Google Earth Engine（GEE）——S2影像异常值

在计算索引并生成高质量马赛克时，有一个明显的图像是异常值。我已经确定了图像，问题在于 B5 波段的值非常低，这使得索引具有非常高的值。

编号：哥白尼/S2_SR/20200829T174909_20200829T175522_T13UFP

代码：

var imageCollection = ee.ImageCollection("COPERNICUS/S2_SR"),
    geometry = 
    /* color: #98ff00 */
    /* shown: false */
    /* displayProperties: [
      {
        "type": "rectangle"
      }
    ] */
    ee.Geometry.Polygon(
        [[[-103.99506389984352, 48.962358795597474],
          [-103.99506389984352, 46.88665586989701],
          [-99.68842327484352, 46.88665586989701],
          [-99.68842327484352, 48.962358795597474]]], null, false),
    geometry2 = /* color: #0b4a8b */ee.Geometry.Point([-102.87049298795543, 48.22935423511586]),
    geometry3 = /* color: #00ffff */ee.Geometry.Point([-101.57471425064051, 48.22316284019277]);

//get veg bare soil water
function maskSCL(image) {
  var scl = image.select('SCL');
  var mask = scl.gte(4).and(scl.lte(6));
  return image.updateMask(mask);//.divide(10000);
}

//calculate red edge chlorophyll index
var getIndices = function(img){
  var im = img.divide(10000);
  var CIre=im.select('B7').divide(im.select('B5')).subtract(1).rename("CIre");
  return im.addBands(CIre).copyProperties(img,['system:time_start']);
}


function addDOY(im){
  var doy = im.date().getRelative('day', 'year');
  var doyBand = ee.Image.constant(doy).uint16().rename('doy')
  doyBand = doyBand.updateMask(im.select('B8').mask())
  
  var yr = im.date().get('year')
  var yrBand = ee.Image.constant(yr).uint16().rename('year')
  yrBand = yrBand.updateMask(im.select('B8').mask())
  
  return im.addBands(doyBand).addBands(yrBand)
}


//max CIre from July-Sept
var dataset1 = imageCollection
.filterBounds(geometry)
.filter(ee.Filter.calendarRange(2018, 2021, 'year'))
.filter(ee.Filter.calendarRange(7, 9, 'month'))
.map(maskSCL)
.map(getIndices)
.map(addDOY)
.select('CIre', 'doy', 'year');

var CIre_max = dataset1.qualityMosaic('CIre');

Map.addLayer(CIre_max.select('CIre'),
{min:0,max:20,palette:['black','indigo','cyan','limegreen','yellow']},'CIre')
Map.addLayer(CIre_max.select('doy'),
{min:183,max:274,palette:['black','indigo','cyan','limegreen','yellow']},'doy')
Map.addLayer(CIre_max.select('year'),
{min:2018,max:2021,palette:['indigo','cyan','limegreen','yellow']},'year')


// var index = dataset1
// .select('CIre')
// .reduce(ee.Reducer.percentile([98]))

// //calculate band to use for quality mosaic of 98th percentile values
// // loop over image collection
// //get absolute difference from 98th percentile
// //convert to 1/abs dif + .1 to use as band for quality mosaic
// var forQM  = dataset1.map(function(x){             
//   // get absolute difference of the percentiles with the CI
//   var selector = x.select('CIre').subtract(index).abs();
//   var selector_fixed = selector.eq(0).multiply(0.1).add(selector)
//   // invert to let the min diff be the largest value
//   var invsel   = ee.Image.constant(1).divide(selector_fixed);
//   // add inverted difference band to the image and name properly
//   return x.addBands(invsel.rename('selector_CIre'))   
// });

// // // // use the new selectors to perform the quality mosaic and add to map
// var CIre_p98QM  = forQM
// .qualityMosaic('selector_CIre')
// .select(['CIre', 'doy'])

// Map.addLayer(CIre_p98QM.select('CIre'),
// {min:0,max:20,palette:['black','indigo','cyan','limegreen','yellow']},'CIre_p98')
// Map.addLayer(CIre_p98QM.select('doy'),
// {min:183,max:274,palette:['black','indigo','cyan','limegreen','yellow']},'doy_p98')


///print trouble image
//ID COPERNICUS/S2_SR/20200829T174909_20200829T175522_T13UFP
//Think B5 is not right
var checkit = imageCollection
.filterBounds(geometry2)
.filter(ee.Filter.calendarRange(2020, 2020, 'year'))
.filter(ee.Filter.calendarRange(240, 242, 'day_of_year'))
print(checkit)
Map.addLayer(checkit.max(),{bands:['B4','B3','B2'], min:100,max:2000},'trouble_image')
Map.addLayer(checkit.max(),{bands:['B5'], min:100,max:2000},'trouble_b5')

///nearby image same time period...B5 better
var checkit = imageCollection
.filterBounds(geometry3)
.filter(ee.Filter.calendarRange(2020, 2020, 'year'))
.filter(ee.Filter.calendarRange(240, 242, 'day_of_year'))
print(checkit)
Map.addLayer(checkit.max(),{bands:['B4','B3','B2'], min:100,max:2000},'nearby_image')
Map.addLayer(checkit.max(),{bands:['B5'], min:100,max:2000},'nearby_b5')