Table 2 Agreement between 11 compared maps as percent of total pixels.

From: How accurate are existing land cover maps for agriculture in Sub-Saharan Africa?

Country

All predict same class

All predict crop

Split prediction

None predict crop (all non-crop)

Kenya

60.6

0.4

5.0

60.2

Malawi

21.8

0.0

17.5

21.8

Mali

69.9

0.0

2.9

69.9

Rwanda

15.8

0.1

24.0

15.8

Tanzania

44.4

0.0

9.0

44.4

Togo

23.8

0.2

17.0

23.6

Uganda

29.1

0.4

11.6

28.8

Zambia

49.5

0.0

4.4

49.5

  1. “Split prediction” describes pixels where the prediction across all maps is approximately evenly split (5 predict crop but 6 predict non-crop, or 6 predict crop but 5 predict non-crop).