Identification of small-scale agricultural clearing and regrowth in the Congo Basin and its impacts on temperature

Abstract

Deforestation in tropical regions plays a significant role in altering local and regional climate dynamics. Most of the deforestation across the Congo Basin is due to small-scale agricultural clearing which results in the rural complex, a dynamic mosaic of active and fallow fields, and secondary forest regrowth. The Congo Basin has received less attention than other tropical regions and the limited studies assessing the climatic impact of deforestation in the region have mostly focussed on future, and often unrealistic, projections of deforestation. Very few of these studies have used observations, and none have focussed on the rural complex, despite its status as the predominant deforested land type in the region. Studies of tropical deforestation typically find that surface roughness and evapotranspiration efficiency are the main drivers of deforestation-induced warming, but small-scale agricultural deforestation has much more diverse land types than other typical deforestation drivers and, therefore, may have different impacts on surface roughness and evapotranspiration efficiency. Large-scale climate conditions can also affect deforestation-induced warming. The region's comparatively drier conditions and the observed long-term warming and drying trend could result in a distinct response to land-use change compared to other tropical areas. The primary aim of this project is to assess the impacts of deforestation on local temperature in the Congo Basin, by first developing a method to identify deforestation and its subsequent evolution, then quantifying warming over the deforested land relative to its surroundings, and finally using a climate model to look at how deforestation impacts on temperatures might change in the future. The secondary aim of this project is to understand how deforestation-induced warming is impacted by interannual and intraseasonal changes in large-scale climate conditions. In Chapter 2, I developed a novel method using remote-sensing observations to detect deforestation events on a sub-km scale on a monthly basis. I then quantified the post-deforestation recovery of vegetation proxies to determine the potential longevity of deforestation impacts. I was able to detect 79% of deforestation events at a 500 m scale, showing that small-scale agricultural clearing, although typically small, often clusters together to form deforested areas large enough to cause atmospheric impacts. I then found that 66% of the deforested locations recovered to the median pre-deforestation vegetation proxy value within a year, showing the rapid regrowth common in the region. The recovery varied by the land type, as the plantation land and rural complex recovered to their median pre-deforestation vegetation proxy value faster than the forested land did. The fallow period was generally very short as ~88% of deforested locations underwent further detectable deforestation within 6 years. This foundation of detailed deforestation detection provided useful insights into the dynamics of the rural complex and was essential for examining the link between deforestation and temperature changes in subsequent chapters. In Chapter 3, I used remote-sensing data to examine how cumulative deforestation affected land surface temperature (LST). This study found a linear relationship between cumulative deforestation and LST, showing that the total area of the rural complex has a strong impact on warming. This linear relationship varied both seasonally and interannually as the warming of a typical 1 km rural complex pixel increased as the dry season progressed, up to a maximum warming of +1.59 degrees Celsius in February. The interannual variability also varied by up to 1 degree Celsius. To understand the cause of this variability I then assessed how the relationship between cumulative deforestation and LST is moderated by large-scale climate conditions by applying Bayesian moderation analysis and found a strong moderation effect from the Enhanced Vegetation Index (EVI), a proxy for vegetation's response to water stress. I also found that the Standardised Precipitation Index (SPI), had a moderate control on deforestation-induced warming but there was no control from the large-scale LST anomaly. This understanding of the relationship between deforestation and temperature paved the way for a comparison of how remote-sensing observations and a convection-permitting regional climate model simulation (CP4-Africa) represent the impact of the rural complex on temperatures (Chapter 4). The observations showed that the rural complex had an average LST warming of 1.16 degrees Celsius, and the model showed an average air temperature warming of the rural complex of 1.4 degrees Celsius, representing the rural complex's warming well. However, the model did not capture the increase in deforestation-induced warming as the dry season progressed and the deforestation-induced warming responded differently to large-scale climate conditions. The CP4-Africa simulation represented the mechanisms driving deforestation-induced well and established that evapotranspiration efficiency is the strongest driver, with minimal impact from changes in surface roughness. The deforestation-induced warming reduced by ~51% in the future high-emission climate scenario due to the disproportionate impacts on the primary forest's surface energy fluxes from the high CO2 future. This research shows that small-scale deforestation has a substantial impact on temperatures despite the presence of a rural complex with large amounts of secondary regrowth. It also highlights the challenges in retrieving high-quality remote sensing data in this region and emphasises the need for increased numbers of ground-based observations in the region to enhance understanding and management of its forests and climate. Future research should focus on extending these findings across the Congo Basin, exploring the impacts of smaller rural complex patches, and investigating the interaction between deforestation-induced warming and rainfall patterns, especially under varying climate scenarios.

Date of Award	9 Dec 2024
Original language	English
Awarding Institution	The University of Manchester
Supervisor	David Schultz (Main Supervisor)