Both the geometry and radiometry of SAR imagery is severely distorted in regions with terrain undulations present. Conventional geometric terrain correction (GTC) transforms image content from radar geometry to map geometry with the help of highly accurate sensor orbit state vectors and the availability of Digital Elevation Model (DEM). However, the radiometric distortion in the original SAR image still exists in the terrain corrected image, i.e. the GTC backscatter estimates contain a mixture of terrain effects and the land cover signatures. To improve the quality of the land cover signature, radiometric terrain correction is applied by modelling the effects of terrain on the backscatter and then applying compensation to normalize or flatten the backscatter amplitude. A robust technique has been developed recently that models the terrain effects by adding together all contributions of the applicable finite DEM facet areas [R-7]. After the flattening step, a single land cover class should manifest relatively consistent behaviour independent of the local terrain-slope. In backscatter amplitude-based land cover classification, the classification accuracy is impacted by terrain-induced backscatter variability and the noise level present in the backscatter estimates. With the application of terrain flattening, the terrain effects can be removed from the backscatter estimates leaving the noise the only factor affecting the classification accuracy. This opens the door for noise reduction using multi-temporal observations. The work presented below generates a composite SAR imagery from a set of multiple radiometrically terrain corrected (RTC) backscatter images and their co-registered local illuminated areas. The resulting composite backscatter image can trade off temporal resolution for improved local spatial resolution while simultaneously increasing the local number of looks. Given that the local illuminated area is inversely proportional to the local image resolution, multiple RTC backscatter estimates may be combined to advantage by weighing each individual contribution by its known local resolution. Let N be the number of RTC images and M be the number of available contributions at a given point (M will be less than or equal to N). The weight assigned to an image i within the set of M is then: (1)where Ai is the local illuminated area for the ith image at the given point. The local resolution weighted composite backscatter is then computed as the weighted sum of the flattened backscatter observations: (2)Images with the highest available local resolution receive the highest local weights. However, all available observations do contribute, reducing noise and increasing the equivalent number of looks (ENL). We will implement the composite SAR imagery generation algorithm in SNAP. Particularly, we will:1. Modify the current “Radiometric Terrain Flattening” operator to add options toa. Output the simulated images in map domain;b. Output the terrain flattened image in map domain;2. Create a new operator that willa. Take a set of multiple radiometrically terrain corrected (RTC) map domain backscatter images as input;b. Compute weight for each image;c. Create weighted average of the input images, and d. Output the map domain composite image.