Infrastructure-as-a-Service (IaaS) cloud providers, such as Amazon EC2, charge users based on a pay-per-use model, typically following an hourly-based pricing structure. Budget-constrained workflow scheduling in such environments remains a significant challenge, as existing strategies often lead to longer makespan and inefficient resource utilization. Although prior approaches, such as GRP-HEFT and PACP-HEFT, have made progress in addressing these challenges, they do not fully take advantage of adaptive resource provisioning strategies. This paper presents IGRP-HEFT (Improved GRP-HEFT) and IPACP-HEFT (Improved PACP-HEFT), enhanced versions of the baseline algorithms that incorporate adaptive resource provisioning. These improved algorithms introduce two distinct scenarios for resource selection: (i) a diverse resource selection strategy to optimize task execution, and (ii) a restricted selection of the fastest resource type to minimize data transfer overhead. This dual scenario approach provides greater flexibility, reducing makespan while offering the potential to decrease costs, all within strict budget constraints. Experimental evaluations on real-world workflows demonstrate that the proposed adaptive resource provisioning mechanism outperforms existing approaches, delivering improved cost efficiency and reduced makespan under hourly-based pricing models.