Analysing the uncertainties, we have to distinguish between different types of uncertainties. Manning and Petit (2003) in their paper for the IPCC identify five types, which all are relevant for LSL projections:
- Aleatory uncertainties are most severe in terms of vertical land motion and the reference frame's connection to the CM, as well as the oceanographic observations. The incompleteness of the LSL observations both in space and time add additional aleatory uncertainties.
- Epistemic uncertainties enter the picture mainly from the climate modeling side, where large epistemic uncertainties exist, for example, with respect to a potential dynamic and even catastrophic response of the ice sheets to global warming. With respect to the solid Earth's and LSL's response to ice melting, we have very little epistemic uncertainties, since loading on the solid Earth is a well studied and understood phenomenon.
- There are many aspects both in climate and in LSL projections where incorrect descriptions of processes, over-simplifications, or even program errors may impact our projections. A particular case is the computation of the fingerprints of ice sheet melting in sea level, where large differences between different groups are found. These differences are most likely due to incorrect description of processes or errors in programs.
- Chaos comes into the picture mainly through the climate system, which includes the atmospheric forcing, the ocean circulation, thermal expansion, and the melting of land-based ice. Chaos does not come in in the computation of the LSL fingerprint of ice sheet melting, which is a rather deterministic and to the largest extent linear problem.
- There is certainly a lack of predictability with respect to the ice sheet response to global warming, and this introduces a large uncertainty into LSL projections.