EGU22-7260, updated on 28 Mar 2022
EGU General Assembly 2022
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

An integrated method for validating spatially non-continuous remotely sensed forest patch dataset for West Africa

Chima Iheaturu, Vladimir Wingate, Felicia Akinyemi, and Chinwe Ifejika Speranza
Chima Iheaturu et al.
  • University of Bern, Institute of Geography, Land Systems & Sustainable Land Management (LS-SLM), Bern, Switzerland (

Remote sensing products of medium to high spatial resolution have emerged as promising datasets for environmental modelling and policymaking across scales. Despite the recent increase in their availability and accessibility,  questions often remain on how to best assess the accuracy of these products, since it is pivotal that these be rigorously validated before they are used for scientific investigation and decision making. There are several methods for validating spatially continuous remote sensing-derived products, including comparisons to field surveys, cross-comparisons and verification of physical consistency using reference data. However, there exist few or no validation strategies for validating spatially non-continuous products such as forest patches. In effect, forest patches, as with many other thematic maps, contain information that is discrete, not spatially continuous, and not normally distributed; thus, a validation strategy that makes these assumptions may be inappropriate for such a product. 

We present an integrative approach for assessing the accuracy of a remote sensing-derived product identifying forest patches found within agricultural landscapes of West Africa. The method is based on the well-established error matrix approach and uses a spatial sampling strategy that determines the sample size based on spatial autocorrelation, select sample points based on spatial uniformity and heterogeneity, and assesses the accuracy by comparing sample points and reference data. Compared to other random sampling approaches, ours ensures that a representative sample is used for the accuracy assessment. This representativeness was achieved by utilizing a stratification method that enabled different categories of forest patches across different ecoregions in the map to be included in the sample size. 

While further tests are required, the preliminary results show that our method has the potential to effectively assess the accuracy of forest patches in West Africa and can therefore be adapted for validating other spatially non-continuous remote sensing products.

Keywords: Remote sensing products, Validation, Spatially non-continuous, Error matrix, Spatial sampling, Forest patch, West Africa

How to cite: Iheaturu, C., Wingate, V., Akinyemi, F., and Ifejika Speranza, C.: An integrated method for validating spatially non-continuous remotely sensed forest patch dataset for West Africa, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7260,, 2022.


Display file