Develop statistical model to predict extreme precipitation through

20
Predict Sri Lanka Extreme Precipitation through El Nino Southern Oscillation R.M.S.P. Ratnayake PGIS/SC/M.Sc./ APS/10/20 MSc in Applied Statistics Post Graduate Institute of Science/ University of Peradeniya

description

Predict Sri Lanka Extreme Precipitation through El Nino Southern Oscillation

Transcript of Develop statistical model to predict extreme precipitation through

Page 1: Develop statistical model to predict extreme precipitation through

Predict Sri Lanka Extreme Precipitation through El Nino Southern Oscillation

R.M.S.P. RatnayakePGIS/SC/M.Sc./ APS/10/20

MSc in Applied StatisticsPost Graduate Institute of Science/

University of Peradeniya

Page 2: Develop statistical model to predict extreme precipitation through

Over view

• Introduction• Motivation and Background• Problem• Objectives• Hypothesis • Methodology• Organization• Time Frame

Page 3: Develop statistical model to predict extreme precipitation through

Introduction

• Sri Lanka economy mainly depend on Agriculture Industry.

• Sri Lankan Agriculture mainly depend on two monsoons.

• Therefore extreme precipitation changes the natural agriculture cycle.

• Expose to Disaster and Hazard potentials.

Page 4: Develop statistical model to predict extreme precipitation through

Problem

• Extreme Precipitation requires extra effort beyond basic Statistical Analysis.

• There is no proper model to predict Extreme Precipitation.

• Heavy Precipitation is a result of multiple courses.

• Sri Lanka climate data are spatially coherent.• Analysis required longer period precipitation

data

Page 5: Develop statistical model to predict extreme precipitation through

Motivation and Background

No of Affected Families 268544

No of Affected People 990471

No of Reported Deaths 18

No of Injuries 24

No of Missing People 3

No of Fully Damaged Houses 4216

No of Partially Damaged Houses 22186

Case Study : Early 2011 rainfall

Department of Metrology : Sri alnka

Page 6: Develop statistical model to predict extreme precipitation through

Objectives

• Identify Relationship between Extreme Precipitation and ENSO.

• Develop a model to relate Extreme Precipitation and ENSO.

• Validate defined model with recent data.

Page 7: Develop statistical model to predict extreme precipitation through

Hypothesis

• Null hypothesis that “There is a significant relationship between

extreme precipitation and ENSO behaviour.” • Against the alternative hypothesis that

“There is no significant relationship between extreme precipitation and ENSO behaviour. ”

Page 8: Develop statistical model to predict extreme precipitation through

Others Work• 2009 – Comparative analysis of indices of extreme

rainfall events: variations and trends from Mexico• 2008 - Predictability of Sri Lankan rainfall based on

ENSO• 1998 – ENSO influence on Intraseasonal Extreme

Rainfall and Temperature Frequency in the Contiguous United State: Implications for Long Range Predictability

• 2011 – Research on the Relationship of ENSO and the Frequency of Extreme Precipitation Events in China

Page 9: Develop statistical model to predict extreme precipitation through

Methodology : Overview

• Data Collection• Defining Threshold value• Analysis– Distribution of Data– Identifying Extreme Percentile– Spatial Distribution of Extreme Precipitation– Correlation Analysis– Time Series Analysis

Page 10: Develop statistical model to predict extreme precipitation through

Methodology : Data Collection

• Quarterly Cumulative Rainfall data • At least 50 years• 11 out of 21 Stations• Treating missing rainfall data : By Multiplying

each year value by multiplying N/(N-m) • NINO 3.4 – monthly data from 1951 to 2002

Page 11: Develop statistical model to predict extreme precipitation through

Methodology : Threshold value

• Gamma Distribution is used.• Rainfall above 95% percentile.• Separately calculated to Individual Stations

and All Island.

Page 12: Develop statistical model to predict extreme precipitation through

Methodology : Analysis

• Distribution of Data– Histogram– Normality Test

Page 13: Develop statistical model to predict extreme precipitation through

Methodology : Analysis

• Correlation Analysis between ENSO and Seasons

January - March

April - June

July - September

October - December

Page 14: Develop statistical model to predict extreme precipitation through

Methodology : Analysis

• Correlation Analysis between ENSO and Different Stations and All Island

Anuradhapura Mannar

Batticoloa Nuwara Eliya

Colombo Puttalam

Hambanthota Ratmalana

Kankasanthure Trincomalee

Katunayake

Page 15: Develop statistical model to predict extreme precipitation through

Expected Results End of the Research

• In JFM/ AMJ/ JAS/ OND Extreme Precipitation days in Anuradhapura/ Batticoloa/ Colombo/ Hambanthota/ Kankasanthure/ Katunayake/ Mannar/ Nuwara Eliya/ Puttalam/ Ratmalana/ Trincomalee/ All Island are significantly More or Less Frequent in El Nino than La Nino

Page 16: Develop statistical model to predict extreme precipitation through

Statistical Software

• R• Excel

Page 17: Develop statistical model to predict extreme precipitation through

Organization

• Irrigation Department • Department of Meteorology of Sri Lanka• Foundation of Environment and Climate

Technology• Institute of Post Graduate Studies – University

of Peradeniya.

Page 18: Develop statistical model to predict extreme precipitation through

Time LineRequirement Analysis

Data Gathering

Data Arranging

Study Existing Approaches

Analyzing DevelopingModel

Testing and Validating

Report preparation

Presentation

Week1

Week2

Week3

Week4

Week5

Week6

Week7

Week8

Week9

Week10

Week11

Week12

Page 19: Develop statistical model to predict extreme precipitation through

Acknowledgement

• Dr. Lareef Zubair at Foundation of Environment and Climate Technologies, Dhigana.

• Eng. R.M.W. Ratnayake at Director (Water Resources) Ministry of Irrigation and Water Resource Management.

• Post Graduate Institute of Science University of Peradeniya

Page 20: Develop statistical model to predict extreme precipitation through

Thanking you

Weather is a great metaphor for life - sometimes it's good, sometimes it's bad, and there's nothing much you can do about it but carry an umbrella. ~Terri Guillemets