The identification of projected changes in temperature caused by global warming at a fine-scale spatial resolution is of great importance for the high-altitude glacier and snow covered Upper Indus Basin. This study used a multimodel ensemble mean bias-correction technique which uses the ensemble empirical mode decomposition method to correct the bias of ensemble mean of seven CMIP6 GCMs outputs with reference to the European Centre for Medium-Range Weather Forecasts Reanalysis 5 (ERA5). The bias-corrected data have a nonlinear trend of seven GCMs but interannual variance and mean climate of ERA5 dataset. The dataset spans from 1985 to 2100 for historical and future climate scenarios (SSP126, SSP245, SSP370, and SSP585) at daily time intervals with a 1 km grid resolution. The result of different scenarios indicates that the increase in maximum (Tmax) and minimum temperature (Tmin) ranging from 1.5 to 5.4 °C and 1.8 to 6.8 °C from 2015 to 2100, respectively. Similarly, elevation-dependent warming is identified in Tmin from 1.7 to 7.0 °C at elevations <2,000 to 6,000 m asl, while the contrary relationship in Tmax is projected under different scenarios from 2015 to 2100. This study provides an insight into how to improve the GCMs projections and can be helpful for further climate change impact studies.