Results
The thesis consists of two main parts: the tool development and the testing and assessment. The development follows the previous research of processing GNSS track records. The automated approach is unique to consolidate the positioning data in one linear layer with passage frequencies. Testing purposes the parameterisation methodology for optimal results. And the assessment part proves the result for a particular level of accuracy.
Automation Tool
The Jupyter Notebooks for the direct interaction (Google Colab) or the indirect through the web application (Voilà in Binder), allow the user to upload the GPX files to process them under specific parametrisation. A simple tool for compression of the GPX data has been developed to handle the large input datasets in a systematic manner. After the calculation of passage frequencies, the user can inspect the processing outputs in two modes of web maps. The final outputs can be saved in two data formats and two web maps. The cloud-based tool is open to an expert user group in the world. The reference network data come from the global database of OpenStreetMap contributors. However, the ability of the result to convey a strong message originates from the credibility and objectivity of the input GNSS track records.
Testing and Assessment
Three different case studies support the importance of human knowledge in the loop of automation. Each case study aims to find suitable parameterisation for the area. Optimal parametrisation plays the most important role in the outcome. The three tables below illustrate the parameterisation matrix with the optimal test run. The best run has been assessed in the frequency correctness rate. Generally, the frequency precision above 90 % grants the gpx2intensity tool to provide an innovative source of information for road infrastructure development, especially for active transport, i.e. hiking, running, cycling etc.
