Dr. Yiyun Chen [email protected] School of Resource and ...

Dr. Yiyun Chen [email protected]

School of Resource and Environmental Science Wuhan University 3 December, 2014

Outline

• Case study 1 (Geographical space)

– Mapping of Cu and Pb Contaminations in Soil Using Combined Geochemistry, Topography, and Remote Sensing: A Case Study in the Le’an River Floodplain, China

• Case study 2 (Geographical and spectral spaces)

– Explore the spatial dependence of soil properties in both geographical space and spectral space

1 Background

Objectives

Materials and Methods

Results

Conclusion Mapping of Cu and Pb Contaminations in Soil Using Combined Geochemistry, Topography, and Remote Sensing: A Case Study in the Le’an River Floodplain, China

Background

Farmland Wetland

Mines

Largest fresh water lake in China

Case study 1

Le’an River Floodplain

Annual rainfall 1900 mm

Objectives

• To map the total copper and lead contents in top soil within a buffer zone of 200 m along the Le’an River and its branches.

• To understand of the roles that topography and land use play in soil pollution in mining area and its downstream region.

Case study 1

Better mapping for a better understanding of the “story”

Flowchart of mapping total Cu and Pb contaminations

ALOS image

October 24,

2009

Total Cu/Pb

content at

points

GDEM

ALOS image

May 10,

2009

ALOS

image of

study area

Mosaic

Interpretation

Built up areaAgricultural

areaMining area

River 200 m bufferingFloodplain

soil area

Interpolation (IDW)

Total Cu/Pb

content in

area

Masking

Total Cu/Pb

content in

floodplain soil

Overlapping

GIS data

Watershed

delineation

Field survey

records

Source identification

and risk assessment

Map for total

Cu/Pb

distribution

Cu and Pb content at

sample sites

Digital Elevation

Model (DEM)

Built up area

Mining area

Agricultural area

Water area

Cu and Pb in soil within a buffer zone of 200m

along rivers

Interpolation

Interpretation

Watershed delineation

Jishui watershed

Dawu watershed

Dexing Copper Mine Tailings Watershed

Yinshan Lead and Zinc Mine Watershed

Overlapping all the features

Field sampling and chemical analysis

• A total of 71 top layer (0–15 cm) soil samples were collected along the Le’an River and its branches.

• About 75% of the samples (53 samples) were collected in the middle and upper reaches of the Le’an River where the Dexing copper mine and Yinshan lead-zinc extraction facility are located

Spatial distribution of Cu

Spatial distribution of Pb

Spatial distribution of Cu and Pb

• With 3D maps generated from multi-source data, the potential sources and transportation routes of Cu and Pb pollutants were inferred through visual inspections and field survey records.

• The inference was thereafter confirmed by mapping together the watersheds of the Dawu River, Jishui River and the mining area, the Cu/Pb content in soil, and sample sites.

Conclusion

• Bearing the idea “better mapping means better and easier understanding”, this study demonstrated the arts of utilizing multi-source data in the mapping of environmental pollution as well as in the understanding of the role that topography plays in the transportation of pollutants.

Case study 1

2 Background Background

Explore the spatial dependence of soil properties in both geographical space and spectral space

Remarks

Background

• Soil spectroscopy

– Visible and near infrared spectra

350 nm 760 nm 2500 nm

Soil

Incident sunlight reflected (reflectance spectra) by soil contains information of soil components (e.g. minerals, organics, moisture and particle size)

halogen lamp

Case study 2

Total nitrogen

If soil can talk…

Farmers Soil fertility

Pedologist Soil properties

Judge by colour (Qualitative) In order to quantitatively understand the words of soil, we need models

Soil classification (Qualitative)

the reflectance spectra in the visible and near infrared region could be some of its words

A rainbow of soil is under our feet; red as a barn and black as a

peat. It’s yellow as lemon and white as the snow; bluish

gray…… --- F.D. Hole, A Rainbow of Soil Words, 1985

Soil Colour

Use cross validation for components selection

Dataset

Soil Sample

Spectra Measurement

Chemical analysis

400 401 … 2450

Sample 1

Sample 2

…

Sample m

SOM

X Y

400 401 … 2450

Sample 1

Sample 3

…

SOM

400 401 … 2450

Sample 2

Sample 4

…

SOM

Calibration set

Validation set

multicollinearity problem

W2

r=0.88 p<0.01

W1

Principal component regression (PCR)

Partial least square regression (PLSR) Explanatory variables are linearly correlated

Geographical space

Spectral space

GPS

Spectrometer

Soil Samples

Soil sample location in different spaces

soil sample with similar spectra would have similar soil properties

Soil samples that are spatially nearer usually tend to have similar soil properties

soil samples that are nearer in the spectral space tend to have similar soil properties

Case study 2

• Objective

–To explore the spatial dependence of soil properties in both geographical space and spectral space

Paddy field Irrigated land

• Land use types could have spatial randomness. • For different land use types, soil properties (e.g. soil

organic content) could be from different populations . • Those soil properties potentially influenced by land use

types could also have randomness in geographical space .

