Developing a geocomputational workflow to check the consistency of volunteered geographic information
H. Mehdi Poor, Raul Zurita‐Milla, M.J. Kraak · 2015
Checking the consistency of volunteered geographic information is a challenging task. Current approaches addressing this challenge are costly and time consuming as they heavily rely on human interventions. Here, we propose a geocomputational workflow based on the availability of relevant contextual geoinformation. The workflow consists of three main steps: 1) dimensionality reduction using the t-SNE to facilitate further analysis and interpretation of the results, 2) model-based clustering to group the volunteered observations according to their context conditions, and 3) boxplots to highlight inconsistent volunteered observations for each of the clusters. This workflow was successfully tested using volunteered observations on the timing of the first flower of lilac plants. Results indicate that several observations were inconsistent (i.e. unusually early/late regarding their climatic context). These observations are not necessarily wrong, but they should be used carefully to prevent introducing unduly biases in subsequent scientific studies.