Sanya-Chauhan / Global_Sea_Level_Change

Analysis and Forecast of Global Mean Sea Level Change due to Global Warming

Global Mean Sea Level Change Due to Global Warming

Time Series Analysis and Forecasting

Key Concepts

• Stationarity of a Time Series Data Set
• Decomposition of Raw Data Set Using Ratio to Moving Average Method
• Mathematical Curves
• ARIMA Modeling
• Ljung Box Test
• Forecasting

Model Workflow

1. Stationarity Testing:

• Visualizing Raw Data & 1st Differences
• KPSS Test with p-value 0.01 implying rejection of H0 and non-stationarity in data

2. Data Decomposition:

• Used Additive Model

• Decomposed Components: Trend, Seasonality, Residual Series with Random/White Noise

  

3. Individual Component Fitting:

• Trend: Fitted Cubic Curve (highest adjusted R^2 of 0.991 & least MSE)
• Seasonality: ACF and PACF Plots
• Random Noise: ARIMA Modelling, Ljung–Box test for goodness of fit

Forecasting

Results and Inference

• The best Fit is AR(1) to forecast seasonality due to geometric decay in the ACF plot and sharp cut-off in the PACF Plot.
• Out of an overall YoY rise of 3.2 ± 0.5 mm, 1.8 ± 0.41 mm is due to Climate Change (~43%).
• The overall rise of 102.5 mm (approx. 4 inches) seen since 1993; the forecasted increase of approx. 109.6 mm till 2025 (i.e., ~ 6.9% rise in change in 4 years).

Future Scope

• Studying regional impact analysis with the inclusion of isostatic causes.
• Extending the study to support the formulation of natural calamity action plans and better the existing understanding of their occurrences

For more details:

• One Page Summary
• Detailed Presentation