Airborne gravimetry represents nowadays probably the most efficient technique to collect gravity observations close to the Earth's surface. In the 1990s, thanks to the development of the Global Navigation Satellite Systems (GNSS), which has made accurate navigational data available, this technique started to spread worldwide because of its capability to provide measurements in a fast and cost-effective way. Differently from other techniques such as shipborne gravimetry, it has the advantage to provide gravity measurements also in challenging environments which can be difficult to access otherwise, like mountainous areas, rain forests and polar regions. For such reasons, airborne gravimetry is used for various applications related to the regional gravity field modelling: from the computation of high accurate local geoid for geodetic applications to geophysical ones, specifically related to oil and gas exploration activities or more in general for regional geological studies. Depending on the different kinds of application and the final required accuracy, the definition of the main characteristics of the airborne survey, e.g., the planar distance between consecutive flight tracks, the aircraft velocity, etc., can be a difficult task. In this work, we present a new software package, which would help in properly accomplishing the survey design task. Basically, the developed software solution allows for generating a realistic (from the observation noise point of view) gravimetric signal, and, after that, to predict the accuracy and spatial resolution of the final retrievable gravimetric field, in terms of gravity disturbances, given the flight main characteristics. The proposed procedure is suited for airborne survey planning in order to be able to optimize the design of the survey according to the required final accuracy. With the aim to evaluate the influence of the various survey parameters on the expected accuracy of the airborne survey, different numerical tests have been performed on simulated and real datasets. For instance, it has been shown that if the observation noise is not properly modeled in the data filtering step, the survey results degrade about 25%, while not acquiring control lines during the survey will basically reduce the final accuracy by a factor of two.