Plant phenology provides an integrative biometeorological indicator of how agroecosystems respond to climate variability and long-term change. The Lanzhou lily (Lilium davidii var. unicolor), a key edible-medicinal bulb crop in Northwest China, is cultivated in a temperate semi-arid region that has experienced pronounced warming and a concurrent decline in sunshine duration over recent decades. Using continuous field phenological observations (1995-2024) and daily meteorological records, we quantified long-term trends in four phenological metrics (emergence, full bloom, shoot senescence, and growing season length) and identified phase-specific climatic sensitivity windows via a sliding-window correlation approach. We then parameterized multiple regression models using the optimal windows to quantify the relative roles of temperature, sunshine duration, and precipitation. Over the 30-year period, emergence, full bloom, and shoot senescence advanced significantly (1.06, 1.23, and 1.69 days per decade, respectively), while growing season length shortened (0.67 days per decade). Temperature exerted a consistent advancing effect across all phases, whereas sunshine duration acted in the opposite direction by delaying full bloom and senescence. Precipitation contributed a minor, late-season influence on growing season length. The identified response windows (from early spring to late summer) provide a practical basis for biometeorological monitoring and phenology forecasting of this perennial bulb crop under ongoing climate change.