Predicting and mapping soil properties using proximal/remote sensing by Zhou Shi, Ji W J, Cheng J L, Li X, Zhou L Q, Huang Q T, Huang M X, Yang HQ

Predicting and mapping soil properties
using proximal/remote sensing
Professor Zhou SHI（史舟）
Ji W J, Cheng J L, Li X, Zhou L Q, Huang Q T, Huang M X, Yang H Q
Institute of Agricultural Remote Sensing and Information System
Zhejiang University, Hangzhou 310058, shizhou@zju.edu.cn
soil remote sensing and information technology http://agri.zju.edu.cn
"Advancing the Science and Technology of Soil Information in Asia —
Launch of the Global Soil Partnership’s Asia Soil Science Network and
GlobalSoilMap.net East Asia Node"

1、Introduction
2、Soil Proximal Sensors
3、Soil Remote Sensing
Outlines

Wet chemistry methods to extract soil properties
are cost effective and time consuming
Accordingly, non destructive technologies, such as
field proximal sensors or remote sensing methods
can provide important vantages.
Accordingly, non destructive technologies, such as
field proximal sensors or remote sensing methods
can provide important vantages.
Precision
agriculture
Quantitative soil-
landscape modeling
Global soil C
monitoring
…
Soil map (soil type or soil properties)
Fine resolution | Grid -based | Functional properties
Soil map (soil type or soil properties)
Fine resolution | Grid -based | Functional properties
How to develop methods that use the
minimum number of soil samples possible to
minimize the costs

The capacity and potential of the proximal sensors / remote
sensing on monitoring and mapping of some key functional
properties of soil , such as soil moisture, organic matter and
carbon.
effect on the interactions
between soil and plants
related to soil quality
plays a leading role in surface
water and energy balance
available for plant utilization

Roger Bales, University of California
Soil moisture (土壤水
分）
Variability in the moisture held in soil is a consequence of the
continuous exchange of water between the land and atmosphere.
Although soil only holds a small percentage of the total global
budget, soil moisture plays an important role in the water cycle.
In the absence of a well-integrated program of sampling and
scaling, the limited areal coverage over large land surface areas
makes it difficult to use in situ measurements for studies more
extensive than field-scale.

Satellite spots soggy soil. Nature news. 2010.6.30
Monitoring and mapping soil moisture by satellite
Monitoring global water cycle
Remote sensing can be advantageous owing to its continuous temporal and spatial
coverage. It is the most promising approach for soil moisture estimation involving large
areas
Soil Moisture and Ocean Salinity satellite (SMOS)
Launched in November 2009, the European Space Agency. SMOS makes global maps of
soil moisture every three days to improve our understanding of the water cycle, benefit to
agriculture and water resource management, climate modeling, forecasting floods.

Post W. M. et al. Soil carbon pools and world life zones. Nature, 1982,298, 156 - 159
Soil carbon （土壤碳）

www.globalsoilmap.net
Global soil map 全球土壤图库
Ten soil properties will be predicted at each location.
These are:
（1）total profile depth (cm)
（2）plant exploitable (effective) soil depth(cm)
（3）organic carbon (g/kg)有机碳
（4）pH
（5）sand (g/kg),（6）silt (g/kg)
（7）clay (g/kg)
（8）gravel (g/kg) 砾石
（9）bulk density (mg/m3) 容重
（10）available water capacity (mm)有效水容量
Additional soil properties ：
（1）ECEC (cmolc/kg) 有效阳离子交换量
（2）EC (dS/m) 电导率

Sanchez P A et al. Digital soil map of the world. Science. 2009,325:680-681

Hengl T. 2011:15-37, Global Soil Information Facilities (book)
Predictions of organic matter
(Meuse data) visualized using
polygons (KML)

Gerbbers R and Adamchuk V I. Precision agriculture and food security. Science. 2010,327:828-831
Precision agriculture and food security
粮食安全与精确农业
Variation in soils and crops
Soil and crop sensing
Apply the right treatment in
the right place
at the right time

