PALS
PALS is a web application for evaluating land surface models and the observed data sets used to test them. It provides data for particular modelling experiments as well as a wide range of diagnostic performance measures once model output created from these data is uploaded. As such, it has two broad categories of users: those who provide observed data sets and are interested in receiving feedback about how models use or interpret their data, and; those who run models using this data and use PALS as a model analysis tool.
Our Involvement
PALS was developed for the Climate Change Research Center of the University of New South Wales. It has the following features:
- A website 'front-end' (this page) that shows the performance of all submitted model output for any dataset or experiment probabilistically. This feature is intended to be something akin to an ongoing, automated PILPS experiment.
- Empirical benchmarking (see Abramowitz et al, J Climate, 2008). A tiered system of benchmarking ranging from 'level zero' (conservation of energy and mass), through several increasingly strict levels of empirical benchmarks to the strictest - model simulations within the range of observational uncertainty.
- Spatially explicit experiments
- Data/experimental setup from existing or previous model evaluation experiments, such as PILPS or GSWP.
Above is a view of the dynamic plots of a Data Set. It enables a data owner or researcher to dynamically drill down through any of the data sets that have been uploaded to manage their quality and spot trends.