I am attempting to use USGS Landsat 7 Level 2, Collection 2, Tier 1 data to calculate NDVI. My first step is to try to create a cloud-free composite image using ee.Algorithms.Landsat.simpleComposite

I’m trying to calculate mean NDVI for the American Southwest from June through September of 2003.

I am trying to calculate an average NDVI layer from June through September 2011 using Landsat 7 data. I’ve been trying to remove the noise lines associated with Landsat 7 data after 2003 using data provided from a similar question. However, this does not seem to be cleaning up my lines. I think the issuse stems from the fact that the example code from the original solution removes the noise lines by first sorting all landsat images within the data by ascending cloud cover, and then selecting the first image in the data (.sort('CLOUD_COVER', true) .first()), which appears to remove all data from the noise lines, though I’m not entirely sure how. The missing data is then filled in using the focal_mean() and blend() functions.

I want to open up a discussion that I can’t seem to find a lot about regarding how data are processed for individuals. When using a software like GEE, with lots of data availability and various products readily available, do people tend to do their own corrections or use what is already available? It seems like for convenience’s sake, one should just use what is already available in the catalog. But the staunch programmer may want to know exactly how the data are processed, and have a larger body of data to pull from initially.