Results
This section presents the results on a year by year basis, and the general characteristics are outlined in the last part.
1999
Our modeling of suitable habitats in
1999 is not particularly accurate. Our model predicts an area significantly
smaller than the area observed to be killed by the MPB. The general pattern,
with only valley locations being susceptible, is to a great extend satisfied,
but particularly the northwestern expansion of the beetle is poorly predicted.
Investigating the effect of the different parameters it turns out that in 1999
the location of susceptible host trees is the most limiting factor, although
the only one of our parameters actually modeling the major part of the observed
area is the mean maximum August temperature above 18.3°C.
2000
The model predicting the suitable
habitats of the MPB in 2000 is not accurate, as it was the case with the 1999
model. Particularly the cluster in the northwestern part of the observed area
killed by the beetle is poorly
predicted but also the area in between the two major valleys, our model fails
to predict the infestation by the beetle. This area is only partly killed (low
ratio of killed pine trees) by the beetle, so this area is not likely to be
caught by our conservative model. Investigating the parameters one by one, it
turns out that none of them includes these two areas, but particularly the mean
maximum August temperature is a limiting factor.
2001
For 2001 the prediction our model
produces is not accurate. The area in the northwestern corner of the infested
area is poorly predicted, although the model predicts the area between the
valleys better than the previous years. In 2001 the area with more than 833
degree days above 5°C and the presence of susceptible hosts are the most
limiting parameters in the area of interest.
2002
The
main pattern in the 2002 prediction is that the whole Western part of
the map is cut out because the spring preciptation was higher than the
long term average in this area, which makes the potential host trees
more resistant.
The area just west of the central valley is better
predicted than the previous years. The most interesting prediction on
this map is that the central valley seems to have very good conditions
for the MPB.
2020 Scenarios
In the two future scenarios the
prediction are significantly better, particularly along the central valley and
southwest of here. The A1FI scenario
is the IPCC scenario with the highest rise in temperatures. B1 is the scenario
with the lowest rise in temperature. This is visible, particularly in the
southern part of the map, but the difference is not significant and the general
pattern is the same. The Hectares BC model predicts a far larger area to be
infested than our model. The most limiting factor in our model in the
projection is the susceptible host trees.
Results - general trend
In
general, our predictions are too conservative. It seems like we have made
our parameters too restrictive, thus reducing the predicted area that could
possibly be affected.
Furthermore, it seems like the level of detail in data and background knowledge was not sufficient to make an accurate prediction of which areas were actually suitable for the MPB.
On the other hand, we find areas, especially in the eastern and northern part that could possibly be affected, but have not been affected yet. Given that our model does not take into account the areas previously affected by the beetle, we can not predict if it would be physically possible for the beetle to actually get to the area with the data on hand. Anyhow, the predictions of suitable areas not yet affected means that especially these areas will have to be protected in the future, since they are home to trees of the right age and type as well as favorable climate conditions.
Furthermore, it seems like the level of detail in data and background knowledge was not sufficient to make an accurate prediction of which areas were actually suitable for the MPB.
On the other hand, we find areas, especially in the eastern and northern part that could possibly be affected, but have not been affected yet. Given that our model does not take into account the areas previously affected by the beetle, we can not predict if it would be physically possible for the beetle to actually get to the area with the data on hand. Anyhow, the predictions of suitable areas not yet affected means that especially these areas will have to be protected in the future, since they are home to trees of the right age and type as well as favorable climate conditions.