YAFFO RESIDENTIAL DYNAMICS IN TERMS OF GAS


Yaffo is a part of Tel-Aviv. It is a residential area of 30.000 people, populated by Jews, Arab Muslems and Arab Christians. In our project (Benenson, Omer, Hatna, 2002) we based a model on data from 1955 and up to nowadays, and were interested in simulating Yaffo residential dynamics at resolution of houses and families. The goal was to understand whether the ethnic groups in Yaffo will segregate or integrate in the future.

According to experimental data of Itzhak Omer (1996), Yaffo residents prefer to locate:
- within their ethnic group and
- in a house of specific architectural style.

The description of Yaffo residential dynamics should thus account for HOUSES, characterized by the architectural style (Block or Oriental), and FAMILIES, which can belong to one of three ethnic groups.
To decide whether to occupy the house, a model FAMILY has to know its style and the neighbors.

FAMILIES are neighbors if they reside in HOUSES, which can be directly seen from one another.

Formally, it means that the Voronoi polygons of the houses should have common boundary.

THE PROBLEM:

- To estimate the ethnic structure of the neighborhood, each model year we have to retrieve HOUSES, the Voronoi polygons of which are neighboring to a Voronoi polygon of a given HOUSE. This is a computationally heavy problem when repeated for 1700 Yaffo houses.
- To know neighboring FAMILIES we have to know the HOUSE that the FAMILY occupies

THE SOLUTION:

- Store the neighborhood relationships between HOUSES as a table of HOUSE-HOUSE relationship and retrieve them when necessary.
- To know location of FAMILIES, store them as a table of FAMILY-HOUSE relationship and retrieve when necessary. When FAMILY leaves the HOUSE the corresponding row in this table should be deleted, when it occupies, a new row should be inserted



THE GEOGRAPHIC SYSTEM BECOMES, THUS, A DYNAMIC DATABASE OF SPATIALLY LOCATED OBJECTS!

A DATABASE VIEW HAS IMMEDIATE ADVANTAGES:

For example, working on the Yaffo model, we wanted to test some abstract idea regarding human behavior during residential search.

The irregular partition was just noise for this problem.

Do you have to write a parallel CA-MAS model in this case? For the dynamic database the answer is NO!

DO THE FOLLOWING:
  1. Build square cellular grid in GIS;
  2. Apply a GIS algorithm defining polygons as neighbors if they have common boundary (as for Voronoi polygons) and build the table of neighborhood relationships between cells of the grid;
  3. Substitute layer of Yaffo houses by grid of cells;
  4. Substitute table of neighborhood relationships between Yaffo houses by the table of relationships constructed for cells.
DO NOT CHANGE ANYTHING ELSE in the model, including rules of FAMILIES residential behavior.