TY - JOUR
T1 - Comparison of spatial modelling frameworks for the identification of future afforestation in New Zealand
AU - West, T.A.P.
AU - Monge, J.J.
AU - Dowling, L.J.
AU - Wakelin, S.J.
AU - Yao, R.T.
AU - Dunningham, A.G.
AU - Payn, T.
PY - 2020/6/1
Y1 - 2020/6/1
N2 - © 2020 The AuthorsIn 2018, New Zealand announced an ambitious effort to plant one billion trees by 2028 as part of its climate change mitigation plan and Paris Agreement targets for 2030 and 2050. Afforestation will be promoted through a series of incentives, including payments for carbon sequestration linked to New Zealand's Emission Trading Scheme. Given that afforestation will be pursued voluntarily, the identification of future afforestation areas is paramount to assist policy formulation and avoid unintended land-use outcomes. We identified spatially-explicit drivers of forest gain and the locations most likely to experience afforestation in the country using two distinct spatial modelling frameworks: logistic regressions and artificial neural networks (ANN). Five of the eight most significant drivers of forest gain that have influenced past afforestation patterns according to our logistic regressions (i.e., erosion potential, distance from exotic forests, woody grasslands, grassland productivity, and slope) were also ranked among the ten most influential drivers by the ANN model. These results indicate an overall agreement between the two modelling approaches, despite substantial methodological differences. In the absence of changes in current incentives for afforestation in the country, simulations suggest that the largest afforestation areas would be in the northeast region of the North Island. These findings can assist policymakers and landscape planners to achieve desirable land-use objectives for New Zealand.
AB - © 2020 The AuthorsIn 2018, New Zealand announced an ambitious effort to plant one billion trees by 2028 as part of its climate change mitigation plan and Paris Agreement targets for 2030 and 2050. Afforestation will be promoted through a series of incentives, including payments for carbon sequestration linked to New Zealand's Emission Trading Scheme. Given that afforestation will be pursued voluntarily, the identification of future afforestation areas is paramount to assist policy formulation and avoid unintended land-use outcomes. We identified spatially-explicit drivers of forest gain and the locations most likely to experience afforestation in the country using two distinct spatial modelling frameworks: logistic regressions and artificial neural networks (ANN). Five of the eight most significant drivers of forest gain that have influenced past afforestation patterns according to our logistic regressions (i.e., erosion potential, distance from exotic forests, woody grasslands, grassland productivity, and slope) were also ranked among the ten most influential drivers by the ANN model. These results indicate an overall agreement between the two modelling approaches, despite substantial methodological differences. In the absence of changes in current incentives for afforestation in the country, simulations suggest that the largest afforestation areas would be in the northeast region of the North Island. These findings can assist policymakers and landscape planners to achieve desirable land-use objectives for New Zealand.
U2 - 10.1016/j.landurbplan.2020.103780
DO - 10.1016/j.landurbplan.2020.103780
M3 - Article
SN - 0169-2046
VL - 198
JO - Landscape and Urban Planning
JF - Landscape and Urban Planning
M1 - 103780
ER -