This paper addresses the problem of localizing the source of an unknown scalar field distribution using a constant-speed UAV with on-board sensors in a planar environment. The scalar field intensity is assumed to decrease monotonically in all directions from the source. The UAV can sense the signal strength of the scalar field at its position at every time-instant. A novel source localization algorithm, named 'Gradient Direction Turn Switching' (GDTS), is developed in this paper, following which the UAV is made to follow circular trajectory-segments (loops) with a nominal constant turn rate and switch its turn direction as it traverses a loop and crosses the last updated gradient estimate direction. Thus, the UAV moves towards the source by estimating the gradient direction of the field and realizing a gradient ascent mechanism facilitating it to reach within a threshold neighborhood of the source. Simulation studies of the proposed algorithm in different scenarios like stationary, moving, and maneuvering sources establish its effectiveness and real-time-implementability. Updated gradient estimates over nominal loops help in improving the performance of the GDTS under noisy sensor data compared to other switched turn strategies in existing literature, which is also validated with extensive simulations studies. © 2022 IEEE.