Wavelet Analysis of Dengue Incidence and its Correlation with Weather and Vegetation Variables in Costa Rica

TL;DR

This study uses wavelet analysis to identify periodic patterns and correlations between dengue incidence and environmental factors like vegetation, temperature, and climate indices in Costa Rica from 2000 to 2019.

Contribution

It applies wavelet coherence analysis to quantify the relationship between dengue and environmental variables across multiple regions and time scales.

Findings

Dengue cycles are mainly 1, 2, and 3 years.

Vegetation indices correlate with dengue in central and northern Pacific regions.

El Niño indices show a strong 3-year correlation with dengue.

Abstract

Dengue represents a serious public health problem in tropical and subtropical regions worldwide. The number of dengue cases and its geographical expansion has increased in recent decades, driven mostly after by social and environmental factors. In Costa Rica, it has been endemic since it was first introduced in 1993. In this article, wavelet analyzes (wavelet power spectrum and wavelet coherence) were performed to detect and quantify dengue periodicity and describe patterns of synchrony between dengue incidence and climatic and environmental factors: Normalized Difference Water Index, Enhanced Vegetation Index, Normalized Difference Vegetation Index, Tropical North Atlantic indices, Land Surface Temperature, and El Ni\~no Southern Oscillation indices in 32 different cantons, using dengue surveillance from 2000 to 2019. Results showed that the dengue dominant cycles are in periods of 1, 2, and 3 years. The wavelet coherence analysis showed that the vegetation indices are correlated with dengue incidence in places located in the central and Northern Pacific of the country in the period of 1 year. Climatic variables such as El Ni\~no 3, 3.4, 4, showed a strong correlation with dengue incidence in the period of 3 years and the Tropical North Atlantic is correlated with dengue incidence in the period of 1 year. Land Surface Temperature showed a strong correlation with dengue time series in the 32 cantons.