Sensitivity analysis of a GIS model for the environmental impacts... forest residues

advertisement
Sensitivity analysis of a GIS model for the environmental impacts of the extraction of
forest residues
Euan Brierley1, Willie Towers2, Ian Truckell3, Ann Malcolm2, Willie Walker4, Tim Brewer3
1
2
3
4
Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire, MK45 4DT, UK.
MLURI, Craigiebuckler, Aberdeen, AB15 8QH, UK.
National Soil Resources Institute, Cranfield University, Silsoe, Bedfordshire, MK45 4DT, UK.
TRADA, Jupiter House, High Street, Tattenhall, Cheshire, CH3 9PX, UK
Abstract
Environmental limitations on the removal of forest residues for bioenergy are not
well documented for UK conditions. Additional information was obtained from the
forestry profession and timber harvesting skill-base and through creation of
geographic information system (GIS) model. The paper presented to the IUFRO
meeting on sustainable forestry in theory and practice considered sensitivity
analysis of this model. The sensitivity of the model was tested in terms of woodland
type, scalar resolution, and against national soil inventory point data. This analysis
assisted in recognising areas of weakness within the datasets on which the model
was developed, principally in addressing the potential loss of soil fertility.
Introduction
The use of forest residues may aid the UK government’s New & Renewable Energy policy,
which requires the generation of 10% of electricity from renewable sources by 2010. However,
there are potential environmental impacts associated with the removal of forest residues. For
the most part, impacts are considered to be negative in comparison with conventional stemwood harvesting. The axiomatic requirement that harvesting systems be designed to minimise
damage was formalised in the Helsinki Forestry Guidelines (Ministry of Agriculture and Forestry,
Finland, 1993) and reinforced in the UK through a number of subsequent statements and
guidelines, such as Forest and Water Guidelines (Forestry Commission, 1993) and the Forest
Landscape Design Guidelines (Forestry Commission, 1994). These philosophies have been
incorporated within the UK Forestry Standard and the UK Woodland Assurance Standard
(UKWAS Steering Group, 2000). Principally, planning of harvesting operations in the UK,
consider the following potential impacts:
• preservation of sustainable forest productivity;
• protection of soil structure and the soil resource;
• maintenance of water quality, through the control of sediment transport;
• maintenance of bio-diversity;
• carbon sequestration; and
• woodland amenity, access.
Methods
Seven case study areas, each of 40 000 ha were selected, on which GIS models were
developed to assess the environmental impacts of forest residue extraction. These embraced a
variety of soil types, in areas where forestry was a major land use. The following datasets,
based on 1:250 000 soil maps, and common to both Scotland and England and Wales, were
incorporated:
•
•
•
•
•
Soil fertility for tree growth. A three class map, using soil pH as a surrogate for nutritional
attributes such as base saturation and C:N ratio;
Critical loads of acidity. A five class map, ranging from ≤0.2 kmol H+ ha-1 year-1 to >2.0
kmol H+ ha-1 year-1;
Soil wetness class. A six class map incorporating soil moisture retention characteristics,
integrated with the inherent dryness/wetness of the climate, expressed as field capacity
days;
Leaching potential. A two tier classification combining the permeability of the underlying
lithology (three classes) with an appraisal of the leaching potential of the overlying soil;
Erosion risk. A seventeen class map displaying inherent geomorphological risk of soil
erosion assumed no vegetation and was abandoned in favour of a topographical
approach.
Scenarios, based on impact thresholds, predicted the proportion of forest areas from which
residues may become available. Only the physical/chemical impacts associated with residue
removal were modelled. Carbon sequestration and biodiversity impacts were excluded, along
with any other issues that may create the policy/management drivers behind decisions to
harvest forest residues.
Thresholds for each data layer were selected, representing an assessment of each individual
impact on the suitability of a site for residue extraction. In practice the terms highly, moderately
and marginally suitable may refer to the likely degree of impact (for example, acidity load) or to
differing periods of time in which site conditions permit residue extraction (such as soil wetness).
The term unsuitable is applied to those areas when typically there is no opportunity during the
year to traffic the soils without environmental damage or at a very limited number of sites which
have an exceptionally high acidification risk. Harvesting of residues which have been used as
brashmats and trafficked is not a realistic prospect because of soil contamination. Currently, the
model is insufficiently sophisticated to enable division between non-harvested brashmat
material and surplus harvestable residues.
