Spatio-temporal modelling of drinking water consumption in Amsterdam; an application and validation of SIMDEUM at the metropolitan level

Organised by Laboratory of Geo-information Science and Remote Sensing

Wed 24 August 2016 09:00 to 09:30

Venue Lumen, gebouwnummer 100
Room 2

By Michael Blok (the Netherlands)

Drinking water is one of the basic human needs. To accurately predict and secure drinking water demand in terms of quantity and quality, modelling can be useful. SIMDEUM (SIMulation of water Demand; an End-Use Model) is a stochastic model and has been developed to predict these future drinking water patterns using a bottom-up approach of demand allocation. Water demand is described based on demographic characteristics, time use data and information on water using appliances. However, the model has never been validated on the scale of a metropolitan area. Therefore, the research objective was to validate SIMDEUM on two spatial scales and several temporal levels: water use per connection and the flow rate of the pumping stations providing the study areas with drinking water, classified as season, day and hour. The model has been applied to two distinct areas: the metropolitan area of Amsterdam and the district of Diemen-Noord.
Validation of SIMDEUM shows that the model is able to predict the amount of drinking water per person and per household well. However, the temporal patterns deviate significantly at times from the measurements. Especially the peak demand in the morning is highly overestimated, but on the other hand water use at night and in the afternoon is underestimated. Weekdays are better predicted compared to weekend days. The mean error is not random, which means that certain variables or patterns are not taken into account or misjudged on the metropolitan level. Though the validation datasets had several limitations, such as deviations, erroneous measurements and incompleteness.

Keywords: demographic statistics; drinking water use; residential water demand; SIMDEUM; simulation; spatial analysis; stochastic model; temporal validation.