In underwater acoustic communications, spread-spectrum signals can be used to obtain frame synchronization in a data packet. However, the underwater channel is highly challenging due to phenomena such as multipath, Doppler effect and noise, so this detection can be difficult under certain conditions. With regards to noise, several sources can be present in the ocean at a certain time, providing a non-white and even non-Gaussian noise signal, as opposed to the usual consideration of additive white-Gaussian noise. In this work, a realistic ocean noise model has been developed and evaluated. This ocean noise model considers wind, rain, ships, thermal excitation of the medium and self noise as noise sources. The influence of these phenomena on the detection of several spread-spectrum signals used for frame synchronization is studied under different conditions of noise, Doppler effect, multipath and carrier frequencies. The results show how the detection performance changes noticeably depending on the noise sources and code length, suggesting that the ocean noise model can be a useful tool to aid in the design of underwater systems, prior their actual deployment.