SIB Tech Rept Template - Canadian Institute of Forestry

NESI develops River Characterization Tool to support efforts in the Far North
To date, very little baseline information has been collected in the Far North. The Far North landscape
is too vast and inaccessible for it to be financially viable to acquire enough field data to adequately
describe the condition of the entire region. Therefore, it is imperative to leverage existing tools and
methodologies that use remotely sensed data for characterizing the landscape. In order to develop
resources in an economically sustainable manner, current and accurate science and information
must be available to guide management decisions.
Northeast Science and Information (NESI) has developed a River Characterization tool as part of
the Far North Information and Knowledge Management initiative. The tool provides a user friendly
ArcGIS toolbar for scientists and resource managers to perform automated characterization of
the riverine landscape across various scales: watershed, river, and reach. This tool integrates and
aggregates various types of information into one characterization tool and draws from concepts
and methodologies of several existing MNR tools, helping improve efficiency in how to use existing
science and information.
As the proposed development activities, such as hydro power and mining, have the potential to
adversely affect the aquatic resources of the Far North, the river characterization tool will provide
a cost effective method to provide initial science and information and determine where more
intensive field data collection is required. This ultimately saves money by optimizing the use of
available digital data and reduces the time required to collect data in the field.
For more information on the river characterization tool, please contact Jennifer Potvin, 705-235-
1260, jennifer.potvin@ontario.ca.
NESI’s Terra Team collaborates with UQAT and CFS on the Regional Fire Regime
Knowledge of the regional natural disturbance regimes (stand replacing fire) is required to develop
a natural benchmark landscape pattern for Ontario’s Boreal forest management planning process.
Detailed fire history mapping is expensive and, for Ontario, is only available for the Lake Abitibi
Model Forest. Dave Etheridge, NESI’s Forest Landscape Specialist, and Gordon Kayahara, NESI’s Forest
Science Specialist, approached Yves Bergeron (Université du Québec en Abitibi-Témiscamingue) and
Sylvie Gauthier (Canadian Forest Service) with an idea to extrapolate their work on the fire mapping
of the Lake Abitibi Model Forest to the rest of Ontario’s Claybelt using regional forest resource
inventories. Dominic Cyr, who worked on his PhD under Professor Bergeron, came up with the novel
method of using Bayesian statistics. Classic statistical approaches like regression are normally used
and deal with average outcomes. For those “older” Trekkies among us, you will be familiar with the
more Bayesian approach of the original Spock who gives the probability of an individual outcome
given various levels of prior knowledge.
In this paper, two classes for fire history were used: time since fire (TSF) < 150 years and TSF ≥ 150
years. A Bayesian Belief Network was then constructed to place each FRI polygon into one these two
fire history classes based upon latitude and the FRI map attributes. The relative abundance of early
(jack pine and poplar) and late (balsam fir) successional tree species were the best indicators of the
two fire history classes. These variables were followed by photo-interpreted age, stocking, height,
and site class. The results suggests that more than 72.4% of the forests located within the Claybelt
originated ≥ 150 years from stand replacing fire. We now have a method that gives us an efficient
estimate of time since fire classes and the resulting amount of old forests for the entire claybelt, and
a process that can be extrapolated to the rest of Ecoregion 3E. We learned that average time since
fire is much longer than currently believed, further confirming obtained results through modelling
exercises used to develop the provincial Landscape Guide. In addition, the exposure to the Bayesian
approach holds promise as a technique for various other MNR tasks such as silviculture effectiveness
monitoring or developing succession rules.
For additional information on this collaboration, please contact Gordon Kayahara, 705-235-1237,
gordon.kayahara@ontario.ca or Dave Etheridge, 705-235-1230, dave.etheridge@ontario.ca