If understand your question correctly, the signal from the water surface is higher (than expected) and as all radar images show speckle the water surface shows also speckle? If it is like that, it could be due the wind condition at the time of image acquisition, because the wind causes the water surface to be rougher which results in a higher signal return to the sensor. This is well known phenomenon when monitoring water surfaces (becomes even more relevant when you want to monitor oil spills). Let me know if I understood your question wrongly.

For future analysis I would suggest a polarimetric analysis. Maybe vegetation introduces some double bounce backscatter (if high enough), while the signal from the wind ripples (if..) is more a surface scattering.

Hope I could help. 

If understand your question correctly, the signal from the water surface is higher (than expected) and as all radar images show speckle the water surface shows also speckle? If it is like that, it could be due the wind condition at the time of image acquisition, because the wind causes the water surface to be rougher which results in a higher signal return to the sensor. This is well known phenomenon when monitoring water surfaces (becomes even more relevant when you want to monitor oil spills). Let me know if I understood your question wrongly.

For future analysis I would suggest a polarimetric analysis. Maybe vegetation introduces some double bounce backscatter (if high enough), while the signal from the wind ripples (if..) is more a surface scattering.

Hope I could help. 

Thank you for your answer Christoph Hütt.

Yes, you understood the question correctly.

Indeed I considered the hypothesis of the wind influence on the water surface. I searched for past weather bulletins and in most cases the wind was relatively low (10km/h), I also looked for screening effects due to the land, but with few success. Nevertheless it is a possible explanation.

Thank you for the suggestion about using polarimetric analysis, I'll do it.

Dear Giulio, Speckle is a multiplicative noise which arises from backscattering of the original wave. This is an area of research where a lot has been done. Our group has been working with speckle noise in ultrasound imaging, which has the same modelling as the SAR speckle, for more than 20 years. See also our Books published. Your question is very and valid and cannot be answered that easily. If it was easy to answer your question then the problem of speckle noise interference would be solved. Unfortunately the answer is not clear. The speckle interference may be caused from different factors, ranging from environmental conditions, climate, insonating waves, setting parameters of the transducers, time of the wave travel and many more. What can be at a first stage done is to perform despeckle filtering on the images acquired, for removing some of the noise component by choosing the appropriate filter (see our book). Then you may perform texture analysis on selected areas and based on the extracted texture features you may distinguish or separate ground with water or other different objects of interest. The best way to approach the problem is to select a large number of images acquired under the same settings. Then you may perform a study using filters and then ask observers to evaluate the images. You may also extact features and quality metrics from the original and the depsekled images (see our publications for methodology). Thus you will be able to find the most appropriate filter/s that can be used for your case. I hope this will help. Good luck with your research.

Even small winds can cause waves in such shallow waters, and I agree with Christoph Hütt's explanation that these might be radar-facing slopes of waves. In my experience with coastal waters, weather bulletins, even if taken from a weather station close enough, will still show significant deviations from the local conditions. 

One way ahead would be to check the possible values of winds (not only from the weather bulletins, but by how much it has been known to deviate in this area). For the wind speed quoted here, Bragg scattering would account for 100% of the backscattering at angles higher than 20 degrees, and still around 10-20% at 0 degrees. I checked the data in Ian Robinson's book "Measuring the Oceans from Space" (Volume 1), published with Springer in 2004. More complex formulae can be used to account for the exact frequency of the radar, its polarisation, its azimuth relative to the upwind direction etc. This book (Chapter 9) shows examples with different radars, and also points to the CMOD4 wind-backscatter transfer function of Stoffelen and Anderson (1997). 

This might help in assessing whether the problem is going to reoccur in other images.

Good luck with the analyses!