CatchScape.

Conflicts over water in Northern Thailand.

An individual based viewpoint about catchment scale issues.

Nicolas Becu (ENGREF), Pascal Perez (CIRAD), Andrew Walker (ANU)

In April 1998, an extraordinary event took place in Northern Thailand. Five thousands lowland Thai farmers from the district of Chom Thong occupied and blocked one of the district's main roads. They were demanding the forced relocation of almost 20,000 hill-tribe villagers whom they accused to cause widespread environmental damage: deforestation, forest fires and streamflow drying. This event was the first sign of a global emerging conflict between lowland and upland farmers of Northern Thailand about natural resources management [Scoccimarro et al., 1999]. On one hand, upland settlers are accused of reducing streamflow, through deforestation, and, on the other hand, lowland farmers are increasing their demand over water. Moreover, the conflict is emphasized by ethnical differences between upland and lowland people. Regarding the complexity of the existing interactions, solutions may only arise from local negotiation. But stakeholders need descriptive, integrative and anticipating tools in order to share a common view and to reach a sound consensus.

Towards a fully integrative model

In order to focus on social interactions and resources management, we have first defined the farmers as cognitive agents (Farmer) and then, the other elements that compose the farmer's environment: the crops (Crop), the river (River), the irrigation canal (Canal) and the village (Village) have been created as reactive entities. The Farmer has to manage four main resources and their interactions: water, labor, cash and land. The catchment has been schematically represented in order to sketch the different levels of organization of the relevant spatial units. First, the Land Units combine soil texture, soil depth and land slope. The second spatial representation concerns Land Use: Paddy, Upland and Forest. the Paddy zone is an irrigated area composed of a multitude of bounded terraces on which farmers mainly crop rice during the wet season. The Upland zone is constituted with rainfed plots spread all over the hillsides and cropped either with rice, soybean or vegetables. Forest, mainly used for gathering and religious ceremony, is described as a sempervirens sole type cover.

Model description

The unit cell (Plot) that composes the modelling grid (44*45), corresponds to a 2-rai farm plot (1 rai = 0.16 ha), which is the average size encountered in the Mae Uam catchment. Farmers are characterized by their family size and labor force. they can initially own upland and/or paddy plots according to their status. A paddy Plot belongs to a Canal. There are six Canals in the system, organized by pair and grouped into two irrigated schemes. A cognitive agent, called Manager, manages the weir controlling a Canal. The irrigated schemes belong to one zone each (Village) corresponding to the upstream and downstream groups of actual villages. The biophysical dynamics are simulated through a distributed water balance model and a hydraulic model [Perez et al., 2001]. First, Farmers choose the most profitable crop according to their constraints in cash, labor force and water availability. Except for rice during the wet season, the Crop choice is based on a simplified Linear Programming model taking into account seasonal farming costs, water and labor requirements.This initial Crop choice may be modified if the growth duration doesn't reasonably fit within seasonal boundaries. The planting date is delayed as long as Farmer's cash and labor force can't match the land preparation and planting requirements. Rice cultivation is mostly motivated by food security strategies. Thus, within the model, Farmers are forced to plant rice at the beginning of the rainy season as long as their yield expectation doesn't reach their family needs, cash and labor resources permitting. At the beginning of the dry season, Farmers have to decide whether they allocate part of their labor force to off-farm activities or not. Finally, Farmers have to make decisions about land dynamics. Three opportunities are offered to the cognitive agents. First, Farmers can buy available Plots. Then, they can decide to install irrigation on rainfed Plots (located in the uplands), in order to farm them during the dry season. At last, Farmers can convert forest Plots into upland Plots. Paired Canals enter into an irrigation rotation as soon as the downstream canal faces water shortage. The upstream Manager is forced to accept the rotation but may stop it if the River's streamflow comes back to normal. Negotiations may involve Managers from different groups of paired Canals and, eventually, from different Villages. In this case, downstream Managers still send requests to the upstream ones but the later are not forced to respond positively.

Preliminary results

Each scenario has been run over a 10-years period, corresponding to the existing climate data set . As several random functions are included in the algorithm, the scenarios have been repeated 20 times in order to estimate the outputs variability. The user can visualize the spatial and social dynamics during simulations according to different viewpoints (Cropping pattern, soil water balance, irrigation canal depletion, individual cash, conditions of negotiation…). At the end of the simulation, a set of indicators can be edited through evolution charts or export files [Bousquet et al., 1999]. In the basic scenario, social inequity has dramatically increased in the catchment at the end of the 10-year period. Approximately 10% of the Farmers enjoy a positive cash position (Rich) and gradually increase their wealth. Some of them are rich enough to invest into upland irrigation (15 cases). Meanwhile, 40% of the Farmers (Poor) enter a permanent and worsening deficit position. The remaining 50% hardly maintain their economic buoyancy. Even if Farmers initially entitled with only 2 upland Plots partly feed the Poor category, the irrigated Plot location along the Canal largely contributes to its owner economic achievements. For example, during the dry season, the upstream Canal (C1) can secure enough water to irrigated approximately 65% of its Plots. But the downstream Canal (C6) can hardly support 20% of the connected Plots. In this case, as Farmers were endowed with the ability to learn from previous failures, gradually, the number of attempts to crop during the dry season decreases.

References

Bousquet, F., O. Barreteau, C. Le Page, C. Mullon and J. Weber, An environmental modelling approach. The use of multi-agent simulations, In Advances in environmental and ecological modelling, (Eds, Blasco, F. and A. Weill), Elsevier, pp. 113-122, 1999.

Perez, P., N. Ardlie, P. Kuneepong, C. Dietrich, W.S. Merritt, CATCHCROP: Modeling crop yield and water demand for Integrated Catchment Assessment in Northern Thailand, Submitted to Environmental Modelling and Software, 2001. Scoccimarro, M., A.

Walker, C. Dietrich, Schreider, A.J. Jakeman, H. Ross, A Framework fir Integrated Catchment Assessment in Northern Thailand, Environmental Modelling & Software 14, 567-577, 1999.

For more information, contact the authors. N. Becu, P. Perez