Hi Andia,
Good questions, here are my responses:
1. When you set a transition target the model targets that amount of area for each timestep independent of how much area is actually eligible. At the start of the simulation 20% of the 1000 acres are conifer ranging in age from 20 to 100 years old. The minimum age for harvest is 40 so initially there will be about 0.75 * 200 = 150 acres eligible for harvest. With a target of 20 acre, the initial probability of harvest will be 20/150 = 0.133. However, the amount of area eligible for harvest will always be changing so when you set a target – the total amount is constant but the probability is not. In the tutorial example, there is not enough old conifer to sustain a harvest target of 20 acres per year so at about year 8, the amount of area harvested drops dramatically.
2. I don’t fully understand your question, are you asking why not use a probability in the pathways for harvest rather than a value of 1? If you set a probability and harvest, the harvest target will override the probability you set. If you would like to see some “harvest” happen with a probability and an additional amount with a target, you need to define two transition types and pathways: for example a type called “probabilistic harvest” that has the probability you intend in the pathway and another called “targeted harvest” where you set a transition target for the annual area undergoing the transition.
3. If harvest is possible in multiple states, you would still see only 20 acres harvested when you set a target of 20 acres. The amount harvested in each state would depend on two things: the amount of area in each state and the relative probability of the pathway in each state. For example if set a harvest target of 20 ac and I had harvest possible in state A with an area of 100 ac eligible and a probability of 0.1 and in state B with and area of 100 ac eligible and a probability of 0.3 then I would expect to see 3 times as much harvest of state B than state A so 15 ac harvested in state A vs. 5 acres in state B. The amount of harvest that you expect to see in state 1 is going to be equal to P1 x A1 x (Target area)/[sumproduct across all cells (area and probability)] where P1 is the probability of your transition in state 1 and A1 is the area in state 1.
4. You do not need to include probabilities for the diagonals (i.e., the “no change” transitions) of the transition matrix. The software will automatically calculate the “no change” probabilities. Also, because STSMs allow for more than one transition per timestep of different types (for example harvest and fire might occur in the same year), it is OK if the sum of your probabilities exceeds 1.
