On the use of smartphones as novel photogrammetric water gauging instruments: Developing tools for crowdsourcing water levels

Publikation: Hochschulschrift/AbschlussarbeitDissertation



The term global climate change is omnipresent since the beginning of the last decade. Changes in the global climate are associated with an increase in heavy rainfalls that can cause nearly unpredictable flash floods. Consequently, spatio-temporally high-resolution monitoring of rivers becomes increasingly important. Water gauging stations continuously and precisely measure water levels. However, they are rather expensive in purchase and maintenance and are preferably installed at water bodies relevant for water management. Small-scale catchments remain often ungauged. In order to increase the data density of hydrometric monitoring networks and thus to improve the prediction quality of flood events, new, flexible and cost-effective water level measurement technologies are required. They should be oriented towards the accuracy requirements of conventional measurement systems and facilitate the observation of water levels at virtually any time, even at the smallest rivers. A possible solution is the development of a photogrammetric smartphone application (app) for crowdsourcing water levels, which merely requires voluntary users to take pictures of a river section to determine the water level. Today’s smartphones integrate high-resolution cameras, a variety of sensors, powerful processors, and mass storage. However, they are designed for the mass market and use low-cost hardware that cannot comply with the quality of geodetic measurement technology. In order to investigate the potential for mobile measurement applications, research was conducted on the smartphone as a photogrammetric measurement instrument as part of the doctoral project. The studies deal with the geometric stability of smartphone cameras regarding device-internal temperature changes and with the accuracy potential of rotation parameters measured with smartphone sensors. The results show a high, temperature-related variability of the interior orientation parameters, which is why the calibration of the camera should be carried out during the immediate measurement. The results of the sensor investigations show considerable inaccuracies when measuring rotation parameters, especially the compass angle (errors up to 90° were observed). The same applies to position parameters measured by global navigation satellite system (GNSS) receivers built into smartphones. According to the literature, positional accuracies of about 5 m are possible in best conditions. Otherwise, errors of several 10 m are to be expected. As a result, direct georeferencing of image measurements using current smartphone technology should be discouraged. In consideration of the results, the water gauging app Open Water Levels (OWL) was developed, whose methodological development and implementation constituted the core of the thesis project. OWL enables the flexible measurement of water levels via crowdsourcing without requiring additional equipment or being limited to specific river sections. Data acquisition and processing take place directly in the field, so that the water level information is immediately available. In practice, the user captures a short time-lapse sequence of a river bank with OWL, which is used to calculate a spatio-temporal texture that enables the detection of the water line. In order to translate the image measurement into 3D object space, a synthetic, photo-realistic image of the situation is created from existing 3D data of the river section to be investigated. Necessary approximations of the image orientation parameters are measured by smartphone sensors and GNSS. The assignment of camera image and synthetic image allows for the determination of the interior and exterior orientation parameters by means of space resection and finally the transfer of the image-measured 2D water line into the 3D object space to derive the prevalent water level in the reference system of the 3D data. In comparison with conventionally measured water levels, OWL reveals an accuracy potential of 2 cm on average, provided that synthetic image and camera image exhibit consistent image contents and that the water line can be reliably detected. In the present dissertation, related geometric and radiometric problems are comprehensively discussed. Furthermore, possible solutions, based on advancing developments in smartphone technology and image processing as well as the increasing availability of 3D reference data, are presented in the synthesis of the work. The app Open Water Levels, which is currently available as a beta version and has been tested on selected devices, provides a basis, which, with continuous further development, aims to achieve a final release for crowdsourcing water levels towards the establishment of new and the expansion of existing monitoring networks.


Gradverleihende Hochschule
Betreuer:in / Berater:in
  • Maas, Hans-Gerd, Mentor:in
  • Kersten, Thomas, Gutachter:in, Externe Person
  • Lindenbergh, Roderik C., Gutachter:in, Externe Person
Datum der Verteidigung (Datum der Urkunde)3 Mai 2021
Herausgeber (Verlag)
  • Bayerische Akademie der Wissenschaften
ISBN's (print) 978 3 7696 5343-4
PublikationsstatusVeröffentlicht - 15 Juni 2021
No renderer: customAssociatesEventsRenderPortal,dk.atira.pure.api.shared.model.researchoutput.Thesis

Externe IDs

ORCID /0000-0003-2169-8762/work/142659493



  • Crowdsourcing, Flood, Photogrammetry, Smartphone, Computer Vision, Hydrology, Gauging, Low-Cost