Gerbbers R and Adamchuk V I. Precision agriculture and food security. Science. 2010,327:828-831
Relevant properties of soil productivity
are soil moisture, clay content, organic
matter content, nutrient availability,
pH, bulk density.
Precision agriculture and food security
粮食安全与精确农业

2.2 Soil spectral library in laboratory
Standard, share 标准/规范/共享
2.3 Soil proximal sensors in field
From lab to field 土壤光谱测试手段走向农田
2.1 Principal of electromagnetic spectrum

NIR VIR FIR ESR NMR
范围
Scale
γ射线
γ ray
X射线
X ray
紫外
UV
可见
光
Vis 红外 Infrared 微波
microwave
射频
Radio
frequency
相互
作用
Inter-
action
核的跃
迁
Nuclear
transition
内层电子
跃迁
Electronic
transition
外层电子跃
迁
Electronic
transition
分子振动
Molecular vibrations
分子转动
Molecular
rotation
磁场中的
自旋取向
Magnetic
field spin
orientation
780nm 2.5μm
波长
Wavelength
12820 3958波数cm-1
Wave number
360nm
激光感生击
穿光谱LIBS
Laser induced
spectrum
X射线荧光
光谱XRF
X-ray
fluorescence
spectrum
漫反射光谱DRS
Diffuse reflective
spectrum
Soil heavy metal
Soil organic matter,
texture, clay,
moisture
Soil roughness,
moisture, salinity
发射率光谱
Emission
spectrum
土壤介电常数
Soil dielectric
constant
依据土壤发射、吸收电磁波以及土壤与电磁
辐射的相互作用来分析土壤特性的技术。
10nm0.1nm 15μm
Soil temperature,
moisture
Electromagnetic spectrum
电磁波波谱
Spectral signatures of materials are defined
by their reflectance or absorbance, as a
function of wavelength in the electromagnetic
spectrum

Viscarra Rossel et al. Advances in Agronomy, 2011
Electromagnetic spectrum
电磁波波谱

土壤遥感与信息技术 http://rs.zju.edu.cnShepherd. Development of reflectance spectral libraries for characterization of soil properties. SSSAJ 66 (2002) 988–998
Soil reflectance spectroscopy
Diffuse reflectance spectroscopy
(DRS)
Procedures of development and
application of soil spectral library
Procedure of soil spectral library

不同土壤的差异能否被漫反射光谱(DRS)所探测?
Soil variation can be characterized using Diffuse
reflectance spectroscopy (DRS)?
不同土壤的差异能否被漫反射光谱(DRS)所探测?
Soil variation can be characterized using Diffuse
reflectance spectroscopy (DRS)?
能否利用现有数据建立全球土壤特性光谱校准模型?
How can we make Global Soil Spectral Calibration of
soil properties using existing data
能否利用现有数据建立全球土壤特性光谱校准模型?
How can we make Global Soil Spectral Calibration of
soil properties using existing data
GSSL (VIS/NIR, mid-IR)全球土壤光谱库
2008，Raphael A. Viscarra Rossel（CSIRO Land and
Water）organized the “The Soil Spectroscopy Group”.
◎ 样品的典型性； ◎测试的规范性； ◎ 应用的可靠性
The global soil spectral library (GSSL)

16000+ spectra come from 90+ countries
All measured with vis-NIR and some with mid-IR
from Raphael VISCARRA ROSSEL
The global soil spectral library (GSSL)

Spectroscopic calibration to soil TOC from GSSL
Calibrations to soil OC after
clustering PLS scores
from Raphael VISCARRA ROSSEL

Supported by Ministry of education of China (MOE)
First stages, The soil spectral library of China (SSLC)collected more than
2000 soil samples, built the linking network among several universities
and institutions in China.
locati
ons
Soil
type
Sample
s
河南潮土 48
水稻土 78
四川
紫色土 56
黑土(1) 418
东北
黑土(2) 90
江西红壤 195
陕西褐土 36
西藏 44
浙江紫色土 12
浙江水稻土 449
浙江滨海盐土 153
湖南红壤 9
新疆水稻土 193
…
The soil spectral library of China (SSLC)

