A place to start with data like Landsat would be to use Band 4 (Near Infrared) to differentiate water from vegetation. Water tends to be very dark in that band, while vegetation is very bright. You could try the following: - Use the Cursor Location/Value tool to look at the general DN (brightness) values for water in a scene - Use the Band Threshold to ROI tool to highlight the areas in a scene that have DN values near that of water (based on what you observed in the first step) by specifying a range of values (min and max) - With the ROI generated, you can go to the ROI Tool window for the display that contains your image and choose Options->Report Area of ROIs and choose the ROI you created. One thing to watch out for with using Band 4 is that shadows in your scene will likely have similar DN values to water, so they could end up in the thresholded ROI. If this becomes a problem, you might have to switch to using a supervised classification technique like Maximum Likelihood, Spectral Angle Mapper, or Support Vector Machine to obtain a region that covers the extent of your lake. At that point, you can turn it into an ROI and obtain an area measurement of it.

Dear Yohan You may also perform an unsupervised classification method (i.e. isodata) with 7 classes. Typically 1st and 2nd class are water bodies. Combine them in one class using envi¢s post classification tools. You may also convert the results in vector format, in order to perform a GIS analysis with other software Best regards phanis