Temporal aggregation is an important query operation in temporal databases. Although the general forms of temporal aggregation have been well researched, some new applications such as online calendaring systems call for new temporal aggregation. In this paper, we study the issue of approximate temporal aggregation with nearby coalescing, which we call NSTA. NSTA improves instant temporal aggregation by coalescing nearby (not necessarily adjacent) intervals to produce more compact and concise aggregate results. We introduce the term of coalescibility and based on it we develop efficient algorithms to compute coalesced aggregates. We evaluate the proposed methods experimentally and verify the feasibility.