Elevation change detection in GEE

Question

I am trying to create a function in Google Earth Engine which would iterate through an ImageCollection consisting of DEMs and return a message saying when a big elevation difference occurs. I believe I am not far from the answer, but I can't manage to make it work.

To what I understand, when you subtract one image from another it does so on pixel-by-pixel basis and returns an image. If you want elevation changes at a specific point, a geometic point can be defined. But what if you want the elevation difference to be calculated on the whole DEM and if at any location a big elevation change occurs, the function detects it? Do you need to explicitly define the region of interests, even though the DEMs themselves act as roi?

Script:

function anomaly(image) {
  // Converting the ImageCollection ''newdems'' to list
  var Imagelist = newdems.toList(newdems.size());
  // Defining the previous image
  var previous = ee.Image(Imagelist.get(-1));
  // Equation to calculate the elevation difference 
  var difference = image.subtract(previous);
  // Conditional statement to return the name of the image after which the big elevation change occurs
  return ee.Algorithms.If(difference.gt(150),print('Anomaly occurs after: ', previous),ee.Image(0));
};

Answer 1

Answering your specific questions: Do you need to explicitly define the region of interests, even though the DEMs themselves act as roi? Yes. I think you should reduce the difference image with its boundingbox, searching for a maximum value of change. for example:

var difference = image.subtract(previous);
var roi = difference.geometry();
var numDiff =  difference.reduceRegion({reducer: ee.Reducer.max(),
                                       geometry: roi,
                                      scale: 250,
                                      bestEffort: true});

But i think there are other problems in your function. For example your previous image will always be the same... If your idea was to substract image 2 with image 1, image 3 with image 2 etc, the code below may work. Its an example with ndvi scenes. (i wish i had a dem time series!)

var col = ee.ImageCollection("MODIS/MOD13A1").filter(ee.Filter.date('2014-04-01', '2014-08-01'))
                                             .select('NDVI')
                                             .map(function(img){
                                               return img.clip(geometry)})
var list = col.toList(col.size());


var indices = ee.List.sequence(0,col.size().subtract(2), 1)

var colDifference = indices.map(function(num){
  num = ee.Number(num)
  var current = ee.Image(list.get(num));
  var add = num.add(ee.Number(1));
  var post = ee.Image(list.get(add))
  var imgDiff = post.subtract(current);
  var numDiff =  imgDiff.reduceRegion({reducer: ee.Reducer.max(),
                                       geometry: geometry,
                                      scale: 250,
                                      bestEffort: true});
  
        return ee.Feature(null, {'diffMax': numDiff})
        
}).map(function(feat){
  return ee.Dictionary(ee.Feature(feat).get('diffMax')).get('NDVI')
});


var clientsideList = colDifference.getInfo()


///////////////////CLIENT SIDE
for(var i = 0; i < 8; i++){
  if (clientsideList[i] > 1100){
          print('anomaly')}
        else {
      print('tutto a posto')
}}

https://code.earthengine.google.com/?scriptPath=users%2Fjjfoguet%2Fdefault%3Asubtracting_consecutive