Specifying the 'CROP' field value in reference polygons

General Discussions
1

Hello,

I generated an L4A product (with in situ data) by indicating my crops as ‘1’ in the ‘CROP’ column of my reference polygon’s attribute table. All in all, the said training data has 3,900+ features (including the non-crop features)

Then I will be using the said L4A data as my mask for my L4B. My question is, will I use the same polygons I uploaded for L4A product creation? Or should I remove the non-crop features and retain only the crops?

For the latter option, will it affect the L4B result since the quantity of features in the reference polygons will decrease?

Regards.

2

Then I will be using the said L4A data as my mask for my L4B.

This is done automatically by the processor, although you probably know that already.

My question is, will I use the same polygons I uploaded for L4A product creation? Or should I remove the non-crop features and retain only the crops?

You can use the same file. The features with CROP=0 will be ignored.

3

Actually I’m using the custom jobs, so I manually choose the mask.

Anyway, I found out that using the same kind of vector polygon (with values of ‘0’ in CROP column), it has a decreased accuracy and there was an overestimation of maize. So I made the features of my vector polygons as CROP = 1, and it resulted to a more acceptable product.

Regards.

4

Oh, okay.

Anyway, I found out that using the same kind of vector polygon (with values of ‘0’ in CROP column), it has a decreased accuracy and there was an overestimation of maize.

That’s… unexpected. The features with CROP=0 should be ignored, as far as I know.