Answers are grounded in this paper and cite their sources.
Extreme weather disrupts bus networks worldwide. Transport agencies need to assess how weather affects reliability while travel demand and operating conditions change. Separating these influences is difficult in pandemic-era records, when restrictions coincided with major weather events. Persistent spatial differences further complicate attribution, because aggregate delay can mask the burden on individual corridors and stops. This study analyses 897 million GTFS-Realtime records covering 673 operating days of the Transport for NSW contracted bus network centred on Greater Sydney, linked to station rainfall and temperature observations. Regressions estimate daily weather–delay associations with pandemic-period controls and heteroscedasticity- and autocorrelation-consistent standard errors. Event recovery is assessed against season- and period-matched baselines, and stop-level tests examine hotspot persistence between dry and rainy days. Within the pandemic-affected 2020–2022 window, routine rainfall and temperature show weak associations with daily mean delay, whereas the mid-2021 lockdown is associated with a 44–50-s reduction. These period indicators capture concurrent changes in demand, traffic, and service supply; they do not identify demand effects. Adding daily smart-card patronage removes the significant lockdown association, supporting a demand-related interpretation. Floods show peak excess delays of 42–74 s, with the defined recovery criterion met within one to two days after events. Severe-delay hotspots persist in the same western and south-western corridors across dry and rainy conditions. The findings identify persistent corridor delays, multi-day event preparedness, and performance monitoring as areas for adaptation appraisal. The assessment can be repeated in other metropolitan bus systems, but the relative importance of these areas requires testing under local conditions and restored post-pandemic demand.
No takes yet. Share an insight, caveat, or question.
Haoning Xi (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: