If your imagery is georeferenced and you do not want to make bank line shape files for each, you could use transparency.

The full valley width tends to be the potential (not actual) boundary.  Soil scientists can often detect the extent of alluvial soils and colluvial soils.  As indicated there is no universal limit, but some valley types are more constrained than others.  Low gradient streams tend to be more sinuous and fine particle substrates more erosive and mobile.  Areas with excess sediment loading may fill the channel and valley so full with sediment, the channel braids.  You might find some helpful information on Dave Rosgen's website www.wildlandhydrology.com on valley and stream channel types, measuring channel and valley features.  Prehistorically, along some major rivers, you may find remnants of abandoned river terraces that were once floodplains and now abandoned.  Look over some of Rosgen's channel evolution information to get an appreciation of what I am suggesting.  And sort of the inverse of braided streams with excess sediment and valley filling are the degradation effects of clean water such as below dams or where the lands are fully vegetated and stable, where the degradation (gully like downcutting of channel) abandons point bars and floodplain.  Over time, the gully channel form may widen sufficiently and eventually rebuild a stable channel with floodplain features.  The former floodplain is now abandoned (terrace).  These features may or may not be evident with aerial photos and remote sensing, and 80 years is a short enough time that you are probably safe to sticking with valley width and not including terrace widths as long as it is recognized that river terraces are abandoned floodplains. 

Correct, there is no set meander width and geologic conditions can restrict or enhance meander formation at any point along the valley floor. Valley width is probably an incorrect indicator, depending on the type of physical environment in which the stream resides. A steep gradient would restrict lateral migration and thus meandering, while a low gradient would promote a wide meander belt. However, a stream's valley floor (or floodplain) often represents its entire history to grade, thus there could be relict courses, overflow channels, etc. (see JPG). By rule a stream's meander belt is generally about 15 times the stream's width for moderately at-grade (or near-grade) water courses. One meander area in the attached JPG showing LiDAR data along an upland mature stream, however, has a width 20X that of the stream. If good LiDAR data is available for your study area, that would be best way to more accurately assess the meander width along the stream. This can be used along with historical maps and aerials to determine channel ages if you need to restrict the data to 80 years.

Dear James

Thanks for your answer to my question. Actually i don't have LIDAR data from my study site. Anyhow, i chose the distance that is 10 times more than the average of active channel and it works:). I think its really complicated to find determine distance from center line of the river. 

Regards

Saleh

I assume that you are using/have access to a GIS program. f you are going to do an average of 10x the width(?) of the river, then all you have to do is draw a polyline down the center of the river and then have the program create a buffer for that distance. You can then do a percentage transparency for the buffered area if you want to show the land surface under it.

Good luck. JAG