How to remove duplicate Images from a Landsat8 Image Collection

Question

I am very new to Google Earth Engine and coding.

My goal is to run statistical analysis on a study & control area looking at Vegetation productivity loss across a time period.

I downloaded the ee.ImageCollection("LANDSAT/LC08/C02/T1_TOA") into my script and renamed it L8Data

I created two polygons for my study area var Somalia and Var Kenya

// Filter Landsat 8 Data temporally and spatially;
 
var SomaliaL8 = L8Data.filter(ee.Filter.bounds(Somalia));


var KenyaL8 = L8Data.filter(ee.Filter.bounds(Kenya));

var L8Som2013 = SomaliaL8.filter(ee.Filter.date('2013-04-01', '2013-06-30'));

var L8Ken2013 = KenyaL8.filter(ee.Filter.date('2013-04-01', '2013-06-30'));

var L8Som2021 = SomaliaL8.filter(ee.Filter.date('2021-04-01', '2021-06-30'));

var L8Ken2021 = KenyaL8.filter(ee.Filter.date('2021-04-01', '2021-06-30'));

I want a composite cloud free image for each of those time periods for both defined areas.

So I used this code below i found in a youtube video.

var S2013 = ee.Algorithms.Landsat.simpleComposite({
  collection: SomaliaL8, 
  asFloat: true
  });

var TCV = {min: 0, max: 0.3, bands: ['B4', 'B3', 'B2']};  

 Map.addLayer(S2013.clip(Somalia, TCV, 'S2013'));

 Map.centerObject(Somalia, 10);

I get the error code

Layer 1: Layer error: ImageCollection.reduce: Error in map(ID=LC08_165060_20130409): Landsat.TOA: Band 'B1' is Type, expected Type<Integer<0, 65535>>.

Also when I go to my console and look at the features in the image collection I made there should only be 6 images per year for my filtered dates but I have between 15-20 in each one. They look like replicates repeating the same dates over and over. This has given me hours of headaches.

I want to remove the duplicates and make my cloud free composites so I can finally run NDVI, NPP, GVI, & RDI on the images (probably in ENVI).

Answer 1

SimpleComposite computes TOA, you can't pass it TOA data. You need to use the raw collection instead:

ee.ImageCollection("LANDSAT/LC08/C02/T1")

Or, use your own compositing, like a median. SimpleComposite is pretty simple.