Drought_Forecasting_TimeSeries

Introduction

Globally, droughts are the biggest concern from climate change. Frequency and intensity of droughts has increased over the last century – Since 1900 Global droughts have affected 2 billion people and lead to more than 11 million deaths

Objectives

• Our Focus: Horn of Africa
• Identified the most appropriate Time-Series method to forecast drought in African countries, acting as a critical early warning for drought managements
• With the goal of maximizing the impact of our predictions we have decided to focus on the region most affected by droughts: Somalia, Ethiopia

Data

• Using Meteorlogical Drought indicator: SPEI
• Monthly SPEI measurements from the capitals of Somalia & Ethiopia 

https://spei.csic.es/home.html

• SPEI: Measures drought severity according to its intensity and duration, and can identify the onset and end of drought episodes
• SPEI takes into account both precipitation and potential evaporation in determining drought, therefore, SPEI captures the main impact of increased temperatures on water demand

Modeling

1.Benchmark Models: Naive, Mean, Seasoan Naive, Naive with drift

2.Exponential Smoothing

3.ARIMA, sARIMA

4.Spectral Analysis

5.VAR, Regression with ARIMA error

6.TBATS

7.ARCH/GARCH