A Vector Autoregressive (VAR) Model for the Hourly Forecasting of Climatological Data in General Santos City
Document Types
Paper Presentation
School Name
Mindanao State University-General Santos Senior High School
Research Advisor (Last Name, First Name, Middle Initial)
Valdueza, Ralph Laurence
Start Date
25-6-2025 10:30 AM
End Date
25-6-2025 12:00 PM
Zoom Link/ Room Assignment
https://zoom.us/j/96395524945?pwd=uMIouA2hMraYDLTYrKqbhW8mCobpfv.1 Meeting ID: 963 9552 4945 Passcode: 843648
Abstract/Executive Summary
Through multivariate time series exploration, the Vector Autoregressive (VAR) model was developed to predict marked climatological data for General Santos City, Philippines. The dataset, which covered the period from January 01, 2001, until June 30, 2024, was obtained from the NASA POWER API and included crucial variables: surface pressure, specific humidity, wind speed, air temperature, precipitation, and UV irradiance. The study recorded significant diurnal and seasonal trends, which could be observed in the surface pressure, specific humidity, air temperature, and UV irradiance that displayed the predictable cycles. Precipitation demonstrated limited seasonality, as it represented the localized tropical rain patterns. The trend analysis mentioned the increasing trend of surface pressure and air temperature while the humidity, wind speed, and precipitation showed a decreasing trend, representing drier and warmer conditions generally. The results of the Granger causality test clearly depicted that among these variables, there were interdependencies, therefore it is suggested multivariate approaches are essential for the climate forecast to be more accurate. The most suitable VAR(168) model produced impressive results for the stable variables, which are, surface pressure, and air temperature with high accuracy. With this model, the city’s climatological dynamics are thoroughly learned, and along the lines, it is possible to establish a basis for the city’s project in climate adaptation and resource management. Subsequent versions of the model may have to zero in on potential refinement to help elevate the accuracy of the UV irradiance and the precipitation variables, which are volatile.
Keywords
vector autoregressive (VAR) model; multivariate time series analysis; climate forecasting; General Santos City; NASA Power API
Research Theme (for Paper Presentation and Poster Presentation submissions only)
Sustainability, Environment, and Energy (SEE)
Initial Consent for Publication
yes
Statement of Originality
yes
A Vector Autoregressive (VAR) Model for the Hourly Forecasting of Climatological Data in General Santos City
Through multivariate time series exploration, the Vector Autoregressive (VAR) model was developed to predict marked climatological data for General Santos City, Philippines. The dataset, which covered the period from January 01, 2001, until June 30, 2024, was obtained from the NASA POWER API and included crucial variables: surface pressure, specific humidity, wind speed, air temperature, precipitation, and UV irradiance. The study recorded significant diurnal and seasonal trends, which could be observed in the surface pressure, specific humidity, air temperature, and UV irradiance that displayed the predictable cycles. Precipitation demonstrated limited seasonality, as it represented the localized tropical rain patterns. The trend analysis mentioned the increasing trend of surface pressure and air temperature while the humidity, wind speed, and precipitation showed a decreasing trend, representing drier and warmer conditions generally. The results of the Granger causality test clearly depicted that among these variables, there were interdependencies, therefore it is suggested multivariate approaches are essential for the climate forecast to be more accurate. The most suitable VAR(168) model produced impressive results for the stable variables, which are, surface pressure, and air temperature with high accuracy. With this model, the city’s climatological dynamics are thoroughly learned, and along the lines, it is possible to establish a basis for the city’s project in climate adaptation and resource management. Subsequent versions of the model may have to zero in on potential refinement to help elevate the accuracy of the UV irradiance and the precipitation variables, which are volatile.
https://animorepository.dlsu.edu.ph/conf_shsrescon/2025/paper_see/32
