Marine environmental surveys using unmanned surface vessels (USVs) can be a challenging task, especially if the surveyed area takes several weeks to cover. There is also the constant risk of depleting the battery before the mission is completed, which is associated with the challenge of vehicle power management. Thrusters in unmanned vehicles are the main power drainers. Waves, currents, and wind unpredictable behavior have a great influence on the motion of the vehicle and, hence, affect whether the vehicle is to be able to fulfill its mission in the allocated time. The primary objective of the present research is to design an algorithm that optimize USV's power consumption by predicting the amount of power devoted to the thruster as a function of time. Thruster power predictions were performed by a genetic algorithm that uses battery, speed, solar power, and wave height as well as wave period information to forecast power supply and demand. Simulation results showed that the proposed algorithm outperforms a human pilot in reducing thruster power utilization per unit distance by 17%. Despite stochasticity in the search process of the genetic algorithm, the produced results were semi-consistent and satisfy mission objectives as well as constraints. With limited battery resources, the algorithm was able to produce a plan that never dropped battery level to zero and maintained energy above 80Wh, about 51% more energy than a human pilot in a real mission.