TraceMe

Under the background of global climate change, the development of earth science makes the structure of earth system model more abundant and the amount of observed data explosively increase. However, there are huge uncertainties in different model simulations, which requires the model evaluation to be more systematic and comprehensive. More complex model structures and massive amounts of data present new challenges to current model evaluation, One of the challenges is that model evaluation methods need to be able to trace the sources of uncertainty between different models from complex model structures; Another challenge is that the uncertainty between the components of different models has a certain transitive relationship and it is necessary to be able to quantify the uncertainty between the components. In order to better understand and explore the uncertainty of model estimation, model estimation requires new assessment methods and comprehensive observational data sets to trace and quantify the sources of uncertainty of model estimation. In view of the new challenges faced by model evaluation, the research group is committed to developing a new model evaluation system based on traceability analysis method (as shown in the left figure). The traceability analysis method is based on the matrix method to simplify the complex model structure and systematically reflect the structural relations of different simulation components. By means of traceability analysis method, model evaluation can trace the structural sources of uncertainty and quantify the contribution of different estimation components to the uncertainty of the target estimator.