Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Assessing the sensitivity of water demand to climate change Joanne Parker* and Rob Wilby *[email protected], Department of Geography, Loughborough University, Loughborough, LE11 3TU The ‘Golden 100’ dataset Overview This study uses Anglian Water’s ‘Golden 100’ and Survey of Domestic Water Consumption (SODCON) datasets for the east of England to examine the sensitivity of metered water demand micro-components to climate variability and change. We initially revisit an earlier study by Atkins (2005) which used a multiple linear regression approach to forecast water demand. Our sensitivity analysis shows the extent to which climatic and non-climatic drivers could shape future peak water demands and hence the degree to which such demand might be managed. The study will provide a basis for testing the robustness of Anglian Water Services’ strategic water planning as well as inform wider debates about balancing water supply and demand. In addition, the study is providing case study material for a regional assessment of the resilience of water supply, distribution and demand to climate change in SE England . The EPSRC-ARCC water project will produce a Regional Water Systems Model (RWSM) which simulates supply intakes, demands and raw and potable water transfers within and between water service areas. The RWSM will be operated within a Multi-Criteria Robust Decision Analysis (MC-RDA), which tests against a series of forcing factors to identify system vulnerabilities under uncertain boundary conditions and thereby help design more robust adaptation solutions. Study aim To examine the sensitivity of long-term water demand micro-components to climate variability and change in order to inform water resource management and enable robust water planning. Background/context Source Anglian Water Services Area Covered East, Lincoln, Ruthamford Duration 1992 – 2002 daily readings Variables 9 micro-components 8 meteorological variables 5 socio-economic variables Climate variability and change threatens UK water security through factors such as altered drought frequency and intensity, changing water demand profiles and damage to infrastructure. These potentially could alter water availability for storage, abstraction and supply and present new challenges for the UK water sector. Furthermore, climate change sits amongst a host of other pressures such as socio-demographic change and population growth. Considerations of the impact of climate change on future water supply and demand are very unbalanced. To date, relatively few studies have examined potential impacts on water demand. The narrow long-term demand forecasting literature base can be attributed to the difficulty in recording, understanding and predicting the complex nature of domestic demand (Memon and Butler, 2006; Medd & Chappells, 2008 ). No. of data fields ~18 million An automated algorithm has been created to undertake a rigorous quality control process on the datasets. For example; • Rogue values (such as when min temp > max temp) were identified and excluded. • Large outliers were identified and excluded using a percentile approach (e.g., 983,020 litre/d for a 3 occupancy household). • Dummy values were assigned to day of the week and month. • Zero total daily PCC values were excluded (as these skew the regression models and imply zero occupants in any case). Example results using the ‘Golden 100’ dataset • The following example shows how a multiple regression analysis of Per Capita Consumption (PCC) for single and four person occupancy households in the Ruthamford region can be used as a diagnostic tool to highlight areas of further exploration and interest. From this model alone many intriguing questions emerge. Some examples are explored below. Currently when assessing the impact of climate change on demand most water companies apply factors from the Climate Change and Demand for Water Revisited project (CCDeW, 2003). This is 7 years old and state an average per capita consumption (PCC) prediction for the UK, ~2-3% PCC increase under climate change for the next 25 years. This average PCC masks inter-house variations in water-use habits resulting from different variables such as occupancy rates, cultural-values, and bill payment methods (Gleick, 2002). Map provided by Anglian Water Services, showing the three regions sampled by the ‘Golden 100’ and SODCON datasets. Water demand management is increasingly being viewed as a robust and low regret adaptation option in the face of uncertainties surrounding climate change and other threats to the UK water sector (Memon and Butler, 2006; Wilby and Dessai, 2010). Metered households with four occupants in Ruthamford consume on average 6.5 litres more per 1oC temperature rise. This value is based on historical data but if this relationship and all other factors remain constant, an average 2oC temp rise would result in a 52 litre increase in household water-use per day. • This poses the question why are such households so much more responsive to temperature than single occupants? • The differential metered response of single and four occupant households to temperature suggests that the use of average PCC could conceal a lot of subtle variation at the level of the micro-components of demand. Even a preliminary analysis of the micro-component PCC suggests that 4 occupant metered households consume more water in the practice of showering than single occupant households. Single occupancy metered households appear to use more water in the kitchen sink per capita than four occupant households. Social scientists within the EPSRCARCC project will be considering the underlying causes of these different behaviours. (T-bars show the standard error of the mean) Longer-term study goals and questions Questions for further consideration There is clearly a weekly cycle in water use the amplitude of which depends on the occupancy of the household and whether it is metered or unmetered. Metered households’ water use is more depressed during midweek in relation to Sunday than unmetered. Similarly, four people households show a more pronounced weekly cycle than single . • All modelled houses consume more water on a bank holiday than a non-bank holiday. How might this behaviour be affected by climate change? These example results are based on average historic conditions. The datasets provided by Anglian Water Services cover several years so it may be possible to evaluate time-dependency in the most important loadings of micro-component water demand. • Metered single and 4 occupant households use ~160litres more water on a bank holiday at other times. Will metered households’ water use always be so responsive to bank holidays? The regression modelling can also be used to investigate how different regions, household occupancy levels, and metered/ unmetered households respond to different climate variables. • On bank holidays metered single occupancy households in particular use much more water in the kitchen sink and for WCs. Why is this? The UKCP09 projections could then be applied to relationships determined between micro-components and climatic variables to evaluate the range of uncertainty in water demand projections under climate change. Downing. T.E., Butterfield.R.E., Edmonds. B., Knox. J. W., Moss. S., Piper. B.S., Weatherhead. E. K. 2003. Climate Change and Demand for Water, Final Report. Gleick, P.H. 2002. Soft water paths, Nature, 418, 373. Medd, W., Chappells, H. 2008. Drought and demand in 2006: Consumers, water companies and regulators, Final Report. Lancaster, UK. Lancaster University. Memon, F.A., Butler, D. 2006. Water consumption trends and demand forecasting techniques, In Butler, D., Memon, F.A. (Eds) Water demand management, pp. 1-26, London, IWA Publishing. Wilby, R.L., Dessai, S. 2010. Robust adaptation to climate change, Weather, 65, 73. (Results are presented with the standard error of the mean) We will also need to consider the transferability of our findings to other regions for which there is no such detailed data.