Dataset

Spatial dependence in different spaces

SOM N Fe P

Permutations: 999

Ge

ogr

aph

ical

sp

ace

Sp

ect

ral

sp

ace

p<0.01 p<0.01

p<0.01 p<0.01 p<0.01 NO

NO NO

Usually, observation dependence is spatial dependence in geographical space; but sometimes, it could be dependence in spectral space

Spatial dependence in different spaces

Permutations: 999

Ge

ogr

aph

ical

sp

ace

Sp

ect

ral

spac

e X, p<0.01 Y, p<0.01

PC5 p<0.01

PC4 p<0.01

PC3 p<0.01

PC2 p<0.01

PC1 p<0.01

Further comparisons

SOM N Fe P PC1 PC2 PC3 PC4 PC5

Moran’ I 0.026 0.105 0.157 0.141 0.184 0.321 0.240 0.617 0.258

SOM

SOM

N

N Fe

Fe

P

P PC1

PC1

PC2

PC3

PC4

PC5

PC2 PC3 PC4 PC5

SOM

N

Fe

P

PC1

PC2

PC3

PC4

PC5

SOM N Fe P PC1 PC2 PC3 PC4 PC5

Pearson’s r p value

Why the PCs of spectra have spatial dependence in geographical space?

Their correlations with soil components (SOM, mineral content) might be an explanation, but…

• The original idea was to explore the independence of observation, and examine the autocorrelation of model residuals.

• There is no spatial autocorrelation in the dataset presented, when the soil chemical properties were estimated using OLS with PCs of soil spectra.

• BUT for the China Soil Spectra Library dataset…

Why introduce spectral space?

China Soil Spectra Library

• 1581 soil samples from 16 soil types

• Spectra recorded in the range 400-2450 nm

• Soil organic matter content for all the samples

Reference:

China Soil Spectra Library

OLS

R-squared : 0.172

Weights were created using PC1 and PC2

Spatial error model

R-squared : 0.180

OLS R-squared : 0.172

More studies are needed

• Y. Liu, Q. Jiang, T. Fei, J. Wang, T. Shi, K. Guo, X. Li, Y. Chen*, 2014. Transferability of a Visible and Near-Infrared Model for Soil Organic Matter Estimation in Riparian Landscapes. Remote Sensing. 6(5), 4305-4322.

• Y. Liu, Q. Jiang, T. Shi, T. Fei, J. Wang, G. Liu, Y. Chen*, 2014. Prediction of total nitrogen in cropland soil at different levels of soil moisture with Vis/NIR spectroscopy. Acta Agriculturae Scandinavica, Section B–Soil & Plant Science 64(3), 267-281.

• T. Shi, Y. Chen, H. Liu, J. Wang, G. Wu, 2014. Soil Organic Carbon Content Estimation with Laboratory-Based Visible-Near-Infrared Reflectance Spectroscopy: Feature Selection. Applied Spectroscopy. 68(8), 831-837.

• T. Shi, Y. Chen, Y. Liu, G. Wu, 2014. Visible and near-infrared reflectance spectroscopy—An alternative for monitoring soil contamination by heavy metals. Journal of Hazardous Materials. 265(0), 166-176.

• J. Wang, L. Cui, W. Gao, T. Shi, Y. Chen, Y. Gao, 2014. Prediction of low heavy metal concentrations in agricultural soils using visible and near-infrared reflectance spectroscopy. Geoderma. 216(0), 1-9.

• X. Peng, T. Shi, A. Song, Y. Chen, W. Gao, 2014. Estimating Soil Organic Carbon Using VIS/NIR Spectroscopy with SVMR and SPA Methods. Remote Sensing. 6(4), 2699-2717.

• T. Shi, H. Liu, J. Wang, Y. Chen, T. Fei, G. Wu, 2014. Monitoring Arsenic Contamination in Agricultural Soils with Reflectance Spectroscopy of Rice Plants. Environmental Science & Technology. 48(11), 6264-6272.

Some publications

• K. Guo, Y. Liu, C. Zeng, Y. Chen, X. Wei, 2014. Global research on soil contamination from 1999 to 2012: A bibliometric analysis. Acta Agriculturae Scandinavica, Section B–Soil & Plant Science. 64(5), 377-391.

• T. Shi, L. Cui, J. Wang, T. Fei, Y. Chen, G. Wu, 2013. Comparison of multivariate methods for estimating soil total nitrogen with visible/near-infrared spectroscopy. Plant and soil. 366(1-2), 363-375.

• Y. Chen, Y. Liu, Y. Liu, A. Lin, X. Kong, D. Liu, X. Li, Y. Zhang, Y. Gao, D. Wang, 2012. Mapping of Cu and Pb Contaminations in Soil Using Combined Geochemistry, Topography, and Remote Sensing: A Case Study in the Le’an River Floodplain, China. International Journal of Environmental Research and Public Health. 9(5), 1874-1886.

• Y. Liu, Y. Chen, 2012. Estimation of total iron content in floodplain soils using VNIR spectroscopy – a case study in the Le'an River floodplain, China. International Journal of Remote Sensing. 33(18), 5954-5972.

• Y. Liu, Y. Chen, 2012. Feasibility of Estimating Cu Contamination in Floodplain Soils using VNIR Spectroscopy—A Case Study in the Le’an River Floodplain, China. Soil and Sediment Contamination: An International Journal. 21(8), 951-969.

• Y. Liu, D. Liu, Y. Liu, J. He, L. Jiao, Y. Chen, X. Hong, 2012. Rural land use spatial allocation in the semiarid loess hilly area in China: Using a Particle Swarm Optimization model equipped with multi-objective optimization techniques. Science China Earth Sciences. 55(7), 1166-1177.

Some publications

Thank you for your attention