Change Detection Analysis of Eco-Sensitive Area using Remotely Sensed Data By Abhijat Arun Abhyankar.

Change Detection Analysis of Eco-Sensitive Area using Remotely Sensed Data

ByAbhijat Arun Abhyankar

Outline of the talk

• Remote Sensing• Introduction and Objective of the study• Study Area• Data• Methodology• Results and Discussion• Future Work

Remote sensingRemote Sensing is the science and art of making measurements of an object or

environment without coming into physical contact with target

Introduction and Objectives

The International Geosphere-Biosphere Programme (IGBP) and the Human Dimensions of Global Environmental Change Programme (HDP) have acknowledged the importance of land use change studies in developing our understanding of global environmental change

Satellite images have inherent advantages

1)Spatial 2)Temporal revisit3)Images of inaccessible areas4)Time less than survey method

This work depicts changes in land-use/ land-cover for the area covering ten kilometre radius around the limestone mining site of Lafarge Surma Cement Company, located in Shella, (situated about 96 km away to the south of Shillong, the Capital of Meghalaya).

Study Area and Data

IRS P6 LISS III: January 7, 2008 and March 9, 2010, field visit

The study are considered is an area of 10 km radius (aerial distance) from the 25o11’18’’ N latitude & 91o37’28’’ E Longitudes. This is mine site of Lafarge Umiam Mining Pvt. Ltd.

The area is entirely rural and sparsely populatedCommunity Development (CD) blocks of Shella Bholaganj and Mawsynram both under the district jurisdiction of East Khasi Hills. The village Nongtrai is about 2.5 km away from the mine area while Shella Bazar and Pyrkan are within the radius of 2 km from the mining zone. The nearest township is at Cherrapunji, known to be the ‘rainiest’ place of the world.

IRS P6 satellite

Launch date October 17, 2003Launch site SHAR, Sriharikota

Launch vehicle PSLV-C5

Payloads LISS-4, LISS-3, AWiFS-A, AWiFS-BOrbit Polar Sun Synchronous

Orbit height 817 kmOrbit

inclination 98.7o

Orbit period 101.35 minNumber of

Orbits Per day 14

Local time of equator crossing

10:30 am

Repetivity (LISS-3) 24 days

Revisit 5 daysLift-Off mass 1360 kg

Attitude and orbit control

3-axis body stabilised using Reaction Wheels, Magnetic Torquers and Hydrazine

Thrusters

Power Solar Array generating 1250 W, Two 24 Ah Ni-Cd batteries

Sensor Resolution Colour

LISS-IV Mono 5.8 m black and white

LISS-III 23 m multispectral

AWiFS 60 m multispectral

Sensor LISS-IIIResolution 23 m

Swath 127 km (bands 2, 3, 4)134 km (band 5 -MIR)

Repetitive 25 daysSpectral Bands 0.52 - 059 microns (B2)

0.62 - 0.68 microns (B3)0.77 - 0.86 microns (B4)1.55 - 1.7 microns (B5)

MethodologyField visit to study area

Identification of different landcover classes and recording these using GPS

Procurement of cloud free satellite data-geocoded

Identification and extraction of sample landcover classes on the satellite imagery (six landcover classes were identified namely, dense forest, sparse forest, barren land, crops, water and dry channel)

Dense forest/Medium forest: canopy cover greater than 40%Sparse forest: canopy between 10 to 40% of canopy coverScrub land: less than 10% of canopy cover (Forest Conservation Act, 1980)

Using supervised classification with Maximum likelihood estimator, preparation of landcover map-temporally

Change detection analysis

False color composite (FCC) of sample landcovers

Scatter plots of sample landcovers

Barren Land

CropDense Forest

Dry Channel

Sparse Forest water

Supervised Classification with Maximum Likelihood Estimator

This method assumes the training dataset selected for each landcover distribution-normally distributed.

Using these parameters, probability of unknown pixel falling in various classes is calculated. We have identified six landcover classes.

Hence for each of the pixel-we obtain have six probability value.

The higher probability value-class is assigned to unknown pixel

Mathematically,

False Color Composite of January 7, 2008 using IRS

P6 LISS III

False Color Composite of March 9, 2010 using IRS P6 LISS III

Landcover map of January 7, 2008 using IRS P6 LISSIII image

Classes 7-Jan-08

Crop land 20.3

Dense forest 127.7Sparse forest/ Scrub

land 90.4

Water 1.3

Dry channel 0.7

Barren land 74.4

Total Area 314.8

Landcover map of March 9, 2010 using IRS P6 LISS III image

Classes 9-Mar-10

Crop land 30.2

Dense forest 137.0Sparse forest/Scrub

land 63.8

Water 6.7

Dry channel 1.5

Barren land 75.6

Total Area 314.8

Results and Discussion

• Dense forest area-increased• Barren land area-no change• Sparse forest/Scrub land-reduced• Crop-increased• Water-increased• Dry Channel-increased

Future work

Accuracy Assessment of landcover map

Discriminant analysis for landcover classification

ANN for landcover classification

Comparison of Landcover results with 2006

THANK YOU and

QUESTIONS