Presentation Type

Poster

Faculty Mentor’s Full Name

Marco Maneta

Faculty Mentor’s Department

Geosciences

Abstract

Droughts are one of the most prevalent natural hazards. They cause severe economic losses and water and food insecurity. While droughts can be observed, the mechanisms that cause a drought to occur and persist are poorly understood. Our aim is to better understand drought behavior by using principal component analysis to determine their causes, as well as further insight to their onset, duration, and effects.

A drought can be defined as an area and duration of below-average precipitation compared to that area’s mean precipitation at that point in time annually. A useful tool for determining and analyzing a drought is the Standardised Precipitation-Evapotranspiration Index (SPEI.) The SPEI uses temperature, precipitation, and evapotranspiration information to calculate the net precipitation anomaly across the globe. Using consecutive-in-time SPEI maps, one can observe the spatio-temporal onset, duration, and end of a drought.

While the SPEI can be used to identify and observe droughts, we used Principal Component Analysis (PCA) to try to understand the teleconnections between droughts and their causes. PCA is a statistical method that creates new, uncorrelated variables (Principal Components, PCs) and allows for clearer interpretation of large datasets of dependent variables and minimizing information loss. The correlation of PCs to the dataset is given by their variance. By applying PCA to the SPEI dataset for the continental United States, we found that the first four PCs had more than 50% of the variance. Studying the trendline of each PC through time, we then fit the trendlines of other climatic forces to see possible causes of drought. From this comparison, we found that the first (primary) PC is influenced by the Southern Oscillation (driver of the El Niño cycle) and the second PC is influenced by the North American Monsoon, which occurs annually and affects the Southwestern states, Texas, and Colorado. These results not only show some of the climatic teleconnections to past droughts, but show that PCA is a viable method for discovering such teleconnections.

Category

Physical Sciences

Share

COinS
 

Understanding Drought: An Investigation Using Principal Component Analysis

Droughts are one of the most prevalent natural hazards. They cause severe economic losses and water and food insecurity. While droughts can be observed, the mechanisms that cause a drought to occur and persist are poorly understood. Our aim is to better understand drought behavior by using principal component analysis to determine their causes, as well as further insight to their onset, duration, and effects.

A drought can be defined as an area and duration of below-average precipitation compared to that area’s mean precipitation at that point in time annually. A useful tool for determining and analyzing a drought is the Standardised Precipitation-Evapotranspiration Index (SPEI.) The SPEI uses temperature, precipitation, and evapotranspiration information to calculate the net precipitation anomaly across the globe. Using consecutive-in-time SPEI maps, one can observe the spatio-temporal onset, duration, and end of a drought.

While the SPEI can be used to identify and observe droughts, we used Principal Component Analysis (PCA) to try to understand the teleconnections between droughts and their causes. PCA is a statistical method that creates new, uncorrelated variables (Principal Components, PCs) and allows for clearer interpretation of large datasets of dependent variables and minimizing information loss. The correlation of PCs to the dataset is given by their variance. By applying PCA to the SPEI dataset for the continental United States, we found that the first four PCs had more than 50% of the variance. Studying the trendline of each PC through time, we then fit the trendlines of other climatic forces to see possible causes of drought. From this comparison, we found that the first (primary) PC is influenced by the Southern Oscillation (driver of the El Niño cycle) and the second PC is influenced by the North American Monsoon, which occurs annually and affects the Southwestern states, Texas, and Colorado. These results not only show some of the climatic teleconnections to past droughts, but show that PCA is a viable method for discovering such teleconnections.