Spatial Dynamical Modelling with TerraME (lectures 3 – 4) Gilberto Câmara Land-Use modelling example Vale do Anari (Rondonia.mdb database) Small-scale government planned rural settlement in Vale do Anari (RO), established in 1982 and land parcels sized around 50 ha TYPOLOGY OF LAND CHANGE ACTORS IN VALE DO ANARI REGION Land use patterns Spatial distribution Clearing size Variable Actors Main land use Linear (LIN) Roadside Irregular (IRR) Near main Small settlements (< 50 ha) and main roads Regular (REG) Near main Medium and Midsized Cattle settlements large and large ranching and main (> 50 ha) farms roads irregular Description Small households Subsistence Settlement parcels less than agriculture 50 ha. Deforestation uses linear patterns following government planning. Small farmers Cattle ranching and subsistence agriculture linear Settlement parcels less than 50 ha. Irregular clearings near roads following settlement parcels. Patterns produced by land concentration. regular Vale do Anari – 1985 source: Escada (2006) Pattern type Geometrical Irregular Linear Vale do Anari – 1985 - 1988 source: Escada (2006) Pattern type Geometrical Irregular Linear Vale do Anari – 1988 - 1991 source: Escada (2006) Pattern type Geometrical Irregular Linear Vale do Anari – 1991 - 1994 source: Escada (2006) Pattern type Geometrical Irregular Linear Vale do Anari – 1994 - 1997 source: Escada (2006) Pattern type Geometrical Irregular Linear Vale do Anari – 1997 - 2000 source: Escada (2006) Pattern type Geometrical Irregular Linear Vale do Anari – 1985 - 2000 source: Escada (2006) Pattern type Geometrical Irregular Linear Can you grow it? Anari -1985 Anari -1995 Anari -2000 1. Simple diffusive model: number of deforested neighbours 2. Diffusive model: : number of deforested neighbours + additional factors 3. Statistical model without neighbours 4. Statistical model with neighbours Can you grow it? Anari -1985 Anari -1995 Anari -2000 -- CONSTANTS (MODEL PARAMETERS) CELL_AREA = 0.25; -- 500 x 500 meters or 0.25 km2 DEMAND= 500; -- 100 km2 Vale do Anari (1985) Vale do Anari (1995) Vale do Anari (2000) Pattern type Geometrical Irregular Linear General outline of land change models Calculate potential for change Order cells according to potential Demand for change Allocate change on cells Spatial Iterator in TerraME it = SpatialIterator { csQ, function(cell) return cell.champion == “Brazil”; end } Ordering cells in TerraME Demand for change Calculate potential for change Order cells according to potential Allocate change on cells -- Step 2: Order cells according to potential it = SpatialIterator { csQ, function(cell) return cell.pot > 0; end, function (c1,c2) return c1.pot > c2.pot; end } -- Step 3: allocate changes to most suitable cells count = 0; for i, cell in pairs( it.cells ) do if (count < num_cells_ch) and (count < it.count) then cell.cover_ = "deforested"; count = count + 1; end Exercise 1 – Simple diffusive model Expansion based on neighbourhood potential More deforested neigbours, more potential for change Exercise 2 – Modified diffusive model Expansion based on five factors: 1. Neighbourhood potential 2. Distance to main road (dist_rodovia_BR) 3. Distance to primary side roads (dist_ramal_princ) 4. Distance to secondary side roads (dist_ramal_sec) 5. Distance to urban centers (dist_urban) main road primary side road secondary side road Exercise 3 – Neighbourhood + regression Expansion based on two factors: 1. Neighbourhood potential (50%) 2. Linear regression (50%) poti= - 0.0012* dist_rodovia_BR - 0.06* dist_ramal_princ - 0.003* dist_ramal_sec (normalize to [0,1]) Simple Linear Regression 20 18 16 14 12 10 8 6 4 2 0 -0.5 R2= 0.43 0 forest 0.5 1 deforested 1.5 Exercise 4 – Spatial regression Expansion based on spatial regression (includes neighbourhoods) poti = 0.173*num_deforested_neigh -0.1 * math.log10 (cell.dist_rodovia_BR/1000) + 0.053*math.log10 (cell.dist_ramal_princ/1000) -0.157 * math.log10 (cell.dist_ramal_sec/1000) (normalize to [0,1]) Exercise 4 – Spatial Regression R2= 0.84