Often for water cycle quantities one gets a better impression by plotting the changes relative to the model climatology. The assumption is that if the model has a strong bias in the mean state it will have similar bias in the climate change signal and the relative change is then a better representation of the signal than the absolute numbers. The relative changes are always plotted in percentages. Note that for fields that include zero, such as P-E, the relative changes may become very large and should be considered with great care in the regions where the climatology is near-zero.
Temperature and pressure fields are never plotted as relative changes.