For the answer to why there is wind, see: Where does wind come from?
That establishes the forces at work driving the wind, namely:
- non-linear advection of momentum
- the pressure gradient force (PGF)
- the Coriolis force
Friction is less over water than over land, and so the wind will tend to flow more parallel to the isobars than they will over land at the earths surface. The force balance between the PGF, Coriolis and friction is easy to conceptualize:
Image taken from University of Illinois at Urbana-Champaign, credits at http://ww2010.atmos.uiuc.edu/(Gh)/abt/aknw/dvlp.rxml
In this we can see that see that wind will flow with low pressure to its left, and cross the isobars slightly toward the low pressure. The angle of the wind crossing the isobars is related to friction. The direction of the wind is related to the orientation of the isobars and friction. The speed of the wind is related to the spacing between isobars (close = fast, far = slow).
However, this only explains a steady-state wind in equilibrium, which is great for theoretical explanations but is too idealized for the real atmosphere. In your observations on the lake, the flow is not steady-state and that pesky non-linear term is at work.
The lowest level of the atmosphere is called the boundary layer, and this extends from the surface to roughly 1 - 2 km. During the daytime this boundary layer is "well mixed", which just means it has a uniform potential temperature and moisture profile and it is turbulent. To really over-simply turbulence, consider swirling motions of different sizes. At the largest sizes these swirls are the size of the depth of the boundary layer. At the smallest sizes the swirls are on the scale of millimeters, and there are swirls at all of the intermediate sizes.
Let us just consider the biggest of the swirls. The top of these swirls are at the top of the boundary layer where the wind is typically faster (it can be slightly faster, or much faster). This swirl will mix some of this momentum down to the surface, at the bottom of the swirl. These swirls are moving around as well, not sitting in one place. Where the swirls are, there will be a different wind velocity than where they arent, and it will depend on the orientation of the wind, the swirl and the momentum being mixed downward to the surface. If you were standing at one place, this would give you the effect you observe, winds that change direction and/or speed.
Now consider all of the different sizes of eddies doing this, and things become a mess (turbulence!). On some days the this effect is small and you might just notice subtle changes in the wind, and on other days you might have strong gusts and "bumpy" air.
Finally, lets get back to the lake. The lake has a temperature flux and a momentum flux with the air above it, which produce changes in the wind. All of this wind is moving mass around and the net result of this is that the wind is causing the pressure field to evolve, and these changes feedback into the wind.
The above effects can be rolled into the non-linear and friction terms in the momentum equation, which are the reasons that your observations of the wind periodically change directions.