Mapping Heat Risks Using A Non-Parametric Multivariate Approach

In the face of ongoing climate changes, the frequency and intensity of unusual weather phenomena, temperature variations, and natural disasters have intensified. Consequently, the associated impacts, including damage and disruption, have escalated as well. The increase in temperature has led to a rise in occurrence of heat waves and tropical nights, leading to various challenges in different regions. The repercussions of heat waves encompass a wide array of issues, ranging from health concerns to energy supply shortages. Addressing these multifaceted risks necessitates a thorough analysis of their distribution and characteristics. Previous studies often treated different aspects of heat waves in isolation, which led to an incomplete understanding of their overall impact. To rectify this limitation, we present a novel non-parametric multivariate approach to assess heat risks. This method combines non-parametric multivariate kernel density estimation and data envelopment analysis (DEA) to evaluate hazard and vulnerability, respectively. For assessing heat hazard, we use the universal thermal climate index (UTCI), which considers factors such as humidity, wind, radiation, and human metabolism, in addition to temperature. By utilizing the maximum daily UTCI values, we derive intensities and durations, which are used to construct hazard maps illustrating the joint exceedance probability of intensity and duration. To determine vulnerability to heat, we select 41 indicators from the 2020 United States Census that pertain to heat mitigation. These indicators are categorized into 35 sensitive and 6 adaptive criteria. The vulnerability index is calculated using DEA, offering insights into each region’s susceptibility to heat-related risks. By combining hazard and vulnerability, with population as a measure of exposure, we calculate the overall risk using a product-based methodology at a county scale. We demonstrated this innovative approach across the contiguous United States, showcasing its potential applicability worldwide. This reproducible methodology holds promise for guiding policymakers in developing effective climate adaptation strategies, with the flexibility to address varying local conditions.