<p>Adverse weather conditions such as haze substantially impair image quality, thereby degrading the reliability of vision-based systems in critical applications, including aerial surveillance, security monitoring, and autonomous navigation. Recent advances in image restoration have been largely driven by deep learning, with dehazing networks demonstrating superior performance compared to traditional model-based approaches. Nonetheless, the progress of such methods is significantly constrained by the scarcity of real-world datasets. Existing synthetic datasets predominantly simulate haze under the assumption of spatial homogeneity, failing to capture the diversity and complexity observed in real hazy environments. This limitation ultimately leads to a synthetic-to-real performance gap. Recognizing the inherent challenges associated with collecting large-scale real hazy/clear image pairs, this paper introduces a multi-range hazing method for synthesizing homogeneous atmospheric haze. By incorporating multiple haze ranges into the synthesis process, our approach better approximates the visual diversity of real hazy scenes and thus provides a more robust benchmark for dehazing algorithms. Experimental evaluations demonstrate that hazy images synthesized with the proposed method present greater challenges to state-of-the-art dehazing models compared to those generated by conventional synthesis techniques.</p>