Decision support tools for forestry using open source software
Stephen Bathgate, Andy Kennedy, Duncay Ray, Louise Sing
and Phil Taylor
03 November 2016
1. A little bit of history on selecting tree species
2. Decision Support Systems for Forestry
3. Ecological Site Classification (ESC)
4. ESC4 Decision Support System
5. Conclusions and Questions
FC Bulletin 30 (1957)
• Climatic maps covering
temperature, wind speed
and moisture (rainfall
• Site prescriptions
according to bioclimatic
Some of the issues
• During 60s and 70s the principle of matching species to
site was turned around, and some sites were matched
to species with ploughing and fertiliser.
• The advent of Sustainable Forest Management, coupled
with increasing costs, led to a reduction in fertiliser use
in the 90s. Objectives changed away from production.
• Skills and staff were lost.
• Land available for afforestation was often degraded,
poor choices about species potential arose.
• The range of site types, tree species and potential
future climates a forester might need to analyse
required digital data and computer processing.
ESC Bulletin 124
• Following research and
development in the 1990s
national scale guidance was
published in 2001.
• Was complemented by a
computer based decision
support system, field
survey pack and training
• ArcView extension.
• Now embedded in many
aspects of GB Forestry.
Decision Support Systems
1.Tools/models that simplify a complex problem so that
users can evaluate the relative merits of different
2. Intended to complement, not replace,
3. Rubbish in, rubbish out.
4. Uncertainty/error/assumptions in models/data.
What does a DSS do?
• In this context
1. query the properties of a site (climate/soil)
2. analyse those properties with a scientific model
3. return the results to the user.
• The climatic data is usually in raster format
• Some models have spatial interactions (e.g. insects
move according to their preferences for certain
• Private sector forestry
• Public sector forestry (operational/policy)
With such a broad spectrum of users ESC tools
have to be relatively agnostic to technology –
web based tools and paper based systems.
• When disseminating data/services:
- avoid imposing technical requirements (and
costs) upon end users (e.g. Windows only
- ensure interoperability of geospatial data and
services across platforms.
• In ESC4 we use web map services (WMS),
GeoJSON, CSV and other open data formats.
Open Source GIS
R with raster plugins to
generate national scale
maps for species potential
QGIS for visualisation,
with qgis2threejs, sld4raster,
Geotools provides set of libraries to
work with spatial data
(geotiffs/geojson) in Java
OpenLayers3 for integration of
spatial operations on the client side
Geoserver to provide web map
services to desktop and web based
Could this simplify the construction
of this type of tool?
PostGIS to store and manage
spatial and aspatial data.
Design/Open Source Tools
Desktop GIS (WMS)
• Stand (Site) scale (e.g. <15ha), point and click
• Strategic scale, WMS maps, precalculated
results from R scripts etc.
• Operational scale with specific site data (e.g.
soil survey)? Need to batch process large
number of individual sites, e.g. to assess a whole
forest block/design plan area.
- Upload a geojson file based on a soil survey,
process and send back to user so they can
integrate with other design plan data.
ESC Site Analysis
• To create an individual site analysis.
- query 6-20 raster layers for point of interest.
• Execute models based on site data.
• Send results to user (web browser, file download depending on
• In use cases involving multiple sites this has to be repeated
many times (including zonal stats/centroid derivation of mean
attribute values for polygons).
• PostGIS raster queries slower than Geotools, so far..
• Need to look at alternate approaches, e.g. rather than rasters
create bioclimatic zones as vector data with attributes (loss of
• Other ideas?
• All very much a work in progress.
• Refining user interface and working with user
• Adding other decision tools.
• Optimisation of operational applications.
• Looking at options for open source survey data
collection e.g. QGIS extension.
Challenges (some historical)
• Forestry Commission technical strategy based on Oracle technology
and proprietary GIS.
• Learning curve and range of open source GIS tools e.g. GRASS,
Gvsig, QGIS + many others, Mapserver/Geoserver. What to choose?
• Licensing constraints around certain tools and data.
• Developing capacity/capability to support an open source
• Range of user communities and their expectations.
• Resistance to open source(or change). Security constraints, one tool
per business requirement, concerns over integration and support.
The creation of parallel open source environments was
necessary to workaround certain barriers and prove the
merits of open source.
• Wind exposure index.
• Based upon
effects and aspect.
• FR conducted tatter
flag surveys to build
Decision Support Systems
ESC is part of a decision making process,
it is NOT the decision.
This principle should apply across all
decision support systems.