In a flood disaster, safe evacuation is essential to save lives. However, there have been problems such as low capacity or biases in evacuation centers for some flood disasters in Japan. Therefore, estimating the number of evacuees in each evacuation center during a disaster is vital for a safe evacuation. Consequently, it is necessary to develop highly accurate evacuation behavior simulation models. This paper proposes an evacuation behavior model using open-source data that are readily available from the internet. As a case study, the evacuation behavior model parameters and coefficients were obtained for a flood in Kuji City, Iwate Prefecture in Japan, where many people were affected by Typhoon Hagibis in 2019. By comparing the actual number of evacuees with the predicted number from the proposed evacuation behavior model and conducting a field survey of the actual evacuation shelters, we measured the validity of the evacuation behavior model and identified possible improvements.