The urban sprawl dynamics: does a neural network understand the spatial logic better than a cellular automata?



consistent with the SOM behavioural rules, shown in the codebooks.

To conclude the SANNs process seems able to capture and predict a sound spatial logic
for future trends in urbanisation and to drive suitable territorial policies towards the
facilitation or the inhibition of the “organic” processes of the considered urban system.

9. Concluding remarks

The experiments presented allow to accept NN as powerful tool for investigation in
urban dynamics. The original aim of the paper was to compare NN performances with
the well known CA forecasting capabilities; therefore information provided is that of a
prototypical CA, which is limited to the local and neighbouring land use conditions.

But the big advantage offered by NN consists in investigating the connections in
whatever other data: density, urban morphology, spatial relationships with central
functions, planning constraints, political view of the local authority and so on. Further
research with enlarged Data Base would produce an important improvement on location
theory and on territorial morphology dynamics studies.

The SOM run was able to show significantly different dynamic behaviours and to
clearly distinguish the spatial location pattern of the urban functions considered:
compact and urban the commercial activities, compact and peripheral the industry, more
scattered and invasive of natural resources the residence, particularly in the last few
years.

In each of the models produced, the codebook points out the degree of relevance of each
variable in explaining the considered behaviour.

The SANN has produced a suitable scenario of the future urbanisation, even if with the
assumption of stability of the transition rules and with only two temporal thresholds
supplied for training. Again, NN should be able to learn more complex dynamics if
provided with larger temporal series of data, unavailable at the moment.

The overview depicted gives an idea of the level of “urbanisation risk” for each
agricultural cell and the level of infilling in urbanised areas, in centres and along the
roads. It seems quite a probable scenario which could steer the territorial policies
towards a sustainable development approach.

16



More intriguing information

1. Neural Network Modelling of Constrained Spatial Interaction Flows
2. Subduing High Inflation in Romania. How to Better Monetary and Exchange Rate Mechanisms?
3. The name is absent
4. The name is absent
5. Restructuring of industrial economies in countries in transition: Experience of Ukraine
6. A Pure Test for the Elasticity of Yield Spreads
7. The name is absent
8. Trade Openness and Volatility
9. A Theoretical Growth Model for Ireland
10. Publication of Foreign Exchange Statistics by the Central Bank of Chile
11. Olfactory Neuroblastoma: Diagnostic Difficulty
12. The name is absent
13. The name is absent
14. LAND-USE EVALUATION OF KOCAELI UNIVERSITY MAIN CAMPUS AREA
15. Life is an Adventure! An agent-based reconciliation of narrative and scientific worldviews
16. Importing Feminist Criticism
17. The name is absent
18. Stillbirth in a Tertiary Care Referral Hospital in North Bengal - A Review of Causes, Risk Factors and Prevention Strategies
19. The name is absent
20. The name is absent