Scientific workflows for parameter-setting support in environmental modelling. GeoComputation 2019

Parameter uncertainty, restricted information and insufficient transparency of many environmental models are often limiting factors in their use and application. The aim of this work is to demonstrate how scientific workflows can be designed and implemented to address these limitations in use and application of environmental models. Specifically, this work shows an application of scientific workflows and their potential in delivering a transparent, reproducible, integrated and automated tool for a parameter setting decision support. The results show how parameter uncertainty can be captured and managed using a freely-available and open-source scientific workflow management system (SWFMS). Specifically, scientific workflows integrate environmental modelling and a machine learning technique to discover and select appropriate parameter settings. This work shows a case study with the particular example of environmental modelling for estimating solar radiation potential of building rooftops. In general, an appropriate user support and management of parameter uncertainty help to broader usage and warrantability of environmental models and their outputs, both in research related work and in interdisciplinary analysis for decision making.