Split100

The transition potential calculation variant Split100 differs from Mnl100 by the replacement of the Multi Nomimal Logit transformation of S_ij and N_ijt respectively by a binomial logit transformation.

SuitabilityLogit_ij is defined by Rules/SuitabilityLogitExpr as the logit of v_ij.

And furthermore

DynamicNeighbour_ijt here is defined by Rules/DynamicNeighbourLogitExpr as the logit of N_ijt, the linear combination of neighbourhood data.

For the rest, all definitions are equal to those for Mnl100.

Split100 most closely resembles the calculation of transition potentials in DynaClue, especially when Neighbourhood Enrichment would be reactivated. The definitions of Mnl100 and Linear100 more closely reproduce what has been estimated by the logistic regressions on suitability factors and neighbourhood enrichment or potential.

Thus, The resulting Transition Potential TP_ijt for allocatable land unit i, allocatable land use type j and time step t is defined as

• if not Allow_ijt then − 5[EUR/m²]
• if Allow_ijt then (1−W_j) * S_ij^L + W_j * N_ijt^L with W_j is the j-th entry of the scenario parameter vector Neighbourhood/Weight.

S_ij^L is defined by Rules/SuitabilityLogitExpr as $exp(v_{ij}) \over {1 + \exp(v_{ij})}$

v_ij is defined by Rules/SuitabilityExpExpr as min_elem(MakeDefined(SuitabilityData_ij, -1000000.0), 80.0) which means that undefined SuitabilityData is replaced by a very negative value and very positive values are capped to a maximum of 80 to prevent numeric overflows in the exponential function.

SuitabilityData_ij is defined by alloc1.txt as a linear combination of factor map data.

N_ijt^L is the combined neighbourhood value, which is defined by Rules/DynamicNeighbourLogitExpr as $\exp(N_{ijt}) \over {1+\exp(N_{ikt})}$

N_ijt is defined by Rules/DynamicNeighbourExpExpr as min_elem(Alloc2/ConstWeight_j + DynamicNeighbourData_ijt, 80.0) which means that very positive values are capped to a maximum of 80 to prevent numeric overflows in the exponent function.