Parameter estimation and sensitivity analysis for the modified Bartlett‐Lewis rectangular pulses model of rainfall

Abstract

This paper investigates the dependence of parameters of a stationary point process model on seasonal and spatial variability of rainfall. Twenty‐four years of hourly rainfall data for two stations from the Arno River basin in central Italy have been used to estimate parameters for each month, assuming local stationarity within the month. The estimated model parameters are physically reasonable, vouching for the model's ability to capture the general temporal and spatial structure of rainfall events. Parameter sensitivity analysis provide useful guidelines for understanding the model structure and devising effective parameter estimation strategies.