This paper presents an inertial-adided vision-based localization and mapping algorithm for an unmanned aerial vehicle (UAV) that can operate in a GPS-denied riverine environment. We take vision measurements from the features surrounding the river and their corresponding points reflected in the river. We apply a robot-centric mapping framework to let the uncertainty of the features be referenced to the UAV body frame and estimate the 3D postion of point features while estimating the location of the UAV. We demonstrate the localization and mapping results with sensors on our quadcopter UAV platform in the University of Illinois at Urbana Champaign Boneyard Creek. The UAV is equipped with a light weight monocular camera, an inertial measurement unit (IMU) which contains a three-axis magnetometer, an ultrasound altimeter, and an onboard computer. To our knowledge, we report the first result of performing localization and mapping by exploiting multiple views with reflections of features in a riverine environment.