4439米海拔，高山草甸(alpine meadow)
3893米海拔，高山冷杉(alpine fir) 3426米海拔，青稞田(hullessbarley)
4161米海拔，中大试验站，灌木(shrub)
Soil sampling in Tibet （色季拉山、雅鲁藏布江等地）

Spectral reflectance of some selected soils in China

Relationship between Soil OM and spectral reflectance

土壤遥感与信息技术 http://rs.zju.edu.cnViscarra Rossel and Behrens, 2010, Geoderma
(1) Linear regressions
•Partial least squares regression (PLSR)
•Multiple linear stepwise regression (MSLR)
•Principal component regression (PCR)
Multivariate calibrations
(2) Data mining techniques
•Multivariate adaptive regression splines
(MARS)
•Boosted regression trees (BT)
•Random forest (RF)
•Support vector machines (SVM)
•Artificial neural networks (ANN)
Data analysis & model calibration

A hand portable, remote sensing FTIR spectrometer (Model 102F) is a
type of Fourier transformed infrared spectrometers produced by
Design& Prototypes company.( Design for in situ measurements)
美国Design& Prototypes公司便携式傅立叶变换红外光谱仪（102F型现
场及工业用FT-IR光谱仪）
Thermal Infrared Spectral Property of Soil (土壤TIR特性)

Emissive spectra feature for different soil types
不同类型土壤的发射光谱特征
Emissive spectra feature for different soil water
不同含水量土壤的发射光谱特征
Emissive spectra feature for
different soil’s OM contents
不同有机质含量土壤的发射光谱特征
Relationship between the emissive spectral and soil property

Soil microwave dielectric properties
Soil dielectric constant was measured and analyzed using Vector
Net Analyzer at different conditions
Coaxial probe method directly calculates the microwave complex
permittivity of a medium by measuring the reflection coefficient of the
probe terminal

Soil samples preparation
Install the test-bed Measure the dielectric constant
Set parameters:
Frequerency:500MHz-
12GHz,concluding L S X C
bands
Bands: 800
Probe pressure: according the
drying condition of the samples

The relationship of the soil dielectric constant and the frequency
（a）（b）
The real part of the dielectric constant is increasing with the increase of
the soil water content and decreasing with the decrease of the frequency.
This trend is obvious at high water content but hardly changing at low water
content.
The imaginary part of the dielectric constant increase with the increase
of the water content at higher frequency.

The relationship of the dielectric
and soil volumetric water content
simulated by Topp model.
The relationship of soil dielectric constant and soil volumetric
water content by Dobson model at different frequencies.

Program:
In situ Characterization of SOM of Paddy Soil with vis-NIR Diffuse
Reflectance Spectroscopy (DRS)
Objective:
To evaluate the feasibility of VNIR-DRS for in situ quantification of
organic matter content of paddy soil
Application of field DRS technology
田间反射光谱技术应用

Paddy soil in China
China is the largest producer of rice in the world

1. 20cm-deep-transect 2. optimize by spectralon panel 3. scanned by spectrometer
4. Moisture measurement5.Soil collection

Predicted model by MLSR
method, Selected bands （454,469,404nm)
n=75, R2=0.701, RMSE=0.215
Soil OM Range [6.02-22.4 g/kg]

Vis-NIR DRS shank module（Veris)
On-the-go measurement by
vis-NIR DRS
Vis-NIR DRS
probe
module(Veris
P4000)

Key fields:
Precision agriculture, Agricultural IT
Sensors and intelligent information processing
Area of priority:
Agricultural Internet of things (IOT)
Intelligent equipment or sensors
National Program for Science and Technology Development in
China (2006‐2020)
Soil proximal technology and sensors

