Utilization of Satellite Images

forBasin Management

and Environmental Conservation in Mekong River Basin

Global Environment Department

29th October 2018

2Location Map Data source: JICA

Yunnan in China,

165,000

Myanmar, 24,000

Laos, 202,400Thailand, 184,240

Cambodia, 154,730

Vietnam, 65,170

Composition of Mekong River Basin

Total area: 795,000 km2

p Catchment area: 795,000 km2

p River length: 4,800 kmp Headwater: Tibetan Plateaup River mouth: State of Ben Tre in

Viet Namp General issues due to

deforestation:Increase of natural disasters such as flooding, drought, collapse of ecosystems, acceleration of global warming and etc.

1. Outline of the SurveyBasic Information

3

Project Name:Data Collection Survey on the Basin Management and Environmental Conservation in Mekong River Basin

Target country: Cambodia, Lao PDR, Thailand, Viet Nam and Myanmar

Main counterpart:Mekong River Commission (MRC) and 4 National Mekong Committees (NMC)

Objectives:1) To understand forest cover areas in Lower Mekong Basin (except for China)2) To clarify triggers of deforestations and issues of forest management3) To propose effective countermeasures and to recommend effective basin

management policy focusing on forestry sector in LMB

Project period: Dec 2017 to July 2019

1. Outline of the SurveyWork Procedure (Evaluation of Deforestation)

4

Collect satellite images

Predict the future forest cover area*

Prepare input data (land use data considering future deforestation)

Set the climate change scenario (sea level rise etc.)

⇒ Identify actual deforestation areas

Execute hydrological and hydraulic analysis

Evaluate the results(for identification of potential vulnerable areas)

*Where man-made deforestation is mainly evaluated while the effects of climate change inducted deforestation is not considered.

Prepare historical forest cover maps in Lower Mekong Basin (LMB)

Step 1 Step 2

Future (with deforestation)Baseline (existing)

Increase in peak discharge (flood risk)

Decline of base flow (drought risk)

Time

Rive

r flo

w

2. Utilization of Satellite Images

5

p Ideally, Preparing forest cover maps by compiling maps supplied by member countries is the best methods for the project purpose. However, there are certain difficulties for this such as lack of maps in some areas, inconsistencies in image resolutions covering the target areas and limited time for processing.

p Therefore, it is practical as well as optimum to utilize global observation data.

p ADCP (Asian Disaster Preparedness Center in Bangkok)* can provide time series Land Cover Map (based on Landsat imagery) for free.

p JICA, MRC and NMC agreed on the utilization of satellite images prepared by ADPC to analyze the historical changes of forest cover areas in LMB at Joint Seminar on June 2018.

*ADPC website: https://www.adpc.net/igo/?

Preparation of Forest Cover Maps

2. Utilization of Satellite ImagesBasic Condition for Preparation of Maps

6

p Duration: 16 years, from 2000 to 2015

p Images: Landsat 4 to 8

p Land Cover Classification:

(1) forest (8 class types), (2) urban area, (3) farm land, (4) paddy field, (5)

others (8 class types), (6) unknown e.g. blocking by clouds

p Data:

RGB (Blue, Green, Red) ,near infrared and middle infrared band of the Landsat

images

p Method:

Utilization of indices such as NDVI (Normalized Difference Vegetation Index),

NDWI (Normalized Difference Water Index), etc. with machine learning (random

forest algorithm) as supervised classification. It is automatic classification

p Reliability: Over 70% even without manual editing

2. Utilization of Satellite ImagesPreparation of Forest Cover Maps

7

Fig- Land Cover Map by ADPC

p Land Cover Classification: Total 19-class (1) forest (9 class types)(2) urban area(3) farm land(4) paddy field(5) others (7 class types)(6) unknown

2. Utilization of Satellite ImagesResult on Image Analysis (1)

8

Fig- Forest Cover Change Rates in LMB(from 2000 to 2015)

p The map indicates decrease/increase of provincial forest area.

p By using global observation data such as satellite images, change of forest cover area can be examined even for broad study area.

p Forest cover change rates and areas can be calculated.

2. Utilization of Satellite Images

9

0100000020000003000000400000050000006000000700000080000009000000

10000000

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Cambodia

Forest Area

Farmland

02000000400000060000008000000

100000001200000014000000160000001800000020000000

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Lao PDR

Forest Area

Farmland

0

2000000

4000000

6000000

8000000

10000000

12000000

14000000

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Thailand

Forest Area

Farmland

0

500000

1000000

1500000

2000000

2500000

3000000

3500000

4000000

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Viet Nam

Forest Area

Farmland

0

100000

200000

300000

400000

500000

600000

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

Myanmar

Forest Area

Farmland

Unit: ha Unit: ha Unit: ha

Unit: ha Unit: ha

Fig-2 Historical Changes of Forest and Agricultural Area (Farm Land and Paddy field)

Result on image analysis (2)

Conversion of agricultural land deems to be one of the reason of deforestation.

2. Utilization of Satellite ImagesUtilization of Image Analysis

10

p Flow regime, ‘hydrological and hydraulic patterns’ in the regions, depends on land use conditions.

p Future flow regime can be estimated by conducting hydrological and hydraulic simulation utilizing predicted land use conditions and the results of this will contributes to river basin management.

p In this JICA project, above analysis will be conducted considering future deforestation to evaluate flood and drought risk.

Increase in peak discharge (flood risk)

Decline of base flow (drought risk)

Future (with deforestation)Baseline (existing)

Time

Rive

r flo

w

3. Findings and recommendationsFindings and recommendations

11

p Trend of deforestation in LMB can be clarified by using satellite images.p By analyzing deforestation drivers (human activities such as commercial

tree logging, conversion to agricultural land, etc.) and historical forest area transition, deforestation near future in LMB can be assumed.

p For better prediction of deforestation in LMB, it is desirable to utilize the past forest cover maps, i.e. 1980’s and 1990’s.*Landsat 1 launched on 1972.

p The proposed method of preparation of forest cover maps in this JICA project can be applied to other countries and regions for better understanding land use changes in broad areas.

12

Thank you for your attention.