Land cover is a conditioning factor that affects landslide because it helps in reducing the impact of rainfall and other external factors on the slopes. For large areas, land cover can be represented through the NDVI or Land Use and Land Cover (LULC) because these factors denote areas of vegetation in an image. However, many studies on evaluating landslide susceptibility using a specific of LULC map while land cover is a time variant factor. So, LULC factor is often less important than other landslide affecting factors. In this study, we consider NDVI as a time variant factor for landslide spatial prediction. NDVI maps are derived from the optical imagery Sentinel 2 before the rainy season of each year (from 2016 to 2020) in the mountainous area of Quang Ngai Province, Vietnam. Additional to LULC and NDVI factors, the landslide events that occurred in the rainy season and other 9 landslide influencing factors are collected for landslide susceptibility assessments using XGBoost model. The results indicate that the case using time series NDVI has a better performance than the case with LULC. Therefore, this study recommends using time series NDVI data for landslide susceptibility assessment to improve the accurate prediction model.