Your location: Home

Compendium on methods and tools to evaluate impacts of, and vulnerability and adaptation to, climate change

CLOUD (Climate Outlooks and Agent-based Simulation of Adaptation in Africa)

CLOUD developed a proto-type multi-agent simulation (MAS) model that is coupled to information about farming activity and climate data.

MAS is a general computational technique that can be used for simulating societies and their interactions with the environment. It is a processed-based modelling approach arising from artificial intelligence research: Individual agents in a given system are identified and sets of behavioural rules are used to evolve the system state forward in time, taking into account the interactions between agents and the feedbacks between agent actions and the changes in state of their environment. For social-system models, agents may represent individual people, households, social groups or larger institutions.

Environmental models can be coupled to the MAS in a variety of ways, depending on the type of physical or ecological system under consideration. In a climate context, such a model forms a test-bed for hypotheses about how environmental change and its perception within society may affect both adaptation to change and the future climate, in much the same way that a physical model allows testing of hypotheses about the physics of the atmosphere and oceans.

Appropriate Use The model is as yet exploratory in nature and not sufficiently sophisticated to use for policy advice. For models that examine forecasting and climate, the CLOUD model has been applied to small subsistence farming villages in dryland farming regions where some seasonal forecast skill might be available.
Scope The model focus was originally Southern Africa and specific to a single village. With appropriate social survey, crop and climate data it could be applied to other regions. However, MAS itself has in principle much broader applicability for social simulation.
Key Output An exploratory model for examining “What if” scenarios. In the case of CLOUD, how belief in seasonal forecasts might affects crop returns and benefit (or otherwise) subsistence farmers.
Key Input For forecast simulations, time series rainfall and temperature data at monthly resolution, crop coefficients suitable for CROPWAT, social information regarding agricultural practice and conditions under which crop choices are made, price and distribution information for crop sales, and climate forecast skill information.
Ease of Use For programming experts.
Training Required Requires expert programming knowledge in C++ or Java and experience in multi-agent simulation.
Training Available None currently available.
Computer Requirements Platform independent self-contained code. For single village simulations, any modern computer has sufficient power to run the model.
Documentation Documentation is currently available as comments within the code.

Subsistence farmers are particularly vulnerable to fluctuations in climate, particularly rainfall. Seasonal, inter-annual and longer-term changes in availability of water all affect their ability to survive. One way in which they might improve their circumstances would be to have access to seasonal forecast information, so as to be able to anticipate the right crops to plant, both for food and for marketing. Dry-land farmer success therefore depends on the availability of rain, and on the prices they can get at market, both of which are also dependent on the climate.

The model was applied to two Southern African villages, Ha Thlaku, where adaptation of practice to planting either maize or sorghum was examined (Ziervogel et al. 2005), and Mangondi (Bharwani et al. 2005), where computer-aided knowledge-elicitation tools were used to determine a set of strategies that the farmers could use to plant crops in a market garden.

An agent-based model was developed that captured these strategies and allowed us to couple them to a crop model (CROPWAT) driven with rainfall and temperature derived from 140-year runs of the UK Met Office coupled climate model. The agents changed their behaviour according to their memory of past climate, their interaction with other farmers, and their belief in a seasonal forecast. The projects studied how these factors influenced the success of the farming community as the climate varied over annual, decadal and longer timescales, including the effect of changes in the accuracy of the seasonal forecasts.

Contacts for Framework, Documentation, Technical Assistance

Dr. M. Bithell

Dept. of Geography, University of Cambridge, England (see

Cost Software is regarded as pre-release at present. Future development envisages a MAS model that will be freely available and open source.

Bharwani, S., M. Bithell, T.E. Downing, M. New, R. Washington and G. Ziervogel. 2005. Multi-Agent Modelling of Climate Outlooks and Food Security on a community Garden Scheme in Limpopo, South Africa. Phil Trans. Roy. Soc. London Ser. B 360:2183-2194.

Bithell, M., J. Brasington and K. Richards. 2006. Discrete-element, individual-based and agent-based models: tools for interdisciplinary enquiry in geography? Geoforum. doi:10.1016/j.geoforum.2006.10.014.

Ziervogel, G., M. Bithell, R. Washington and T. Downing. 2005. Agent-based social simulation: a method for assessing the impact of seasonal climate forecast applications among smallholder farmers. Agricultural Systems 83:1-26.