Model Autoregressive Distributed Lag Pengaruh Faktor Cuaca Kota Bogor Terhadap Tinggi Muka Air di Jakarta

Abstract

Model Autoregressive Distributed Lag (ARDL) merupakan analisis deret waktu multivariat yang mampu menangkap hubungan jangka pendek dan jangka panjang antara peubah. Salah satu indikator penting dalam sistem peringatan dini banjir di Jakarta adalah Tinggi Muka Air (TMA) di Pintu Air Manggarai. Kota Bogor, yang dikenal sebagai "Kota Hujan", berperan signifikan dalam siklus hidrologi kawasan Jabodetabek. Tingginya curah hujan di Bogor menyebabkan aliran air besar menuju Jakarta, yang pada akhirnya memengaruhi kondisi TMA. Penelitian ini menggunakan data harian dari tahun 2022 hingga 2024 dengan peubah respon berupa TMA, serta peubah penjelas meliputi curah hujan, suhu rata-rata, kelembapan relatif, dan kecepatan angin di Kota Bogor. Pemilihan lag optimal dilakukan menggunakan Cross Corelation Function (CCF). Model ARDL yang dibangun memiliki kemampuan peramalan yang baik dengan nilai Mean Absolute Percentage (MAPE) sebesar 2,83%. Hasil analisis menunjukkan bahwa peubah curah hujan, suhu rata-rata, dan kelembapan relatif berpengaruh positif terhadap TMA, sementara kecepatan angin berpengaruh negatif.

The Autoregressive Distributed Lag (ARDL) model is a multivariate time series approach that allows for the estimation of both short-run and long-run dynamics among variables. One of the key indicators in Jakarta's flood early warning system is the water level at the Manggarai Water Gate. Bogor City, known as the "Rain City," plays a significant role in the hydrological cycle of the Jabodetabek region. High rainfall in Bogor contributes to substantial water flow toward Jakarta, ultimately affecting the water level conditions. Daily data from 2022 to 2024 are used in this study, with the response variable being water level at the Manggarai Water Gate, and the explanatory variables including rainfall, average temperature, relative humidity, and wind speed in Bogor City. The optimal lag selection was conducted using the Cross-Correlation Function (CCF). The constructed ARDL model demonstrates strong forecasting performance, with a Mean Absolute Percentage Error (MAPE) of 2.83%. The analysis results indicate that rainfall, average temperature, and relative humidity have a positive effect on water level at the Manggarai Water Gate, while wind speed has a negative effect.