Add locked out constraint

Add constraint to a restoration problem (restopt_problem()) object to specify that certain planning units cannot be selected for restoration activities.

Usage

add_locked_out_constraint(problem, data, touches = FALSE)

Arguments

problem

restopt_problem() Restoration problem object.

data

terra::rast() or terra::vect() Either a raster object containing binary values hat indicate which planning units cannot be selected for any restoration (i.e., cells with a value equal one are locked out from the solution), or a vector object whose features correspond to the locked out areas. See the function add_available_areas_constraint() to get a locked out constraint from allowed restoration areas.

touches

logical If the locked out data is a vector, define wether the rasterization must include all pixels touching the polygons. (see terra::rasterize()). Useless if the data is raster data.

Value

An updated restoration problem (restopt_problem()) object.

Details

Locked out constraints can be used to incorporate a wide range of criteria into restoration planning problems. They can be used to account for existing land-use practices, feasibility of restoration activities, and stakeholder preferences. For example, locked out constraints can be used to ensure that urban areas are not selected for restoration. Additionally, if restoration activities can only be implemented depending on certain conditions -- such as places where landscape slope is not too steep -- then locked out constraints could be used to ensure restoration activities are not prioritized for places where they could not be implemented. Furthermore, if stakeholders require solutions that do not prioritize particular places for restoration, then locked out constraints can also be used to achieve this. See add_available_areas_constraint(), which achieve the same as the locked out constraint by defining areas that ARE available for restoration.

See also

Other constraints: add_available_areas_constraint(), add_compactness_constraint(), add_components_constraint(), add_connected_constraint(), add_min_iic_constraint(), add_min_mesh_constraint(), add_restorable_constraint()

Examples

# \donttest{
# load data
habitat_data <- rast(
  system.file("extdata", "habitat_hi_res.tif", package = "restoptr")
)

locked_out_data <- rast(
 system.file("extdata", "locked_out.tif", package = "restoptr")
)

# plot data
plot(rast(list(habitat_data, locked_out_data)), nc = 2)


# create problem with locked out constraints
p <- restopt_problem(
    existing_habitat = habitat_data,
    aggregation_factor = 16,
    habitat_threshold = 0.7
  ) %>%
  set_max_iic_objective() %>%
  add_restorable_constraint(
    min_restore = 5,
    max_restore = 5,
  ) %>%
  add_locked_out_constraint(data = locked_out_data) %>%
  add_settings(time_limit = 1)

# print problem
print(p)
#> ----------------------------------------------------------------- 
#>                          Restopt                          
#> ----------------------------------------------------------------- 
#> original habitat:     habitat_hi_res.tif 
#> aggregation factor:   16 
#> habitat threshold:    0.7 
#> existing habitat:     in memory 
#> restorable habitat:   in memory 
#> ----------------------------------------------------------------- 
#> objective:            Maximize integral index of connectivity 
#> ----------------------------------------------------------------- 
#> constraints:          
#>   -  restorable (min_restore = 5, max_restore = 5, min_proportion = 1, unit = ha) 
#>   -  locked out (data = in memory) 
#> ----------------------------------------------------------------- 
#> settings: 
#>   - precision = 4
#>   - time_limit = 1
#>   - nb_solutions = 1
#>   - optimality_gap = 0
#>   - solution_name_prefix = Solution  
#> ----------------------------------------------------------------- 

# solve problem
s <- solve(p)
#> Note: The current solution is the best that the solver could find within the time limit. However, the solver had not enough to prove whether it is optimal or not. Consider increasing the time limit if you need a better solution (solving time = 1.03 s)

# plot solution
plot(s)

# }