Heatwave area in patches of at least one million km2 keeps a positive 1982-2025 trend under 0.14 C SST changes in a one-degree OISST sample

We tested which trends of connected marine heatwave area survive bounded changes in daily SST. We combined bounds for new seasonal percentile thresholds, five-day runs, two-day gap joins and fixed spatial graphs. The method gives safe OLS trend bounds for every changed record in the input class. We used 44 NOAA OISST annual files for 1982-2025 and 29,120 fixed ice-free native-pixel samples at one-degree spacing between 60 S and 60 N. For every input change of at most 0.14 C, annual mean area in daily patches of at least one million km2 retains a positive trend with both tested baselines and both four- and eight-neighbour graphs. The four lower bounds are +1.410, +1.342, +0.266 and +0.159 million km2 per decade. The raw rates and allowed ranges are much larger. The planned primary metric, mean largest daily patch area, has a positive lower slope at 0.10 C with both baselines; at 0.14 C it is certified only with the earlier baseline. Ending in 2021 weakens the result: with the later baseline, the 0.10 C area signs are unresolved, while all four metrics remain positive at 0.05 C. This extends earlier local event-day bounds to connected area and separates area growth from unstable event counts. Exhaustive small-record tests, independent graph searches, rebuilt-threshold checks, an external event reference and aggregate audits pass. These are finite-record sensitivity results, not confidence intervals or measured error bounds. Actual SST errors are not shown to satisfy the budgets. The sample is one product and does not reproduce native-resolution or tracked three-dimensional heatwave systems. Proofs, code, annual evidence, figures, exact source hashes and retrieval steps are included. Raw data stay outside the bundle.