Getting List elements from .reduceRegion() results in Google Earth Engine?

Question

I have a FeatureCollection that I am using .reduceRegions() over the Hansen Forest Loss dataset using ee.Reducer.fixedHistogram() to get the count of pixels per year/per feature.

The result is a feature which I can see the results in the 'histogram' property but can't select those results to store in another variable to do further analysis.

How can I select the results from the reducer (ex: how can I select 3.011764705882353 in the first element)?

I keep getting the following error:

ComputedObject (Error)
List.get, argument 'list': Invalid type. Expected: List<Object>. Actual: 
Float<dimensions=2>.

Here is the code I have:

var forest_years = forest.select('lossyear')

var test = forest_years.reduceRegions({
  collection:GB_patches,
  reducer: ee.Reducer.fixedHistogram(0,18,18),
  scale:30,
})


var first = ee.Feature(test.first())
var patch_years = ee.List(first.get('histogram'))
print (first)
print (patch_years)
print (patch_years.get(1))

Link to GEE code: https://code.earthengine.google.com/9ab9b35f4cc49b45f29386c70a0cb2cc

Answer 1

Yes, certainly it's a weir behavior. When you print(patch_years) it clearly says it's a List, but when trying to get a value it says that it is a Float<dimensions=2>, which matches with the column named 'histogram' in the FeatureCollection.

A workaround would be to bring it to the client side, and get it there:

var forest_years = forest.select('lossyear')

var test = forest_years.reduceRegions({
  collection:GB_patches,
  reducer: ee.Reducer.fixedHistogram(0,18,18),
  scale:30,
})


var first = ee.Feature(test.first())
var patch_years = ee.List(first.get('histogram'))
var patch_years_list = patch_years.getInfo()
print (first)
print (patch_years_list)
print (patch_years_list[1])

Answer 2

The output of a fixed histogram is an array with dimension 0 and 1. You should cast the output to an array and then slice over the 1-axis to get only the buckets means or counts:

var forest_years = forest.select('lossyear')

var test = forest_years.reduceRegions({
  collection:GB_patches,
  reducer: ee.Reducer.fixedHistogram(0,18,18),
  scale:30,
});

// rewrite the bucketMeans and counts from the histogram
test = test.map(function(feat){
  feat = ee.Feature(feat);
  // cast to an array and get the bucket means and counts
  var hist = ee.Array(feat.get('histogram'));
  var means = hist.slice(1, 1, 2).project([0]);
  var counts = hist.slice(1, 0, 1).project([0]); // eventually cast to a list using toList()
  return feat.set('means', means, 'counts', counts);
});

print(test);

link code. Note that I draw some geometries as a sample feature collection.