Background
Water companies need accurate information on impervious urban surfaces
A significant part of precipitation ends up in the public sewerage system and must be cleaned together with wastewater. Water companies (who are responsible for water distribution and wastewater treatment) charge rainwater treatment fees to the owners of areas which create run off. The calculation of rainfall in the sewerage system is specified by law and is based on the size of impervious surfaces on a given plot of land. Precipitation that falls on such areas (e.g. roofs or car parks) cannot seep into the ground and runs off into the sewerage system.
Client-reported data is not always reliable
Information on impervious surfaces is not kept in the land register or in any other official register. Water company networks often cover hundreds of square kilometres of built-up areas, so it is impossible to physically check every plot of land. Therefore, in most cases they must rely on client-reported data. Unfortunately, such data tends to underestimate the actual situation.
Our approach
Monitoring of large areas requires an aerial view
Aerial photographs and satellite images provide accurate information on the actual situation. They permit remote inspection of any location and provide savings as regards time and money when compared to a physical visit. However, they are not a solution for large areas that cannot be checked manually.
AI and computer vision make it possible to check hundreds of square kilometres in a few hours
At PwC, we have developed a fully automated solution addressing all the above issues. It is based on a computer vision model able to distinguish surface types from aerial photographs. Model outputs are then compared with the land register and water company data. In this way, we obtain a list of specific cases where the surface determined from an aerial photograph does not match with client-reported data for the area. So, water company employees do not have to manually check all the client data and can focus on cases where there is a reasonable suspicion that data is inaccurate.
End-to-end solution for maximum control automation
In addition to the analysis of aerial photographs, our solution performs other steps required for automation. One of these is input data quality control. Internal client records often contain different recording formats, missing evidence, or data input errors. We can correct most of these automatically. Communication with clients about found discrepancies is another important step. We automatically generate client letters with inspection results and the visualization of inspected areas.
Benefits
Our analysis revealed that, in some locations, up to 60% of data on impervious surfaces is underestimated. This can result in shortfalls in rainwater treatment fees of up to 40%. Since it is a sector with regulated prices, other clients must bear the unmet rainwater treatment costs.
We tailor the complete technical PwC solution to a client’s specific requirements. In addition to checking data on rainwater treatment payments, we can also carry out an overall inspection of unreported surfaces and public areas.
Contact us
