fr1.t09.5 - gis and agro- geoinformatics applications

Post on 23-Feb-2016

68 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

FR1.T09.5 - GIS and Agro- Geoinformatics Applications. Feature Analysis of Groundwater Discharge Points in Coastal Regions around Mt. Chokaisan, Japan by Using ALOS PALSAR DATA. Yoichi KAGEYAMA, Hikaru SHIRAI, and Makoto NISHIDA. Department of Computer Science and Engineering, - PowerPoint PPT Presentation

TRANSCRIPT

Feature Analysis of Groundwater Discharge Points in Coastal Regions around Mt. Chokaisan, Japan by Using ALOS PALSAR DATA 

FR1.T09.5 - GIS and Agro-Geoinformatics Applications

Yoichi KAGEYAMA, Hikaru SHIRAI, and Makoto NISHIDA

Department of Computer Science and Engineering, Graduate School of Engineering and Resource Science, Akita University, JAPAN

2

Table of Contents

1.Motivation2.Study area3.Data analysis4.Results and Discussion5.Summary

Submarine groundwater discharge

Rain or Snow

Groundwater flows

mountain

Sea

Submarine groundwater discharge

-A key role in linking land and sea water circulation

-Collecting water directly-Water quality, amount of discharge, and discharge location are quite different.

previously presented studyUse ALOS AVNIR-2 data

†1Y. Kageyama, C. Shibata, and M. Nishida, “Feature Analysis of Groundwater Discharge Points in Coastal Regions around Mt. Chokaisan by Using ALOS AVNIR-2 Data”, IEEJ Trans. EIS, Vol.131, No.10 (in press)

properties of the AVNIR-2 data acquired in different seasons were well able to retrieval the sea surface information†1.

spreading of the groundwater discharge

・ ALOS AVNIR-2 (Advances Visible and Near Infrared Radiometer type 2)are passive sensors- the data will be affected by clouds- the limited data are available.

・ ALOS PALSAR (Phased Array type L-band Synthetic Aperture Radar) are active sensor - we use the data regardless of the weather conditions.

Analyzes features of the groundwater discharge points in coastal regions by using the ALOS PALSAR data as well as the AVNIR-2 data

⇒ use of textures calculated from co-occurrence matrix ⇒ classification maps regarding the textures were obtained

with k-means. ⇒ comparison the PALSAR classification maps with the

AVNIR-2 ones.

Purpose

6

Table of Contents

1.Motivation2.Data used and study area3.Data analysis4.Results and Discussion5.Summary

Coastal region in Japan SeaAround the Mt.Chokaisan

Groundwater discharge at Kamaiso(Aug. 3, 2010)

Study area

Well known as the origin of Crassostrea nippona⇒  Groundwater discharge can affect the Its growth

ALOS PALSAR data

Winter data(Jan. 30, 2010)

Autumn data(Oct. 7, 2009)

ALOS AVNIR-2

Autumn data(Sep. 20, 2009)

Winter data(Feb. 25, 2010)

(R,G,B:band3,2,1)Band 1 0.42 ~ 0.50

blueBand 3 0.61 ~ 0.69

redBand 2 0.52 ~ 0.60

green Band 4 0.76 ~ 0.89

NIR(μm)

1270 MHz(L-band)

Survey points・ Kisakata beach(2 points)・ Fukuden(3points)・ Kosagawa beach(3points)・ Kosagawa fishing port (1point)・ Misaki(3points)・ Kamaiso(1point)・ Gakko River(2points)

Ground survey

Date: Aug 3, 2010

Comparison of sea and spring water in each water quality

  Sea water Spring water

pH 8.09 7.37

Dissolved oxygen 6.85mg/L 10.2mg/L

Electric conductivity 4.21S/m 0.002S/m

Salinity 27.6% 0%

Total Dissolved Solids 45.6g/L 0.1g/L

Sea water specific gravity 1.023sg 1.002sg

Water temperature 26.0℃ 13.3℃

Turbidity 7.78NTU 5.05NTU

●:Sea Water●:Spring water●:Sea and spring water

11

Table of Contents

1.Motivation2.Data used and study area3.Data analysis4.Results and Discussion5.Summary

Preprosessing-Geometric correction-Masking

Grayscale conversion-16,32,64,128,256,512

For PALSAR data

Textures computed from co-occurrence matrix

k-means algorithm to create the resulting classification

- second order conformal transformation - cubic convolution ⇒average RMS error was 0.41

吹浦

Winter data(Jan. 30, 2010)

Autumn data(Oct. 7, 2009)

Geometric correction

Preprosessing-Geometric correction-Masking

Grayscale conversion-16,32,64,128,256,512

Textures computed from co-occurrence matrix

k-means algorithm to create the resulting classification

Masked images

Masking

Land area-Various DNs-DNs are larger

A hydrology expert’s commentjudged from the scale of Mt. Chokaisan,the submarine groundwater discharge exist ranging from land regions to 500 meters offing.

