Hydrological forecasting is a critical component in addressing challenges in drought mitigation, water resources management, hydropower generation, and flood control. The increasing complexity brought on by climate change has underscored the need for more accurate and reliable forecasting methods. Hydrological models, which rely on skillful meteorological forecasts as inputs, are essential tools in achieving this goal. However, raw ensemble forecasts, which are a common input, tend to be biased and under-dispersed, necessitating post-processing to enhance their utility. Member-by-Member post-processing (MBMP), a method that undertakes bias and dispersion correction, emerges as a promising approach. Unlike other methods that may disrupt the spatiotemporal dependencies within the ensembles, MBMP strives to preserve these critical features. Traditionally, MBMP variants have relied on regression for bias correction, a technique that does not fully eliminate conditional type-1 bias. This abstract introduces a novel variant of MBMP that seeks to improve forecast quality by focusing on maximizing reliability during the bias-correction process. This innovative approach has been tested on temperature forecasts with lead-times of 1-10 days, demonstrating superior performance in terms of reliability and accuracy when compared to existing MBMP variants.