Hydrological drought and wildfire in the humid tropics

Drought is a recurrent hazard, which has happened throughout human history, and it is anticipated to become more severe in multiple regions across the world. Drought occurs in all climate regimes from humid to dry and from hot to cold. Drought is often viewed through its impact on environment and society, including wildfire, which is the topic of this study. The nature of such impacts differs remarkably from region to region. Although drought does not directly cause wildfire, it provides favorable conditions for wildfire ignition and spread. When drought coincides with strong El Niño events in the humid tropics, e.g. Southeast Asia, the impacts worsen through uncontrolled forest fires affecting the global carbon cycle. These include reduction of the carbon stock, intensifying the haze hazard, and other severe socio-economic impacts in Southeast Asia, including areas far away from the burnt area, e.g. Singapore because of fires in Sumatra. There still remains a serious lack of scientific understanding about the fundamental role of drought in fire-generating processes. Most research, so far, suggests that climate controls wildfire occurrence in the humid tropics. However, this climate-centered approach, which is reflected in contemporary drought-fire related indices, overlooks soil and hydrological processes beneath the surface across the humid tropics. There is also uncertainty about the relative roles of climate variability and human activities in influencing the nature and distribution of drought-related wildfires. Hence, the general objective of this PhD research is to examine how characterization of hydrological drought under natural and human-modified conditions can improve understanding of wildfires in general in the humid tropics. Chapter 2 discusses the contribution of humans to an increase of hydrological drought severity in the tropical peatland of Southeast Asia. Climate variability induces drought in the region, however, human activities (human-modified drought) may increase its severity. Analyzing long time series of simulated historical groundwater levels from selected regions in Southeast Asia, which were validated against some years with observations, revealed that human interference (through canalization and land-use change) has amplified drought severity. The drought amplification due to human interference was at least double that of climate-induced drought. The amplification is even higher when peatland is converted into acacia plantation. Further, research findings suggest that even if the Paris Agreement target is met, drought risk of peatlands remains high unless sustainable water management receives top priority in the region. Chapter 3 deals with how an existing, well-known drought-fire related index, i.e. the Keetch-Byram Drought Index (KBDI), is modified to improve applicability in the humid climate environment of Southeast Asia. The improvement includes: (i) adjustment of the drought factor to the local climate, and (ii) addition of the water table depth as a dynamic factor to fine-tune the drought index. The results indicate that the modified Keetch-Byram Drought Index (mKBDI) performed well in predicting fire hazard. Furthermore, the research identified a critical water table depth, which represents maximum fire hazard (0.85 m for the wetland forest of South Sumatra). Below this value hazard does not increase anymore. The mKBDI could be more widely applied, if pedotransfer functions would be developed that link easily-obtainable soil properties to the parameters of the water table factor. Chapter 4 shows that wetland transformation (i.e. through canalization and land-use change) not only affects hydrological drought (Chapter 2), but also influences fire behaviour. In Southeast Asia, expansion of agricultural cropland and forest plantations has changed the landscape of wetlands. The findings showed that the transformation into acacia plantation has amplified the fire hazard from 4% (under natural conditions) to 17%. An even higher amplification (40% fire hazard) is expected under poor water management, that is, uncontrolled drainage. The findings derived from this observation-based modeling experiment suggest that improved water management (controlled drainage with higher dry season surface water levels) can minimize fire susceptibility. Chapter 5 explains the importance of hydrology for fire hazard studies. Borneo is selected to investigate the added value of including hydrological variables in fire hazard prediction approaches. More than 300 statistical models were tested, and the results showed that models that include hydrological variables better predict area burnt than those solely based on climate indicators/indices. Further, modelling evidence shows amplifying wildfires and greater area burnt in response to El Niño Southern Oscillation (ENSO) strength, when hydrology is considered. These results highlight the importance of considering hydrological drought for wildfire prediction. I recommend that hydrology should be considered in future studies of the impact of projected ENSO strength, including effects on tropical ecosystems and biodiversity conservation. The contributions of this thesis research to science are summarized and synthesized in Chapter 6. First, the research identified that fire hazard studies would benefit from adding hydrology, which is reflected in the improved model performance when hydrological variables are integrated. Next, the research revealed that humans play a substantial role in modifying groundwater drought characteristics, hence amplifying the fire hazard in Southeast Asia. Further, the chapter identified several relevant research findings, including the model choice, which should consider the simplicity and the applicability of the model. Another finding demonstrated that controlling canal water level through canal blocking is a practical water management tool to restore degraded wetland. This restored wetland would benefit some endemic species. However, the restored wetland still faces high drought severity. Hence they remain more fire-prone until the un-impacted hydrology condition is achieved. Finally, this research suggest that currently widely-used drought indices (such as FWI) require improvements in their model structure, which means integration of hydrological variables to increase their applicability for fire hazard studies in the humid tropics