While climate change is a global issue, the change in itself is not homogeneously distributed over the globe. It is well established that near-surface Arctic warming is on average 3-4 times larger than the global-average warming. This so-called Arctic Amplification is due to a number of positive feedbacks in the Arctic, some of which are poorly understood. On a basic level, Arctic climate is determined by a balance between inflows of energy from the south and the net loss of energy by radiation at the top of the Arctic atmosphere. Both are large while their difference is small. The inflow of heat from the south occurs in both the atmosphere and ocean. From a numerical modeling perspective it occurs on sufficiently large spatial and temporal scales that it is considered resolved, however, a disproportionately large fraction of the heat transport into the Arctic happens in discrete localized events, sometimes referred to as atmospheric rivers. The net energy flux at the top of the atmosphere also has a very large annual cycle: positive but small in summer, when the solar radiation is at its maximum, but large and negative in winter when the sun is absent. The net radiative flux at TOA depends on a number of processes, including sea-ice cover, surface temperature and albedo, atmospheric chemical composition, clouds and aerosols etc. All of these have in common that they are not resolved in numerical models and hence have to be parameterized, described parametrically as functions of larger-scale resolved variables. Different in different models, models typically have substantial systematic but sometimes compensating errors in these descriptions and to a large extent this explains the spread in climate model projections of future climate and systematic errors in weather forecasts. Arctic Ocean near-surface air temperature, as a proxy for climate, goes through a substantial annual cycle with two main states; these can be characterized as either freezing or melting. Physically, in some sense, the Arctic Ocean surface only has two seasons – the melt season and the freeze season. In winter with surface temperature below the melting point, the surface temperature reacts to changes in the surface energy budget, hence, it features large and fast changes in response to changes mainly in incoming radiation. In summer, or the melt season, the surface temperature is prevented from increasing above the melting point by the phase change of melting, as long as there is substantial ice and snow remaining, and all excess energy goes into melting rather than into warming. Consequently, the summer near-surface air temperature varies only a little. How much ice melts over the melt season is directly related to the length of the melt period but also indirectly to what happens in winter. If the melt season becomes longer it follows that sea ice extent at its minimum in September will decrease. The length of the melt season is therefore one important component of the Arctic climate system, and studying and understanding the so-called shoulder seasons – the transition between melt and freeze both in spring and autumn – is of great interest in order to understand the Arctic climate system. Historically, icebreaker-based expeditions, capable of performing scientific-grade process-level observations have occurred in summer or early autumn because the ice is easier to navigate in the Arctic Ocean, when melting. Hence, a number of Oden expeditions have been able to observe the transition from surface melt to surface freeze in late August or early September. However, only a few have collected such observations at the melt onset. Hence, on a process level, there are no relevant observations of the melt onset. The ARTofMELT expedition was conceived to rectify this, studying the relationship between this onset and atmospheric rivers.