Wikiversity:Fellow-Programm Freies Wissen/Einreichungen/A user-friendly open source software package for statistical climate reconstructions
A user-friendly open source software package for statistical climate reconstructions[Bearbeiten]Making statistical climate reconstruction methods more transparent Projektbeschreibung[Bearbeiten]Climate science gains great public interest due to the anthropogenic climate change. Paleoclimatology studies the past climate on time scales from centuries to millions of years using indirect observations like plant remains or isotope compositions. Thereby, it provides new insights into the processes that drive the climate system. It is an interdisciplinary scientific community that connects physicists, geologists, biologists, and statisticians, amongst others. The sparseness of data and the complex relationship between climate states and corresponding indirect observations demands the use of advanced statistical methods for climate reconstructions. However, many sophisticated methods are rarely used due to missing user-friendly software solutions and a lack of understanding by non-statisticians. Driven by the requirement of transparent research to inform political decision-making for example through the IPCC reports, current efforts within the paleo-community strive for open access to data and publications. Similar concerted efforts for open and transparent research methods are missing. This hampers reproducibility of research results and the wide-spread adoption of innovative statistical methods. In the project, an open access software package is developed which provides a set of methods for statistical climate reconstructions. Bayesian statistics are used, which are well-suited for applications in paleoclimatology because of their ability to combine multiple data sources and quantify reconstruction uncertainties. The package will contain a chain of functions from data processing, over the statistical reconstruction techniques, to the visualization of results. The package will be programmed in a user-friendly way such that it can be used by non-statisticians. A modular framework will make it easy to integrate new data types and corresponding methods. A focus will be on interactive visualization techniques which contribute to a more transparent use of methods. This includes interactive and animated maps to illustrate reconstruction results, graphs of data processing chains to make reconstruction methods more transparent and traceable, and a "shiny app", which provides a graphical interface to explore reconstruction methods. The package will be available from CRAN, the leading R package management platform. It will be used in a block-seminar at Heidelberg University, a short course at the EGU 2020, the largest geosciences conference in Europe, and a PAGES Early-Career Network webinar. It will be promoted on conferences and in international projects like PAGES, PMIP, and PalMod. Visualizations with the package will be made available on Wikimedia Commons and contribute to Wikipedia articles. The interactive shiny app can be used online by the interested public. Autor/in[Bearbeiten]
|