Accurate mapping of actual irrigated cultivated land is essential for understanding the interactions between human agricultural activities and regional agricultural ecosystem, particularly in large basins with strong climatic and ecological heterogeneity. As a typical arid and semi-arid region, the Yellow River Basin (YRB) is facing prominent contradictions between agricultural water use and ecological conservation, making long-term irrigation monitoring a critical indicator of agricultural water use for assessing basin-scale agricultural health and sustainable development. This study developed a basin-scale framework for long-term irrigation monitoring in the YRB from 2002 to 2020 by improving the LST–NDVI feature space and integrating the Temperature Vegetation Dryness Index (TVDI) with precipitation constraints. Quadratic polynomial dry and wet edge equations were constructed for different climatic zones to better represent nonlinear LST–NDVI relationships. Moreover, TVDI derived from MODIS data was then combined with CHIRPS precipitation using a 7-day cumulative precipitation threshold to distinguish irrigation-induced wetting signals from rainfall-driven soil moisture anomalies. Compared with linear dry–wet edge fitting, the quadratic fitting achieved statistically significant improvement ( p < 0.01) in boundary estimation, with an average R 2 rise of 18.6% and overall R 2 above 0.7 across different climatic zones. The extracted irrigated area exhibited strong agreement with prefecture-level statistical records. Basin-wide irrigated area remained relatively stable during 2002–2022, fluctuating between approximately 6500 × 10 3 hm 2 and 7800 × 10 3 hm 2 , while moderate expansion occurred mainly in marginal upstream and midstream regions. The irrigated areas of the Hetao and Qingtongxia Irrigation Districts in the upstream both exhibited a significant, continuous upward trend over the past 20 years, with annual growth rates of 11.37 and 2.08 × 10 3 hm 2 , respectively. The irrigated area of the Weishan Irrigation District in the downstream displayed an upward trend with stronger interannual variability driven by precipitation fluctuations. The proposed framework provides a scalable and physically interpretable approach for long-term irrigation monitoring in large heterogeneous basins. It offers reliable technical support for agricultural water allocation and sustainable water resource management in ecologically fragile basins under climate change.