Hello Narayan, check in Google "Andrew fall mountain pine beetle". This should help you for a start. Cordially, Frédéric.

I would say that the question needs refinement before it can be answered definitively. You will need to clarify if you are looking to account for past, current, or future pine beetle infestations, as well as the strategic decisions that you plan to model.

Approach for modelling future pine beetle attacks would would depend on the intensity of the attack, and level of stochasticity you wish to model.

For example, you could easily model the effect of future deterministic beetle outbreak as an action in Woodstock (or as a treatment in Patchworks... identical concept, different syntax). This would be very similar (in terms of effect on growing stock) to modelling a partial cut or clear cut harvesting treatment. Of course, this "beetle" treatment would have to be removed from the solution basis (i.e., not appear as decision variables in the Woodstock matrix, or be available to the scheduling heuristic in Patchworks), or you will end up "scheduling" beetle outbreaks as part of the optimal solution. Flagging a treatment as off-limits to the scheduler is trivial in Patchworks (you can specify this in the treatment definition in the model logic input file). Achieving precisely the same effect is possible in Woodstock, but is considerably more complicated. Once you have the beetle treatment defined in the model, you simply force simulation where and when you want, lock it, and proceed with your wood supply analysis. You would likely also have to add a special "salvage" treatment to (optionally) salvage some stands in later periods. 

Note that cost of salvaging beetle wood (for an industrial timber consumer) tends to be relatively high (compared to a regular stand), and value tends to be low. Consequently, the unit value of beetle wood is often negative, so be careful that you are not being to optimistic about simulating the salvage treatment, or you might end up with a wood supply strategy that depends on the allowable cut effect of "magic" salvage treatments without the real-world government policy in place to make sure it will actually happen. This would increase risk of future policy failure (i.e., unpredicted wood supply failure), which would defeat the entire purpose of wood supply planning.

Bogle and van Kooten (2012) describe a bilevel optimization problem that models strategic forest policy decisions for beetle-infected Crown land using a principal-agent approach. Their bilevel model includes principal decision variables relating to timber price, harvest level, and tenure type, for the specific case of liquidating large volumes of timber that has recently been damaged by the beetle. There is no way to implement the bilevel model from Bogle and van Kooten in Woodstock or Patchworks.

A stochastic simulation can be modelled in Woodstock or Patchworks by randomly simulating certain levels of the beetle treatment described above, combined with rolling-horizon replanning simulation (with or without mid-period recourse option, as you wish).

The cost of implementing this stochastic rolling-horizon version of this model can blow up very fast. First of all, you need some some of automated mechanism to simulate the wood supply plan, simulate random application of the beetle treatment, update the model state, (optionally) simulate mid-period recourse, grow the forest for one period, repeat for as many cycles as required, and automatically compile the results into usable output. Neither Woodstock nor Patchworks platforms supports these functions directly.

Of the two platforms, it is far easier to implement this in Patchworks than in Woodstock, because Patchworks features a powerful scripting environment that makes extending functionality in this way very straightforward. I have implemented such a scripting module for Patchworks in the past (very similar problem, but for blow-down in hardwood forests). Patchworks also allows direct control over the spatial pattern of beetle (blowdown, fire, whatever) outbreak.

Regardless of the software environment you choose, the above-described stochastic rolling-horizon simulation approach will require a large number of replicates, making the automation all the more essential (otherwise, the cost of doing this manually is prohibitive). The minimum number of replicates required to achieve a target precision at a chosen confidence level will be a function of the variability of your random beetle process.

Also, note that both Patchworks and Woodstock are fundamentally deterministic in their model structures. You can only use these models to estimate the a posteriori effect of a random beetle attack on your optimal solution. It is not possible to explicitly account for the beetle attacks in your objective function, or both models with end up "scheduling around" the "random" events, which would defeat the purpose of the model.

You could also model future random beetle attacks by a priori considering them in the objective function. You could do this using a stochastic programming approach, or a robust programming approach (depending on your objective of the analysis, one or the other will be better). Forget about Woodstock and Patchworks (or any other commercial wood supply modelling platform, for that matter). You will require a custom model. If you want to go this route but are not certain how to proceed, perhaps I could help. You know where to find me.

Finally, the SilviLab modelling platform, which is developed by the FORAC research consortium, was recently extended to automate the iterative plan-disturb-grow rolling-horizon workflow described above, for the case of forest fires. You could probably re-use this technology as-is to model impact of pine beetle outbreaks on wood supply plan feasibility. You would probably only have to select different input parameters to capture the difference between fire (current use) and beetle. I cannot say for sure if this would work without more information about your problem and your objectives. 

Good luck!

Article Why mountain pine beetle exacerbates a principal-agent relat...

Thank you Greg for the time you took to explain and reply my question. A friend in need is a friend indeed! Thank you Frederic for the suggestion. Two papers suggested by you and Greg provided me the basis to start my model. Thank you very much.

While not specifically related to insect problems, I published a paper several years ago that addressed how other disturbances (fires) might be accounted for in a forest planning effort.

Bettinger, P. 2010. An overview of methods for incorporating wildfires into forest planning models. Mathematical and Computational Forestry & Natural-Resource Sciences. 2(1): 43-52.