Due to computational limits, temporal details within Energy System Optimisation Models are often reduced, for example by reducing the time horizon or by resampling via Time Series Aggregation (TSA) techniques. In high RES energy systems, this may lead to undersizing of Long-Duration Energy Storage (LDES) capacities, necessary for system flexibility, due to the omission of long-term interannual weather effects. Via comparative analysis between the capacity expansion results for different subsets of weather years, this paper shows the extent to which single year models underpredict LDES. but also that a small cluster (n=2, 3) of weather years can adequately capture key system-defining weather patterns. Identifying these weather years ex-ante is non-trivial, as there is no obvious correlation with how well they describe the full set of weather years. As this assumed correlation underpins current time series aggregation techniques, new techniques are required.