For the past year, we have been measuring NDVI, i.e. biomass on reforested land, using images from the European Union’s Sentinel II satellite. To validate this tool, we need to ensure that the satellite measurement values match the reality observed in the field.
That’s what we wanted to find out during our latest supervisory visit. We visited 15 sites to be reforested in 2021 and 2022, some of the most extreme in terms of biomass measured by satellite: basically the best and worst sites. Of this non-representative sample, the observations on 11 plots correspond to the satellite values (70%). For the remaining 4 sites, the satellite values are too negative in 3 out of 4 cases.
Satellite measurements could therefore be used to help our beneficiaries by telling them where their land is situated, in terms of biomass, in relation to all the other land. This will enable them to take informed action, for example by enriching their plots with new seeds or wildings if the biomass is below average.
Here are a few examples of good matches:
Field 2021-034, near Oussoudougou. The good NDVI value for February 2023 in the dry season (0.211) reflects the reality on the ground:
Field 2021-015, near Diaragberela. The poor NDVI value for February 2023 in the dry season (0.156) reflects reality in the field:
Field 2021-016, near Diaragberela. The relatively good NDVI value for February 2023 in the dry season (0.203) reflects reality in the field:
Field 2021-056, near Konko. The poor NDVI value for February 2023 in the dry season (0.143) reflects the reality on the ground:
Field 2021-044, near Fansan. The poor NDVI value for February 2023 in the dry season (0.183) reflects reality on the ground:
Field 2021-031, near Talinko. The poor NDVI value for February 2023 in the dry season (0.174) reflects the reality on the ground:
Field 2022-035, near Talinko. The good NDVI value for February 2023 in the dry season (0.262) reflects the reality on the ground:
And here are the 4 cases of non-correlation:
Field 2021-004, near Linko. The NDVI value for February 2023 in the dry season (0.152) is poor and does not fully reflect the reality on the ground, which is better, perhaps due to the effect of the rainy season:
Field 2021-009, near Linko. The NDVI value for February 2023 in the dry season (0.163) is poor and does not at all reflect the reality on the ground, which is much better:
Field 2022-150, near Linko. Field observations do not correspond with the correct NDVI value for February 2023 in the dry season (0.226):
Field 2022-191, near Diaragberela. Observations in the bushy terrain do not correspond to the incorrect NDVI value (0.182):
