Recently, green tides dominated by Ulva prolifera have had a major impact on coastal ecosystems of China, as a result of the algae's high surface area to volume ratio, high rate of nutrient uptake, low nutrient half-saturation coefficient, and restriction of other algae. Here, a U. prolifera growth model was developed to analyze the growth process and key limiting factors of U. prolifera. To date, ecological modeling related to green macroalgae has mainly concentrated on variability in adult plant biomass; in this study, biomass of early life stages and adult plants were considered separately to clarify the growth pattern of U. prolifera. In addition, nutrient limitation and temperature functions in the growth model were adapted to the environment of coastal China. The model was established with STELLA, and the simulation results reveal the growth pattern of U. prolifera. Parameter uncertainty is the basis of model uncertainty and thus should be assessed; parameter optimization based on sensitivity analysis could improve the precision of the growth model. Sensitivity analysis is also an important tool to improve marine ecological models. In this study, global sensitivity analysis of the Morris method based on statistical sampling was implemented in the U. prolifera growth model. Compared with local sensitivity analysis methods, global sensitivity analysis has the advantage of assessing correlations among parameters and of analyzing the sensitivity of all parameters simultaneously. Compared with other global sensitivity analysis method, the Morris method is efficient. Nevertheless, adequate sampling repetition must be performed. Parameters are sampled from the entire defined domain and are ranked according to the mean and standard deviation of elementary effects to assess their global sensitivity and qualitative correlation. Thus, parameter optimization strategies can be established to improve the precision of high-ranking sensitive parameters while insensitive parameters are used as empirical values. Here, the parameters were sampled 10,000 times to reduce random error, and the mean and standard deviation of elementary effects were ranked with radar graphs. The sensitivity analysis results showed that optimum temperature for growth (T_opt), optimum light intensity for photosynthesis (Is), maximum germination rate (G_max), nitrogen half-saturation constant for growth (kqn), and maximum nitrogen uptake rate (V_maxn) were sensitive in the growth model, which meant that U. prolifera is mainly limited by temperature, light intensity, and nitrogen. The precision of these five parameters should be improved by further parameter optimization; maximum growth rate (μ_max) and reproduction rate-biomass lost by sporulation (Reprod_rate) were not sensitive and could be kept as empirical values. Local and global sensitivity analyses were compared, which revealed that global sensitivity analysis was much reliable because the sensitivity results were not affected by the initial parameter values. Correlation analysis showed that the sensitive parameters were correlated with other parameters.