Marius Waelchli (Switzerland) 1
1 - ETH Zürich
The term ‘breadcrumb data’ describes data generated as a by-product of human behavior or other activities outside of the scientific realm. Such data is often of unknown quality and more uncertain than standardized data normally used in science. For instance, climatological datasets are generated through complex networks of calibration and subsequent processing. Whether such datasets represent the state of the climate system adequately can only be assessed if the data generation is transparent in terms of technologies, calibration procedures, and theories used. Hence, strategies to reduce uncertainty of climate data include peer-reviewed publications backed up by well-established theoretical knowledge.
Breadcrumb data is subject to the same uncertainties. Additionally, risks that prevent access for the scope and time period required are higher. Such risk factors are for instance of regulatory or legal nature. Despite the uncertainties and risks associated with breadcrumb data, the increasing ubiquity of data generated by e.g. cheap sensors makes it desirable to make use of it for scientific practices. A promising approach is to use such data to increase the spatio-temporal measurement coverage of variables that can only be measured on coarser resolution with classical measurements. This is for instance the case for urban temperature or impact-related variables. However, as it is not always possible to reduce the uncertainty of breadcrumb data by establishing transparency of the data generation process, pragmatic accounts to use such data need to be developed.
In my talk, I will present two case studies based in Zurich, Switzerland, showing how breadcrumb data can be used in urban climate change adaptation studies. The first case study uses private weather stations for urban temperature measurements, the second one uses activity data from a fitness tracking app in order to measure urban mobility behavior under different weather conditions. I will present how standardized reference data can be used to reduce the uncertainty of breadcrumb data by serving as a validation benchmark of the breadcrumb data. I will provide consideration on (i) the extent to which reference data helps to reduce uncertainty of breadcrumb data, (ii) the role of background knowledge, and (iii) why reliable data access remains a major risk. In conclusion, breadcrumb data can provide additional insights into the spatial distribution of climate and impact related variables relevant for urban climate adaptation.