The Challenges of Mapping Dryland Forests

Accurate forest cover mapping is the foundation of jurisdictional REDD+ accounting. Under VMD0055, these maps form the basis for activity data and deforestation risk allocation across a jurisdiction. Most forest mapping research has focused on closed-canopy forests, and decades of methodological development have produced reliable, well-tested tools and workflows for these systems. Dryland forests however have received less focused methodological attention, and their open, fragmented structure and persistent data limitations make them considerably harder to map accurately, though the sources of this difficulty are increasingly well understood (Reiner et al., 2023).

National forest definitions in dryland regions typically set canopy cover thresholds as low as 10—20%, compared to 30% or more in closed-canopy systems, which leads to a wide range of sparse, scattered vegetation qualifying as forest. In satellite imagery, much of this variation looks similar to shrubland, savanna, or degraded land, making it difficult for classification algorithms to reliably distinguish forest from non-forest (Bastin et al., 2017; Brandt et al., 2020). Approaches that estimate forest cover change by comparing maps across two time points are less reliable in dryland forests, where mapped change may reflect classification variability rather than actual deforestation on the ground.

Beyond classification challenges, dryland mapping is also constrained by data availability. Accurate mapping requires imagery from both wet and dry seasons, since a single time point can misrepresent actual land cover conditions. A dry season image, for example, can make a healthy forest appear deforested, while a wet season image can make degraded land appear vegetated. However, multi-seasonal coverage is often limited. Cloud cover frequently obscures wet season imagery, and because there are significant gaps in historical archives across much of Africa and South America, complete coverage for a given year isn’t always available (Wulder et al., 2016; Zhang et al., 2022). The same archive limitations affect the high-resolution imagery needed to verify map accuracy at individual sample points (Lesiv et al., 2018), and these data gaps shape both the maps themselves and their validation.

As a result of these complicating factors, building and validating an accurate map in dryland forests takes more time and iteration than in regions with denser data coverage.

These challenges are well-recognized, and Verra is working with partners to adapt the VMD0055/VT0007 framework for open-canopy landscapes. Current efforts focus on subnational accuracy assessments to identify where targeted improvement would be most valuable, subnational risk maps that account for differences in forest density across a jurisdiction, and alternative modeling approaches that perform better in local contexts. These steps aim to strengthen dryland forest accuracy within the existing methodological framework.

Figure 1. Open-canopy dryland and closed-canopy forest landscapes in Tanzania across wet and dry seasons. (A) Open-canopy dryland forest near Lindi, Tanzania, shown across wet and dry seasons. Although actual forest cover is relatively stable across these points in time, the appearance appears substantially different between seasons. (B) Closed-canopy rainforest at the foothills of Mount Kilimanjaro, Tanzania, across comparable time points. The closed canopy remains visually consistent across seasons.

While most of the area in both panels would meet Tanzania’s 10% canopy threshold for forest, the dryland landscape in panel A shows sparse, fragmented cover, while panel B shows continuous, unbroken canopy. Seasonal variability of the type shown in (A) demonstrates how a single satellite image can misrepresent dryland forest conditions. Imagery: Esri World Imagery Wayback.

Works Cited:

^​​Bastin, J. F., Berrahmouni, N., Grainger, A., Maniatis, D., Mollicone, D., Moore, R., Patriarca, C., Picard, N., Sparrow, B., Abraham, E. M., Aloui, K., Atesoglu, A., Attorre, F., Bassüllü, Ç., Bey, A., Garzuglia, M., García-Montero, L. G., Groot, N., Guerin, G., & Castro, R. (2017). The extent of forest in dryland biomes. Science, 356(6338), 635–638.  https://doi.org/10.1126/SCIENCE.AAM6527;JOURNAL:JOURNAL:SCIENCE;WGROUP:STRING:PUBLICATION

Brandt, M., Tucker, C. J., Kariryaa, A., Rasmussen, K., Abel, C., Small, J., Chave, J., Rasmussen, L. V., Hiernaux, P., Diouf, A. A., Kergoat, L., Mertz, O., Igel, C., Gieseke, F., Schöning, J., Li, S., Melocik, K., Meyer, J., Sinno, S., & Fensholt, R. (2020). An unexpectedly large count of trees in the West African Sahara and Sahel. Nature 2020 587:7832, 587(7832), 78–82. https://doi.org/10.1038/s41586-020-2824-5

Lesiv, M., See, L., Bayas, J. C. L., Sturn, T., Schepaschenko, D., Karner, M., Moorthy, I., McCallum, I., & Fritz, S. (2018). Characterizing the Spatial and Temporal Availability of Very High Resolution Satellite Imagery in Google Earth and Microsoft Bing Maps as a Source of Reference Data. Land 2018, Vol. 7, Page 118, 7(4), 118. https://doi.org/10.3390/LAND7040118

Reiner, F., Brandt, M., Tong, X., Skole, D., Kariryaa, A., Ciais, P., Davies, A., Hiernaux, P., Chave, J., Mugabowindekwe, M., Igel, C., Oehmcke, S., Gieseke, F., Li, S., Liu, S., Saatchi, S., Boucher, P., Singh, J., Taugourdeau, S., … Fensholt, R. (2023). More than one quarter of Africa’s tree cover is found outside areas previously classified as forest. Nature Communications 2023 14:1, 14(1), 2258-. https://doi.org/10.1038/s41467-023-37880-4

Wulder, M. A., White, J. C., Loveland, T. R., Woodcock, C. E., Belward, A. S., Cohen, W. B., Fosnight, E. A., Shaw, J., Masek, J. G., & Roy, D. P. (2016). The global Landsat archive: Status, consolidation, and direction. Remote Sensing of Environment, 185, 271–283. https://doi.org/10.1016/J.RSE.2015.11.032

Zhang, Y., Woodcock, C. E., Arévalo, P., Olofsson, P., Tang, X., Stanimirova, R., Bullock, E., Tarrio, K. R., Zhu, Z., & Friedl, M. A. (2022). A Global Analysis of the Spatial and Temporal Variability of Usable Landsat Observations at the Pixel Scale. Frontiers in Remote Sensing, 3, 894618. https://doi.org/10.3389/FRSEN.2022.894618/TEXT