Spatiotemporal Analysis of Pre-Monsoon Severe Storms in Northern Thailand using Integrated Crowdsourced and Official Data (2015–2024)

Abstract

Severe convective storms provide a significant weather threat in Northern Thailand, especially during the pre-monsoon season (March–May), however a comprehensive storm database has been unavailable. The absence of comprehensive records complicates the research of storm features, the validation of numerical weather models, and the enhancement of radar-based detection methods. To address this deficiency, we established a spatial-temporal database that integrates official report records with crowdsourced social media data, documenting 259 distinct storm days at the regional level and 478 documented instances at the province level across a decade (2015–2024). Temporal study indicated a distinct seasonal pattern, characterized by an escalation in storm frequency from early March to late April, followed by a reduction in late May. Interannual variability was apparent, with 2018 exhibiting anomalously low activity (6 storm days) in contrast to peak years in 2020–2021 (33–35 days). Statistical analysis revealed substantial disparities in storm occurrences between early and late March (p = 0.0073), while Mann-Whitney U tests demonstrated that 2018 had significantly less storms than the following years (p < 0.05). The majority of storms impacted individual provinces (141 occurrences), with a diminishing frequency as the spatial expanse expanded. GIS-generated spatial anomaly maps indicated provincial variations in storm frequency compared to the decadal average. An investigation of atmospheric sounding from ten exceptional multi-province occurrences revealed elevated CAPE values (1215–1876 J/kg) and negative Lifted Index values, indicating conditions favorable to severe convective development. The database differentiates between hail-producing and non-hail-producing storms, facilitating the detection of unique meteorological variables linked to each category. This research establishes a systematic framework for storm recording in areas with limited observational networks, hence enhancing early warning systems, disaster preparedness, and the validation of radar-based storm detection. Future applications encompass comparative research of hail and non-hail storm situations, as well as integration with Thailand’s national radar mosaic to improve severe storm identification.