500m

For PALSAR data

Preprosessing-Geometric correction-Masking

Grayscale conversion-16,32,64,128,256,512

Textures computed from co-occurrence matrix

k-means algorithm to create the resulting classification

-Noise reductionPALSAR data (2bytes)

⇒  16,32,64,128,256,512

  gray levels16

3264

128256

512

For PALSAR data Grayscale conversion

Preprosessing-Geometric correction-Masking

Grayscale conversion-16,32,64,128,256,512

Textures computed from co-occurrence matrix

k-means algorithm to create the resulting classification

Textures computed from co-occurrence matrix

小砂川

吹浦

小砂川

吹浦

Eight features-Mean, -Entropy, -Second moment, -Variance,- Contrast, - Homogeneity, - Dissimilarity, - Correspond

e.g., meanAverage the DNs of points around

),(1

0

1

0

jiPin

i

n

For PALSAR data

Preprosessing-Geometric correction-Masking

Grayscale conversion-16,32,64,128,256,512

Textures computed from co-occurrence matrix

k-means algorithm to create the resulting classification

k-means

小砂川

吹浦

小砂川

吹浦

For PALSAR data

The processing was ended: -the number of the maximum repetition amounted to 100 times,-moved pixels between clusters became 5% or less of the whole pixels.

k was set from 2 to 20.

17

Table of Contents

1.Motivation2.Data used and study area3.Data analysis4.Results and Discussion5.Summary

3×3 7×75×5 9×9 11 × 11

Filter size (e.g., mean)

(a)mean (d)variance(b)entropy (c)second moment

Select of feature

(f)homogeneity(e)contrast (g)dissimilarity (h)correlation

Select of feature

(16 gray levels; mean; K=7)

Autumn PALSAR results

air 18.7℃Wea water About 21℃

Spring water

About 10.5℃

†1http://www.jma.go.jp/jp/amedas/

Weather information during the data acquisition†1

large difference of temperature between spring water and air

Kosagawa

Misaki

Kamaiso

The red clusters exist in Kosagawa, Misaki, Kamaiso.The green and blue clusters are also formed⇒a spread of spring water.

8.2 ℃

Autumn and winter PLASAR results

In kosagawa, Amount of submarine groundwater discharge has been reduced in January to March.

Autumn data(16 gray levels; mean; K=7)

Kosagawa

Misaki

Kamaiso

Kosagawa

Misaki

Kamaiso

Winter data(16 gray levels; mean; K=7)

the red clusters are decreasing in winter

Autumn data

Kosagawa

Misaki

Kamaiso

Kosagawa

Misaki

Kamaiso

Winter data

  Autumn data Winter data

air 18.7℃ 2.4℃Sea water About 21℃ About 12℃

Spring water About 10.5℃ About 10.5℃†1http://www.jma.go.jp/jp/amedas/

Weather information at the data acquisition†1

the difference of temperature between Sea and spring water in the winter data is smaller.

Autumn and winter PLASAR results

(16 gray levels; mean; K=7)

10.5 ℃1.5 ℃

PLASAR and AVNIR-2 results in Autumn

Kosagawa

Misaki

Kamaiso

AVNIR-2 data(band1,2,3; k=7)

Kosagawa

Misaki

Kamaiso

The red clusters exist in Kosagawa, Misaki, and Kamaiso as well as the PALSAR classification results.

PALSAR data(16 gray levels; mean;

K=7)

PLASAR and AVNIR-2 results in Winter

AVNIR-2 data(band1,2,3;k=7)

Compared with the autumn data, the cluster of red is reduced

Kosagawa

Misaki

Kamaiso

PALSAR data(16 gray levels, mean, K=7)

Misaki

Kamaiso

Kosagawa

The conditions consistent with a decrease in the amount of submarine groundwater discharge in winter

Summary

This study has analyzed the features regarding the groundwater discharge points in the coastal regions around Mt. Chokaisan, Japan.  -The experimental results suggest that the Mean obtained from the co-occurrence matrix was good in extraction of the features of the groundwater discharge points from the ALOS PALSAR data. -The ALOS PALSAR data has the possibility of extracting the groundwater discharge points in the study area. -The k-means clustering results in the PALSAR and AVNIR-2 data agreed with the findings acquired by the ground survey.

Thank you for your attention!

top related