The proliferation of IoT-based environmental monitoring networks has led to an exponential increase in air quality data, creating a significant challenge for effective and timely risk assessment. Traditional analysis methods often struggle to convert this vast volume of raw data into actionable insights for decision-makers. This paper introduces a novel big data framework designed to automate the process of environmental risk assessment through a two-stage analytical approach. First, we propose a Composite Hazard Index (CHI), a quantitative metric that normalizes and aggregates data from disparate pollutants into a single, interpretable risk score for each monitoring location, incorporating expert-defined weights to reflect the relative toxicity of different substances. Second, this index, along with normalized pollutant profiles, is used as input for an unsupervised clustering algorithm (K-Means) to automatically stratify locations into distinct risk tiers (e.g., Low, Medium, High). A key novelty of our approach is a descriptive analysis method that identifies the primary pollutants, or "key risk drivers," responsible for the elevated hazard levels in high-risk clusters. The framework was validated using a real-world dataset from a public air quality monitoring network. The results demonstrate the system's ability to successfully identify high-risk "hotspots," quantify their danger level, and pinpoint the specific pollutants contributing to the threat, thereby providing a powerful, data-driven tool for proactive environmental management.