Description

Rainfed lowland rice growing is practiced in northeastern Thailand under a climate with a six-month dry season and coarse textured soils. The drought risk is still high at the beginning of the wet season. The seasonal labour migrations have for decades been an adaptive mechanism widely used by smallholdings to face up to climate risks and the low rice productivity in the poorest region of the country. With certain migrations becoming definitive and authorities wishing to improve rice growers' access to irrigation, an improved understanding of the interaction between labor management on and off the holding and the use of land and water is crucial to designing well-adapted hydraulic schemes and to foresee to whom they would benefit. Warong Naivinit is carrying out his doctoral research on this subject by applying the ComMod approach to understand this interaction in the central part of the Lam Dome Yai watershed.in southern Ubon Ratchathani province.

The approach has been implemented through several iterative loops between the field and the laboratory since 2004:

1. Review of the literature on the local agrarian system, especially rural worker migrations. Construction of the first UML diagrams to conceptualise the interactions between the agricultural system components.
2.  Design, test and use of a first role-playing game in the Ban Mak Mai village. This game represented the researcher perception of the interaction under study during key phases of the annual rice growing cycle (labor management at transplanting, harvesting and in the dry season) and interactions between types of rice-growing holdings. It was replayed subsequently with the migrants and the gaming sessions led to modifications of the initial UML conceptual diagrams.
3.  Construction of a second game, more explicit about the cropping calendar and a computerised application playing the game under CORMAS. Observation of the adaptation to very dry periods and irrigation (from canal, individual or community ponds).
4.  Gradual construction of a multi-agent model representing the interaction under study for four different rice farms by alternating programming sessions and submission of successive versions of the prototype for review by rice growers.
5. Validation of the improvements of the multi-agent model requested and simulation of scenarios of interest to its users.

 The collaborative modelling approach adopted could be spaced out over a long period and was ideal for multiple exchanges of viewpoints on the simulated interaction between the researcher and a dozen rice farming households, some offering work force, others hiring it out during peak of labour demand periods in the rice-growing cycle. Work on the various effects of the approach on the rice growers who took part heralds the end of this research.