We use a model-ensemble approach for simulating permafrost on the Tibetan Plateau. We identify the uncertainties across models (state-of-the-art land surface models) and across methods (most commonly used methods to define permafrost).

We differentiate between uncertainties stemming from climatic driving data or from physical process parameterization, and show how these uncertainties vary seasonally and inter-annually, and how estimates are subject to the definition of permafrost used.