Drought poses a major risk to Taiwan’s industries, especially semiconductor production. However, Taiwan’s rugged terrain makes it difficult to measure land surface temperature (LST), which is crucial for drought monitoring. In this study, we use machine learning to improve LST accuracy by correcting it with surface condition data. Based on the improved LST, we develop a new drought index called the Surface Water Availability-Temperature Index (SWATI). We tested SWATI across Taiwan from 2001 to 2023 and found it performed better than other drought indices, especially in mountainous areas. SWATI offers a simple and effective tool for tracking drought and supporting water management in complex landscapes.