Spatial Variations in Mortality Due to Heat Waves in Southern Spain

Francisco Viciana, Junta de Andalucia
Diego Ramiro-Fariñas, Spanish National Research Counsel (CSIC)
Dariya Ordanovic, ESRI Spain

This study provides a comprehensive analytical insight on the effects of extreme summer temperatures on human health and survival on mid and short-term time scales. We gathered relevant meteorological and air quality measurements for 2000-2016 and performed time-space data mining to capture local patterns of heat wave phenomenon across Andalusia. Specific attention was payed to temperature fluctuations with particular focus on large urban areas. We explored causal relationship between the cumulative exposures and mortality by age, sex and cause of death in various locations at a daily time-scale. Individual demographic and socioeconomic data were derived from the Longitudinal Statistics on Survival and Longevity of Andalusia and aggregated to cells with smallest resolution of 10 sq. km. To perform a complex analysis of spatial and non-spatial data we developed a series of integrated R-Python models which showed a robust performance and flexibility.

Background.

The increase in mean annual temperatures remains a robust trend globally. Data released by NASA on changes in global surface temperatures for 2000-2016 witness a sharp increase from 2011 onwards with 2016 ranking as the warmest on record (+0.9 above the long-term average). According to historical data on mean annual temperatures in Andalusia - the most southern autonomous community of Spain with the population of 8.4 million people in 2016 - an overall increase in approximately 0.6 degrees Celsius has occurred over the past century. In the nearest future, a strong increase of mean maximum summer temperatures by 6 degrees is predicted for the Iberian Peninsula, as shown by simulations from high-resolution regional climate model of the Danish Meteorological Institute under the A2 emission scenario. A handful of studies focusing on summer heat waves in the future point out to their foreseen extended duration, increased frequency and intensity, especially in large metropolitan areas.

There is no commonly established definition of a heat wave used by researchers, health practitioners or policy makers due to a great variety in study purposes and therefore definition criteria. According to the World Meteorological Organization survey, 34 out of 191 WMO members had a national definition of heat wave with an established threshold of temperature, heat or heat-health index. Spanish National Meteorological Agency (AEMET) defines a heat wave as an episode of at least 3 consecutive days, in which at least 10% of the stations registered maximum records above the 95% percentile of the daily maximum temperatures series for July and August during the reference period.

According to an official statistical report provided by AEMET, since 1975 up until 2015 Spain experienced 47 heat waves in total. On average, these heat waves lasted 5 days with an average maximum temperature slightly exceeding 37.5 degrees Celsius. Across this period the occurrence of heat waves has varied: until 1990 heat waves were registered every 1.3 years, in the next decade – every 1.6 years, and from 2000 onwards the frequency has increased and heat waves were registered every 1.2 years on average.

The effects of extreme temperatures on mortality in Spain are commonly studied using total daily mortality broken down by age and cause of death at a single or multiple city level. The analysis of heat waves on daily and total mortality for larger areas involved discrete meteorological measurements aggregated to province or autonomous community levels.

Data and methods.

Environmental data:

  1. Contemporary (2000-2016) climatic data (https://opendata.aemet.es/centrodedescargas/inicio): hourly measurements of temperature and precipitation; daily measurements of humidity, atmospheric pressure and wind, collected from 2007 meteorological stations across country.
  2. Hourly measurements of ozone, particulate matter and nitrogen oxides for 2000-2016 are collected for 128 operating stations (http://www.juntadeandalucia.es/medioambiente/site/portalweb/).
  3. Seamless vector “snapshots” of land use classes at 3 levels of generalisation for 2000, 2006 and 2012 (https://www.eea.europa.eu/data-and-maps/data/).
  4. Contextual spatial data for Andalusia including governmental services, urban systems, roads, communication and transport infrastructures (https://www.juntadeandalucia.es/institutodeestadisticaycartografia).

Demographic data: statistical grid with the minimal resolution of 250 sq.m gathers various demographic and socioeconomic characteristics of population (an example of source layer with standardized mortality ratios (SMR) for all causes of death, aggregated for the entire period of 2002 – 2013 http://www.juntadeandalucia.es/institutodeestadisticaycartografia/longevidad/mapa/index.htm).

Each cell contains an aggregated value corresponding to the total number of inhabitants within the cell (with information on nationality, age, educational attainment, unemployment rates) and SMR (estimates by age, sex and cause of death). To avoid the disclosure of confidential information through spatial display, IECA provided daily SMR for the periods from June to September for 2000-2016.

To model spatial relationship between environmental conditions and mortality at the grid level, we used the R-ArcGIS Bridge connector to deploy integrated models designed to perform extensive data processing (e.g. co-Kriging interpolation), spatial data mining (e.g. hot spot analysis) and statistical modelling (e.g Poisson regression).

Expected results

Commonly, studies estimating susceptibility of population to climatic variations, acute heat stresses and its effect on mortality have used aggregated measures at municipal, borough or, at maximum, census track level. Studies using large scale individual level data to measure these effects are rare. This paper aims to overcome these caveats by using individual level data combined with detailed environmental characteristics to a) compare performance of models based on discontinuous and continuous environmental measurements, and b) to test the following series of hypothesis:

  • Locations where extreme temperatures occur repeatedly every year have higher mortality rates in comparison with adjacent areas. A time lag in increment in mortality after heat wave event occurs.
  • Physical environment determines health outcomes in Andalusia if SES is equal.
  • Age-dependent relationship with the number of hot days is spatially heterogeneous across Andalusia.
  • Differences in the response to acute heat stress by gender are significant.
  • Extreme temperatures correlate with concentrations of certain contaminators which can potentially contribute to a higher heat-related mortality in urban heat islands of Andalusia.

Presented in Session 1222: Environment, Development, and Space