Comparison of methods for initialization of the Open Global Glacier Model (OGGM)

In order to assess glacier mass balance on large temporal and/or spatial scales, numerical modelling is an essential tool, complementing ground observations and remote sensing methods. For a reliable simulation of a glacier’s development over time, knowledge of its initial state is fundamental. Attaining this information entirely through empirical evidence is impossible due to a lack of data, hence the need for alternative, numerical methods. In this study, three methods of varying complexity are applied to initialize the Open Global Glacier Model (OGGM) for 254 glaciers. These glaciers have a minimum of 5 years of in-situ mass balance observations, allowing for direct comparison with modelled values. The initialization methods comprise, in brief, i) a basic spin-up, starting from present-day conditions, running the model for 200 years with a random climate, representative of the period 1900-2000 ii) a cold climate spin-up, allowing the glacier to grow and create a more representative initial condition for e.g. the year 1901 and iii) a synthetic experiment based on present day glacier observations and past climate information, used to generate a large set of physically plausible initial states, which are then evaluated. Using each method, we reconstruct the glaciers’ initial states and set up a forward run from which to extract mass balance values over the time period 1970-2014, used for validation purposes. The overall aim is to identify an initialization approach that can be successfully applied to our current set of 254 glaciers, as well as areas with even sparser data available, expanding the range of scale for glacier modelling.