Functional time series forecasting of air temperature using a functional autoregressive model with electricity demand as an exogenous variable

Short-term air temperature (AT) forecasting is crucial for climate monitoring, integration of renewable energy, agricultural management, and environmental decision-making. The inherent complexity of the temporal structure of AT (nonlinear dynamics, high diurnal periodicity, the interactions between various meteorological variables) makes the conventional statistical and machine learning models (which typically view AT observations as discrete time series and do not make effective use of the continuous functional structure of AT) less effective in predicting AT. This study proposes a functional autoregressive model with exogenous variable (FARX(1)) for predicting daily AT curves in a functional time series (FTS) context, overcoming these limitations. The proposed model is an extension of the classical functional autoregressive model of order one (FAR(1)), which maintains the time dependence and smooth functional structure of the daily temperature measurements; while additionally incorporating electricity demand as the exogenous variable, given its established association with diurnal temperature dynamics. The hourly AT data is approximated by smooth functional curves that enable efficient modeling of the diurnal temperature dynamics and observational noise in the data, and is obtained by an approximation of the data with Fourier basis functions. The ability of the proposed framework to forecast the performance is evaluated by comparing it with five widely used benchmark models: ARIMA, VAR, ANN, ARNN, and FAR(1) over extensive overall forecasting, hourly forecasting, weekly forecasting and monthly forecasting scenarios, employing the Mean Absolute Scaled Error (MASE), Symmetric Mean Absolute Percentage Error (sMAPE), and Root Mean Square Error (RMSE). The experimental results demonstrate the effectiveness of the proposed FARX(1) model in minimizing the conventional FAR(1)’s forecasting error for all the evaluation scenarios, highlighting the significance of using the exogenous variable of electricity demand in the model for forecasting accuracy. The results also highlight the potential of representing AT as functional observations to obtain a flexible and statistically sound representation of complex temporal variations, which enhances the reliability and robustness of short-term AT forecasts. The concept is proposed as a solution for the meteorology community in operational forecasting to the weather community and as climate monitoring and decision support to weather dependent applications such as renewable energy management.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-21
DOI
https://doi.org/10.1038/s41598-026-78090-4
Primary Topic
Energy Load and Power Forecasting
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Functional time series forecasting of air temperature using a functional autoregressive model with electricity demand as an exogenous variable

Huda M. Alshanbari, Faheem Jan, Naveed Iqbal, Ahmad Shafee et al.
4 links
Scientific Reports
Energy Load and Power Forecasting
article

Functional time series forecasting of air temperature using a functional autoregressive model with electricity demand as an exogenous variable

Huda M. Alshanbari, Faheem Jan, Naveed Iqbal, Ahmad Shafee, Gulagha Jan Assar, Mehwish Tahir
article en

Abstract

Short-term air temperature (AT) forecasting is crucial for climate monitoring, integration of renewable energy, agricultural management, and environmental decision-making. The inherent complexity of the temporal structure of AT (nonlinear dynamics, high diurnal periodicity, the interactions between various meteorological variables) makes the conventional statistical and machine learning models (which typically view AT observations as discrete time series and do not make effective use of the continuous functional structure of AT) less effective in predicting AT. This study proposes a functional autoregressive model with exogenous variable (FARX(1)) for predicting daily AT curves in a functional time series (FTS) context, overcoming these limitations. The proposed model is an extension of the classical functional autoregressive model of order one (FAR(1)), which maintains the time dependence and smooth functional structure of the daily temperature measurements; while additionally incorporating electricity demand as the exogenous variable, given its established association with diurnal temperature dynamics. The hourly AT data is approximated by smooth functional curves that enable efficient modeling of the diurnal temperature dynamics and observational noise in the data, and is obtained by an approximation of the data with Fourier basis functions. The ability of the proposed framework to forecast the performance is evaluated by comparing it with five widely used benchmark models: ARIMA, VAR, ANN, ARNN, and FAR(1) over extensive overall forecasting, hourly forecasting, weekly forecasting and monthly forecasting scenarios, employing the Mean Absolute Scaled Error (MASE), Symmetric Mean Absolute Percentage Error (sMAPE), and Root Mean Square Error (RMSE). The experimental results demonstrate the effectiveness of the proposed FARX(1) model in minimizing the conventional FAR(1)’s forecasting error for all the evaluation scenarios, highlighting the significance of using the exogenous variable of electricity demand in the model for forecasting accuracy. The results also highlight the potential of representing AT as functional observations to obtain a flexible and statistically sound representation of complex temporal variations, which enhances the reliability and robustness of short-term AT forecasts. The concept is proposed as a solution for the meteorology community in operational forecasting to the weather community and as climate monitoring and decision support to weather dependent applications such as renewable energy management.

Scientific Reports
Princess Nourah bint Abdulrahman University (SA), Shaheed Benazir Bhutto Women University Peshawar (PK), Bacha Khan University (PK), Public Authority for Applied Education and Training (KW), University of Ha'il (SA), Polytechnical University of Kabul (AF)
Climate action
Energy Load and Power Forecasting
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon