0
print(ui.Chart.image.series(col, geometry,ee.Reducer.mean(),20).setOptions({
title: 'TimeSeries analysis',
lineWidth: 1,
pointSize: 3 }));

enter image description here

print(ui.Chart.image.series(col, geometry,ee.Reducer.mean(),200).setOptions({
title: 'TimeSeries analysis',
lineWidth: 1,
pointSize: 3 }));

enter image description here

when I set the parameter scale in different values, the chart is different(I mean at the same time while the value is different), so I want to know the meaning of the "scale" parameter and ee.Reducer parameter? I try to find the meaning in the GEE guide help, but can't understand.

Ryan M
  • 18,333
  • 31
  • 67
  • 74
XIN
  • 1
  • 1
  • Of course the chart it different, you told it to scale differently. In what way is it different is such a way that it's not just scaled? **Edit** the question and show images of the differences, so we can see what you're talking about. – Andreas Nov 05 '20 at 09:37
  • Yes ,the scale will have some difference,but I wonder the reason of value is different? – XIN Nov 05 '20 at 12:04

1 Answers1

1

Scale refers to the pixel resolution used to analyse the image. The larger the scale the more aggregated the pixels will be, thus outputting different results. Say you are analyzing a vegetation index on a plot of land. If you want to know the max value, analyzing it by pixels of 20 m by 20 m will lead to a different result than if you analysed it in pieces of 1000 m by 1000 m. The smaller the resolution, the more accurate the results, but the slower the calculation and vice-versa.

More info on this here https://developers.google.com/earth-engine/guides/scale

MadeleineN
  • 148
  • 6
  • How does the resolution of the satellite influences the size of pixels to be analysed? For example, MODIS Terra is 500m resolution and I can still choose to use scale<500 in ui.Chart, would that be wrong? What is the best practice? Thanks @Madeleine – Beto Mar 22 '21 at 01:40
  • I haven't actually verified this myself but I believe if you choose a scale lower than 500 you will get the same results for every option, thus only slowing the calculation. However going higher than 500 will change the results. – MadeleineN Mar 22 '21 at 16:30
  • In my experience the best practice depends on the precision and outcome you are looking for. Smaller scales (down to the pixel size) should be applied if possible, but for some very large study areas this might exceed GEE capacity or simply slow the code a significant amount. My advice is try smaller and increase if that doesn't work. – MadeleineN Mar 22 '21 at 16:32