Control charts for improved decisions in environmental management: a case study of catchment water supply in south-west Western Australia


  • Aaron D. Gove,

  • Rohan Sadler,

  • Mamoru Matsuki,

  • Robert Archibald,

  • Stuart Pearse,

  • Mark Garkaklis

  • All authors are Ecologists at Astron Environmental Services, East Perth, Western Australia.

  • This article came about through discussion with research, industry and government colleagues on monitoring tools required to improve assessments of environmental risks using a tool that could provide managers with one metric that displayed trends and decision points in a simple format.


Environmental monitoring is becoming increasingly sophisticated with the widespread adoption of data loggers, sensor arrays and remote sensing, leading to larger scale, higher resolution and superior quality data. However, interpreting monitoring data and deciding when and how to apply environmental management remains a subjective and underdeveloped area of research. Control charts, developed in industrial settings to identify when manufacturing processes were beyond the acceptable bounds of production quality, represent one solution. Despite their potential utility, control charts have rarely been adopted in environmental monitoring. In theory, they are able to identify undesirable trends early and provide transparent and broadly consensual criteria for defining when management action should take place, that is action is triggered when parameter values are observed beyond the agreed control limits of the process. Once triggered, a predetermined management action is implemented. Possible actions are many and varied, and range from investigation and increased monitoring to intervention in the system. Here, the utility of control charts in monitoring water supply in south-western Australia from 1911 to 2010 is examined, and their ability to provide an early, transparent and easily understandable means of triggering management action is assessed. Two control chart types are applied: the X-bar chart and the CUSUM chart. X-bar charts varied widely in their ability to trigger action and were insensitive to many traditional threshold criteria (of which there are many to choose from). In contrast, standard CUSUM charts are specifically designed to detect subtler shifts away from a mean trend and hence provided a more consistent warning of the decline in water supply. While managers were aware of the decline in water supply from an early stage, we believe that control charts could have clearly communicated this earlier, enabling consensus among decision makers to be reached more rapidly.