Seasonal Variability of Deforestation-Induced Warming in the Congo Basin Using Remote-Sensing Data

Deforestation impacts in the Congo Basin remain significantly understudied compared to other tropical regions. The main driver of Congo Basin deforestation is small-scale industrial agriculture, which leads to the formation of the rural complex; a mosaic patch of deforested land comprising small fields at different stages of regrowth being deforested repeatedly. Transition from primary forest to rural complex may induce lesser changes in albedo, Bowen ratio, and surface roughness than primary forest to cropland, suggesting the impacts of deforestation on temperatures in the Congo Basin will differ from those in other rainforest regions. The Basin's long-term warming trend and possible ongoing drying could exacerbate warming due to deforestation. It is therefore essential that we understand how the specific nature of deforestation in the Congo Basin influences temperatures, and how this is affected by changes in the large-scale conditions driven by global climate change.

In this study, we used MODIS satellite data for LST and EVI, CHIRPS2 for rainfall, and the Global Forest Change dataset for deforestation analysis from 2000 to 2019 to assess how observed deforestation is affecting LST in the Congo Basin and how the deforestation-induced warming varies with climate anomalies, LST and rainfall (SPI), and Δ EVI (deforested EVI – surrounding forest EVI). Due to limited data availability, caused by the prevalence of cloud cover throughout much of the year, our focus narrowed to the most data-consistent dry season (DJF), where land-atmosphere interactions are also likely to be strongest.

We found a linear relationship between cumulative deforestation and warming over deforested land, which varied in intensity by month. A typical 1 km rural complex pixel within the region will warm by +0.33 °C in December, +0.85 °C in January, and +1.54 °C in February, relative to the surrounding forest. We then assessed the cause of the strong seasonal differences by looking at the deforestation-induced warming as a factor of the climate anomalies and Δ EVI. The amount of warming of a typical 1 km rural complex pixel did not show a relationship with the LST anomaly or SPI for the individual months. However, when considering all months collectively, a correlation emerged with the LST anomaly, suggesting a seasonal evolution where the LST anomaly acts as a proxy. We then found a link between the warming of a typical 1 km rural complex pixel and Δ EVI which is present for each month; this partially explains the interannual variability of the results, but it doesn’t explain the seasonal evolution. Comprehensive and high-quality observations are needed over the Congo Basin to fully untangle these relationships. Accurate soil moisture data could be crucial in understanding the pronounced seasonal differences in warming. These findings suggest that even though the rural complex differs from cropland, and might be expected to have a smaller impact, the additional warming can still be substantial (+1.54 °C), although it has a strong seasonal dependency.