Development of statistical and geospatial-based framework for drought-risk assessment

Development of statistical and geospatial-based framework for drought-risk assessment

Drought is an insidious, complex and one of the least understood natural phenomena resulting from a deficiency of water resources. While droughts cannot be prevented, its impacts, however, can be mitigated through proper design of water storage infrastructure and management strategies. A comprehensive drought management plan necessitates the development of a framework that can help reduce the drought-related risk. In Australia, there are limited drought vulnerability and risk assessment models that must (1) include the drought monitoring index that measures the supply-demand balance of water resources, (2) incorporate large-scale climate drivers influencing amplitude of drought events in the statistical prediction models, and (3) objectively quantify the drought-risk on both temporal and spatial scales. The goal of this study is to apply statistical and geospatial tools in developing a framework for assessing drought-related risks in light of improving the drought mitigation strategies.

A new, temporal and spatial-explicit analytical framework for drought-risk assessment is developed based on three objectives focussed in the drought-prone southeast Queensland (SEQ) region. (1) Evaluating and affirming the suitability of the Standardised Precipitation-Evapotranspiration Index (SPEI) for the characterisation of drought events. (2) Developing a copula-based statistical, probabilistic model for predicting the SPEI and the jointly distributed drought properties (i.e., durations, severities and intensities) conditional on the large-scale climate mode indices. (3) Developing a spatially descriptive drought-risk index by combining the drought hazard, exposure and vulnerability factors using a fuzzy logic algorithm.

The first objective of this study demonstrates the scientific relevance of the SPEI as a robust drought assessment metric that incorporates the influence of water supply-demand balance on drought events. Subsequently, the severity (S; accumulated negative SPEI in a drought-identified period), intensity (I; minimum SPEI) and the duration (D; number of months with continuously negative SPEI representing the below average water resources) based on run-sum approach are enumerated to identify historical water deficit periods. Significant disparities in the identified D-S-I affirms the significance of SPEI for regional drought impact assessments. Accordingly, this study advocates the SPEI as a convenient metric for detecting drought onsets and terminations, including its ability for drought ranking and drought recurrence evaluations that are considered vital for water resource management.

The second objective models the joint behaviour of SPEI and D-S-I properties using copula model, conditional upon the pertinent climate mode indices (i.e., El-Niño Southern Oscillation indicators). The vine copula algorithm is employed to derive the bivariate and trivariate joint-distributions of drought variables for conditional probability-based predictions. The results yield marginal differences between the observed and the predicted drought properties, elucidating the effectiveness of copula functions in drought-risk modelling. The results have implications for drought and aridity management in agricultural regions where complex relationships between climate drivers and drought properties are likely to exacerbate the risk of a future event.

The third objective develops a methodology using vulnerability, exposure and hazard indicators to provide a spatio-temporal framework for drought-risk assessment. The conditional joint probability of each drought indicator is estimated using the Bayes theorem. Various fuzzy membership functions are then applied to standardise and aggregate the indicators to derive drought vulnerability, exposure and hazard indices. The resulting indices are integrated with fuzzy GAMMA overlay operation to generate optimal drought-risk maps. The maps reveal varying levels of drought risk in different austral seasons and annually that is well represented by the drought hazard index, i.e., rainfall departure. The validation of the method with respect to the upper and lower layer soil moisture reveal significant correlations with the spatial drought-risk index. It is therefore prudent to state that the fuzzy logic-based analytical technique applied for spatio-temporal drought-risk mapping can be considered as a practical tool that can enable better drought management, drought mitigation and relief-planning decisions.

The statistically and spatially relevant drought-risk assessments frameworks formulated in this study provides promising outcomes that are valuable for the mitigation of drought impacts, and therefore, sets a pathway to construct strategic planning procedures and management of water resources in drought-prone, arid or semi-arid regions.