Abstract: | Abstract The probability-distributed catchment model, as originally proposed by Moore &; Clarke (1981), is re-examined from a maximum statistical entropy viewpoint. The distribution of water within a catchment is treated as a problem of statistical inference and resolved using an entropy maximization technique. A simple runoff generating mechanism is employed, which, together with the catchment mass balance equation, yields a catchment model involving just one dynamic parameter, y, and two constants, k and λ. The parameter y determines the temporal variation of catchment storage V and runoff q. The latter is nonlinearly related to V through q = k(1—λyV), where y provides the nonlinear departure from the simple linear reservoir q = kV. |