Transportation agencies report the localization of roadway anomalies that could cause serious hazards to the traveling public. However, the high cost and limitations of present technical prevent scaling the road monitoring to all roadways. Especially the unpaved road, because of the complexity of unpaved road. Using smartphone application as road condition data collection tool offer an attractive alternative because of its potential to monitor all roadways in real time and its low cost. However, the sensor sensitivity and sampling frequency of different smartphones may vary significantly, which challenge the confidence of using smartphones for actual pavement condition assessment applications. This study tends to solve this challenge by calibrating different smartphones using two different calibrating methods including calibrating towards reference or average road roughness.