Sensitivity analysis of a moist 1D Eulerian cloud model using automatic differentiation

An automatic differentiation tool (ADIFOR) is applied to a warm-rain, time- dependent 1D cloud model to study the influence of input parameter variabil ity, including that associated with the initial state as well as physical a nd computational parameters, on the dynamical evolution of a deep convectiv e storm. Storm dynamics are found to be controlled principally by changes in model i nitial states below 2 km; once perturbed, each grid variable in the model p lays its own unique role in determining the dynamical evolution of the stor m. Among all model-dependent variables, the low-level temperature field has the greatest impact on precipitation, followed by the water vapor field. M ass field perturbations inserted at upper levels induce prominent oscillati ons in the wind field, whereas a comparable wind perturbation has a negligi ble effect on the thermodynamic field. However, the wind field does influen ce the precipitation in a more complex way than does the thermodynamic fiel d, principally via changes in time evolution. The simulated storm responds to variations in three physical parameters (th e autoconversion/accretion rate, cloud radius, and lateral eddy exchange co efficient) largely as expected, with the relative importance of each, quant ified via a relative sensitivity analysis, being a strong function of the p articular stage in the storm's life cycle.