<p>Daily precipitation time series exhibit intermittent periods of high variability separated by periods of no rain, posing challenges to correct projected precipitation. To improve projected changes in probabilities of flooding and drought, it is critically important to improve temporal correlations of the precipitation time series. Previous work introduced a Time Variability Correction (TVC) method, which quantified and corrected time variability errors at differing time scales. This study extends TVC to post-process daily precipitation projections from 28 CMIP6 models over Australia, introducing a new mean adjustment procedure to eliminate negative precipitation values while ensuring that both the mean and variability of the final series aligns with the observations in the historical training period. The new TVC mean-adjusted (TVC-ma) method preserves each model’s projected change in timescale covariances, and our analysis reveals interesting differences among CMIP6 projections of changes in time-scale-dependent variances. TVC-ma is evaluated using a leave-one-out model-as-truth setup. Results reveal that, in most cases, TVC-ma significantly improves the mean, variance, lag correlations, and projections of climate indices related to persistent, heavy, and low rainfall extremes compared to raw models. When applied to future precipitation projections for Australia, TVC-ma projects pronounced increases in prolonged dry periods and maximum 1-day and 5-day precipitation amounts under the high-emission scenario relative to the low-emission scenario. Compared to the historical period, corrected projections under the high-emission scenario show drier conditions in parts of Western Australia, greater variability, extended durations of consecutive dry days and increased multi-day precipitation extremes across most regions of the continent.</p>