Identifying Biases in Global Tree Cover Products: A Case Study in Costa Rica

Abstract

Global tree cover products are widely used in analyses of deforestation, fragmentation, and connectivity, but are rarely critically assessed. Inaccuracies in these products could have consequences for future decision making, especially in data-poor regions like the tropics. In this study, potential biases in global and regional tree cover products were assessed across a diverse tropical country, Costa Rica. Two global tree cover products and one regional national forest cover map were evaluated along biophysical gradients in elevation, precipitation, and agricultural land cover. To quantify product accuracy and bias, freely available high-resolution imagery was used to validate tree and land cover across these gradients. Although the regional forest cover map was comparable in accuracy to a widely-used global forest map (the Global Forest Change of Hansen et al., also known as the GFC), another global forest map (derived from a cropland dataset) had the highest accuracy. Both global and regional forest cover products showed small to severe biases along biophysical gradients. Unlike the regional map, the global GFC map strongly underestimated tree cover (>10% difference) below 189 mm of precipitation and at elevations above 2000 m, with a larger bias for precipitation. All map products misclassified agricultural fields as forest, but the GFC product particularly misclassified row crops and perennial erect crops (banana, oil palm, and coffee), with maximum tree cover in agricultural fields of 89%–100% across all crops. Our analysis calls into further question the utility of the GFC product for global forest monitoring outside humid regions, indicating that, in tropical regions, the GFC product is most accurate in areas with high, aseasonal rainfall, low relief, and low cropland area. Given that forest product errors are spatially distributed along biophysical gradients, researchers should account for these spatial biases when attempting to analyze or generate forest map products.

A lesson unlearned? Underestimating tree cover in drylands biases global restoration maps

“…In the case of study 2, their methodological error was using the Hansen et al (2013) data to estimate existing tree cover in drylands. Bastin, Berrahmouni, et al (2017) Cunningham, Cunningham, and Fagan (2019) found that the HGFC product begins to dramatically underestimate tree cover below approximately 2,270 mm of mean annual precipitation (e.g., tropical dry forest). In this study, I show that the HGFC product underestimates tree cover even at very low cover in hyperarid systems (Figure 3), and that this bias in drylands has a large cumulative effect on global canopy cover estimates.…”

“…(2019) found that the MODIS VCF product underestimated tree cover by 5.6% in shrublands and savannas, very close to the 5.9% bias observed in this study for the HGFC product. This particular challenge likely arises from multiple causes (Cunningham et al., 2019; Sexton et al., 2015; Smith et al., 2019). First, it may be caused by intra‐ and inter‐annual variability in tree and herbaceous phenology in response to rainfall, which can lead to variable estimates of cover over time (Smith et al., 2019).…”

“…Alternatively, estimates of tree restoration potential in drylands could make more conservative area estimates by using a higher threshold for identifying potential tree restoration (e.g., more than +5.9% between existing and potential cover). Identifying levels of bias in tree cover estimates across discrete regions (this paper) or across continuous measurements of rainfall (e.g., Cunningham et al., 2019) could limit the false‐positive identification of potential restoration in drylands.…”

Assessing Accuracy of Land Cover Change Maps Derived from Automated Digital Processing and Visual Interpretation in Tropical Forests in Indonesia

“…This low accuracy of the ADP method may be partially attributed to the reliance on optical images. Previous studies have shown that the use of optical images makes it difficult to differentiate between forest and dense canopy crops such as mature oil palm [17,39] as well as perennial agricultural crops and plantations of less than 5 m height such as pineapple, tea, and soybean as tree cover [3,40] due to similar spectral signal.…”

Deciphering the many maps of the Xingu – an assessment of land cover classifications at multiple scales

“…Disagreement between MapBiomas and TerraClass has been shown for classes other than forest (Körting et al 2020) whereas others have criticized the products for not following good practice guidelines in accuracy assessment (Zalles et al 2019). Others such as Cunningham et al (2019) critically assess the utility of the GFC dataset with recommendations to use with caution in area that do not meet criteria including high aseasonal rainfall, low relief, and low cropland area. Likewise, Galiatsatos et al (2020) found that the GFC dataset should not be used for precise estimates of forest cover change in countries with large areas of forest cover and low levels of deforestation.…”