The DP3 model of leaf coloring, formulated according to the leaf development process, considerably contrasts previous models and allows to set up new hypotheses, e.g., regarding earlier onset and longer duration of senescence predicted for warmer conditions. Comparing the accuracy of the DP3 model to that of previous models and the Null model (average observed date of leaf coloring) suggests that leaf coloring data are noisy, which is why observation protocols and methods should be revised.