Modelling relied heavily on the quality of the 1:250 000 soil map and the interpretations made
from it. More detailed soil maps at the 1:25 000 were available for some of the study areas and
a comparison of the outputs generated using higher resolution data were considered in order to
assess the confidence which might be placed on the coarser data. The sensitivity analysis was
relevant not only to this study but related pan-European work at a coarser scale. The sensitivity
of the models to woodland type was considered also for one case study area in the west of
Scotland, with forest cover containing 17% broadleaf areas. Another test of the models was
undertaken using national soil inventory point data for the soil attributes of fertility (pH), erosion
(slope) and compaction (wetness class) at the three case sites in England and one in Wales.
Results
The model output differentiated clearly between opportunities for residue harvesting between
upland forestry in the north and west of the UK and lowland forestry in the south. This reflects
the different biophysical conditions but, whilst cumulatively the distinction between upland and
lowland forestry is reasonably clear, in fact there was considerable variation in the way
cumulative impacts were derived.
The model was generally insensitive to scale but some difference in the outputs was generated
as a result of a higher resolution of soil fertility. The original analysis used data with a single
value attributed to each 1 km grid square. At 1:250 000, combinations of soils are often
represented as soil complexes, but at 1:25 000, fertility values were attributed individually to
areas differentiated by soil series. Hence, pH values in excess of 7 may be anomalous, for data
were related to areas defined by the soil series and data applied to the forest areas may in fact
be derived from neighbouring agricultural land of the same soil series. Discreet areas for each
soil series appeared as polygons which present a more realistic representation of the spatial
arrangement of soils. Hence soil fertility is one attribute where the use of the 1:25 000 scale
data would enhance the precision and accuracy of the output. However, the benefit remains
uncertain for whilst general relationships between soil attributes and soil fertility are known,
forest scientists agree that there are no consistent variables relating soil parameters and tree
growth. The adoption of a Forestry Commission dataset ‘soil fertility for tree growth’ constructed
around soil pH offered a degree of industry acceptance. We acknowledge, however, the
limitations this imposed and threshold values within the dataset were not as we would have
chosen. Thresholds around a pH 5-7 range is wide and the impact on soil fertility, following
residue removal, will differ between the soils at either end of this range.
The analysis of the significance of woodland type illustrated broad similarities between suitability
assessments for conifer and broadleaf woodlands.
Broadleaved woodlands had been
anticipated to predominate on better soils and offer better opportunities for residue extraction.
We attributed the finding to the coarseness of the underlying soils data. Most of the land
identified as unsuitable for residue harvesting, on the grounds of two or more impacts, was
within the two highest wetness classes and was deemed also to pose an unacceptable risk of
sediment transport to water courses. This combination of attributes is unlikely in reality and was
probably an artefact of the under-pinning data: the coarseness of the soils data did not delineate
the very steep gullies where broadleaved woodland is often found.
Testing point survey data to the modelled output provided only 65% agreement with the model
outputs but the comparisons were based on relatively few data (72 points). The paucity of soil
inventory data for forest areas had been the primary motivation in the adoption of derived
datasets for the model.
Discussion
A GIS model has been created which enabled the consideration of the environmental impacts f
extracting forest residues. The physicochemical components may be spatially modelled using
soil, climate and topographic data. We did not consider it possible to incorporate at this stage
consideration of carbon sequestration, nature conservation, or amenity and access. It would be
straightforward, if the model was to be developed further, to add layers of geographic
information related to protected habitats and landscape character. The limitations imposed by
the scale of the under-pinning data meant that the model was not appropriate for the creation of
a site management tool. Nonetheless, it was developed at a considerably finer resolution than a
similar project undertaken by Lindner et al. (2005) for the European Union. We advocate
national development of the model, to enable user-friendly investigation of the environmental
constraints on forest residue availability within a selected radius of any grid reference to the
benefit of policy makers and bio-energy generating sector. Inclusion within the model of yield
prediction, for different tree species, and therefore of the associated residues, would be
required. Some further research is needed to underpin the setting of threshold values for
classifying site suitability for residue extraction.
Reference
Forestry Commission (1993). Forest and Water Guidelines. 3rd edition, HMSO, London.
Forestry Commission (1994). Forest Landscape Design. Forestry Commission, Edinburgh.
Lindner, M, Meyer, J., Eggers, T. and Moiseyev, A. (2005) Assessing ecologically constrained
potentials for bio-energy from European forests. European Forest Institute, Joensuu, Finland.
(seen in draft form)
Ministry of Agriculture and Forestry, Finland (1993). Sound forestry – sustainable development
proceedings of the ministerial conference on the protection of forests in Europe 16-17 June
1993. Ministry of Agriculture and Forestry, Helsinki, Finland.
UKWAS Steering Group (2000). Certification standard for the UK Woodland Assurance
Scheme. Forestry Commission. Edinburgh.
Download