Regional 10 m wind-speed products extend coverage beyond sparse station networks, but their errors vary with wind regime and location. This study evaluates a station-trained LightGBM correction of a 5 km WRF/HNR wind-speed product over Hainan Island. Raw and corrected fields were collocated with stations using a four-neighbour inverse-distance operator. A spatially separated 2022 holdout retained 67 stations after five coordinate-overlap exclusions. Across 582,610 common hourly pairs, corrected RMSE decreased from 3.198 to 1.211 m s⁻¹ and Pearson R increased from 0.368 to 0.651. The paired RMSE reduction was 1.986 m s⁻¹ [95% CI 1.839, 2.122]. Improvement was largest below 3.4 m s⁻¹ (2.203 m s⁻¹) and remained positive for 3.4–7.9 m s⁻¹ (0.566 m s⁻¹). In contrast, corrected RMSE increased by 0.990 m s⁻¹ for 8.0–10.7 m s⁻¹ and by 1.656 m s⁻¹ for ≥10.8 m s⁻¹. A time-only adjustment reduced pooled RMSE to 1.870 m s⁻¹ but also degraded the two upper strata. The 2016–2022 maps show a lower corrected product climatology, with a mean corrected-minus-raw increment of −2.273 m s⁻¹ across the Hainan buffer. Temporal partitions and AWS-selected station groups show variation across season, time of day, and setting. The correction reduces station-sampled errors for light-to-moderate wind speeds; it changes speed magnitude while retaining the raw-product wind direction.

Heavy rainfall records over the South China Sea accompanies weather systems that can also alter sea-surface temperature (SST), but the magnitude, heterogeneity, and scale dependence of the associated post-event cooling remain unclear. Using 1,192,517 TRMM-defined heavy-rain grid-cell/day records and matched same-grid, same-calendar-month controls during 1998–2015, we quantified the 1–3–day post-event SST contrast and examined rainfall–SST correspondence at daily, monthly, and pooled spatial resolutions. The mean event-minus-control contrast was −0.228 °C (cross-year sensitivity interval: −0.243 to −0.214 °C). Cooling varied geographically, from −0.427 °C at Dongsha to −0.169 °C at Nansha, and seasonally, from −0.315 °C in March–April to −0.130 °C in December–February. Same-grid overlap between SST-response and heavy-rain records was weak on the exact date (F1 = 0.059) and remained limited within ±2 days (F1 = 0.276). Across 216 held-out months in reciprocal temporal folds, basin-wide monthly TRMM heavy-rain counts and weighted SST-response record equivalents co-varied strongly (r = 0.854), whereas their de-seasonalized anomalies were only weakly associated (r = 0.197). Pooled January and July maps retained broad spatial structure after combining 18 held-out instances. Because atmospheric and oceanic covariates were not controlled, the SST contrast represents a conditional association rather than an isolated rainfall effect. These findings identify geographically and seasonally heterogeneous post-event cooling, while showing that strong correspondence in aggregated diagnostics largely reflects shared seasonal structure rather than daily record-level or monthly anomaly agreement.