Cover

Page 4 - Risk Models Procedure – Step 2: Creation of sample units
  • Create 50 m long x 40 m deep cutblock boundary segments from logging history coverage using custom-built scripts.
  • Calculate edge exposure scores using UTM coordinates of segments and custom-built scripts.
Small-scale satellite imagery


arger-scale image showing logging history


Page 5 - Risk Models Procedure – Step 3: Windthrow detection and mapping
  • Identify and map potential edge windthrow on recent 1:15 000 colour aerial photos.
  • Transfer to map and digitize.
  • Verify windthrow status of segments in field on ground or from helicopter.
  • Rate damage severity and add to database.
Map Cutblock Edge


Verify Mapping

Page 6 - Risk Models Procedure – Step 45: Determination of topographic variables
  • Produce TIN.
  • Determine topographic variables aspect, elevation, and slope.
  • Calculate TOPEX-to-distance scores using custom-built scripts.
Topographic and Management Variables


Build DEM



Page 7 - Risk Models Procedure – Step 5: Construction of segment database
  • Overlay coverages with edge segments and extracted segment database.
  • Keep segments with forested boundaries for analysis.
Page 8 - Risk Models Procedure – Step 6: Model fitting and testing
  • Import database into statistical package (SAS).
  • Calculate percent of segment damaged, and create a set of response variables.
  • Fit logistic regression models using 60% of the edge segments.
  • Test predictions using the other 40% of the data.
Stocking Class Grid


Topex2000 Grid


Probability of > 10% damage to

Page 9 - ArcView GIS Procedural Summary

Summary of procedures used to build maps in ArcView GIS.




Obtain forest cover and TRIM data.


Edit database.


Create grid coverage (100 x 100 m cells) for forest cover variables.


Build 3D model based on TRIM 20 m contours for the whole study area including the areas represented by the adjacent maps.


Convert DEM coverage to grid cover using the 100 m cell size.


Select the area (map sheet).


Calculate Topex2000 score for each grid cell for selected distance = 2000 m for eight cardinal directions (see description below).


Use Map Calculator to input the regression equations. The outputs are probability grid covers with different value range for each map sheet. To unify the value range and intervals, set the lower boundary to 0.00, the upper boundary to 0.90, and the number of classes to 6.The management variables in the regression equations are held constant to represent south facing cutblock boundary segments located on the north side of large openings which have had eight years of exposure following harvesting.


Risk estimates are unreliable for stands less than 10 m tall, so stands in Height Class 1 are masked by overlay in grey colour.


Areas with no-data are shown in white.


Areas labelled as non-forest or NSR types are beige and brown.


Create the windthrow risk overlays: grid covers overlaid with polygons representing recently logged sites, non-forest, and no data.