Exporting a global map from Google Earth Engine

Question

I want to export statistics from one of the MODIS ImageCollection to an image as I'm using these values on regular basis. The geometry associated with these datasets is the whole world meaning that on any of the images of the image collection I got:

image.geometry().coordinates().getInfo()

{'type': 'Polygon',
 'coordinates': [[[-180, -90],
   [180, -90],
   [180, 90],
   [-180, 90],
   [-180, -90]]]}

Then when I try to export it no pixels are actually selected and my image becomes blank (the js version)

ndvi = ee.ImageCollection("MODIS/061/MOD13Q1").select("NDVI")
history = ndvi.filterDate("2000-01-01", "2019-01-01")
image = history.min().rename("min").addBands(history.max().rename("max"))

ee.batch.Export.image.toAsset(
  image = image,
  description = "min-max ndvi",
  assetId = "projects/ldc-main-crop/assets/MODIS/MOD13Q1_NDVI",
  region = image.geometry(),
  scale = image.projection().nominalScale().getInfo()
)

Why can I not use the image geometry as an export region? How do I set an export parameter to keep the same footprint for my end image as the MODIS one I don't want to get a weird pole excluding -179,99999999999°, -89,999999999999°, I'm looking for the exact same footprint.

Answer 1

The issue here seems to be with the way Earth Engine handle geometry around the poles not with an image footprint (although it's still problematic).. Both of the following would not return a bounded region regardless of any image (which is fair to relief us from discussing projection)

Map.addLayer(ee.Geometry({'type': 'Polygon',
 'coordinates': [[[-180, -90],
   [180, -90],
   [180, 90],
   [-180, 90],
   [-180, -90]]]}))
Map.addLayer(ee.Geometry.Polygon([
        -180, 90, 
           0, 90 ,
         180, 90 ,
         180, -90, 
           0, -90,
        -180, -90], null, false))

The other thing is the output of image collection reducer e.g., min/max is a mosaic that has no specific scale and is in 'EPSG:4326' projection. It doesn't really have a footprint either (like an ee.Image(0): it has a world boundary but with infinite coordinates repeating itself)

image.projection().nominalScale().aside(print)

returns 111319.49079327357 image.geometry().coordinates().aside(print) returns

List (1 element) 0:
 List (5 elements) 
  0:["-Infinity","-Infinity"] 
  1:["Infinity","-Infinity"] 
  2:["Infinity","Infinity"] 
  3:["-Infinity","Infinity"]
  4:["-Infinity","-Infinity"]

If there is no specific reason to include the poles areas (you actually don't have data over the poles anyway), you can just work around this by exporting for a defined bounding box that exclude the poles and define projection and scale

Export.image.toAsset({
  image: image,
  description: "min-max_ndvi",
  region: ee.Geometry.Polygon([
        -180, 85, 
           0, 85, 
         180, 85, 
         180, -62, 
           0, -62,
        -180, -62], null, false),
  crs: 'EPSG:4326',
  scale: 231.65635826395828, //history.first().projection().nominalScale()
  maxPixels: 1e13,
})

PS check if you really need the original MODIS scale ~250 or coarse scale is acceptable for global image, my try took >1 h to export min/max