Agricultural sensors
For laboratory, in situ field, on-the-go measurement
Static and dynamic observation
Robust, low power consumption, wireless communication
Plant
•Nutrient index: Chlorophyll , water, N, P, K, trace element etc.
•Physiology : Protein, amino acid, enzyme etc.
•Ecology : LAI, VI, shape, Leaf temperature, stem flow etc.
Soil
• Physics : moisture, water potential, temperature、tension,
compaction, EC, salinity
• Chemistry: pH, organic matter, N, P, K, heavy metal
Environment •Temperature, humidity, Solar radiation , CO2
…………
National High Technology Research and Development Program
of China (863 Program) ------- “Digital Agriculture”

Outlines
1、Introduction
2、Soil Proximal Sensors
3、Soil Remote Sensing

Predicting and mapping soil properties using proximal/remote sensing by Zhou Shi, Ji W J, Cheng J L, Li X, Zhou L Q, Huang Q T, Huang M X, Yang HQ

Shepherd. Development of reflectance spectral libraries for characterization of soil properties. SSSAJ 66 (2002) 988–998
Going from point (spectrometer) to image spectrometry
(remote sensing) is not only a journey from micro- to macro-
scales but also a long stage that encounters problems such
as dealing with data having a low signal-to-noise level,
atmospheric interference, shade and shadow effects,
mixtures of materials within pixels, the BRDF effect,
variation in soil moisture content, and more.
(大气干涉、地面覆盖物阴影、粗糙地表的二向反射影响、混合像元、表土
水分空间异质性等)
Problems to soil remote sensing

3.2 Soil moisture remote sensing
3.3 Application of soil remote sensing
3.1 Soil OM/Carbon remote sensing

CASI 1500
288 bands
0.4-1.05μm
*ATLAS： digital aerial images by NASA ATLAS sensor
* AVIRIS: Airborne Visible / Infrared Imaging Spectrometer
Soil OM/Carbon remote sensing
土壤有机质/碳遥感获取

b)田间水分(Filed moisture,FM), (c)有机质(OM), (d) 饱和含水量(Saturated moisture, SM)
Ben-Dor,E. et al. l. Int. J. Remote Sens. 2002, 23, 1043–1062.
DAIS-7915
0.4-12.6 μm
79 bands
Hyper-spectral aerial remote sensing
高光谱航空遥感

Predicted percentages of organic (left side) and inorganic (right side) carbon for a spatial
subset of the Guadalentin HyMap image. The black areas represent vegetated surfaces that were
excluded from the analysis.
Thomas, J. et al. 2007. Proceedings of 5th EARSeL Workshop
Validation results for the organic (left side) and inorganic (right side) carbon contents
estimated from the HyMap image data. The red lines show the linear regression between the
modelled and measured values. The dashed lines indicate a linear 1:1-relationship.
Hymap, 0.45-2.5 μm
128 bands
Detecting soil OC and IC
土壤有机和无机碳提取

Sensor types Methods Advantages and disadvantages
Optical
remote
sensing
Statistical relational model-based The relationship with the soil moisture
spectral characteristics is not closely.
interfered by vegetation layer ,weak ability
to penetrate the surface , seriously
affected by atmosphere and clouds ,
rarely applied in soil moisture detection
Thermal
infrared
remote
sensing
(1)Bare soil or low vegetation coverage
areas：Thermal inertia
(2)High vegetation coverage areas：
TVDI,CWSI
With the advantages of macro and high
time resolution, it has superiorities in the
large range of soil moisture detection and
river basin drought early warning.
Passive
microwave
remote
sensing
(1)Empirical model: Inversion soil moisture
by using brightness temperature and
establish a linear relationship
(2)Theoretical model based on radiative
transfer equation
With the definite rationale, high time
resolution and relatively-low spatial
resolution(about 10-50km)，it fits soil
moisture monitoring by remote sensing
on the regional even global scales.
Active
microwave
remote
sensing
(1) Theoretical model：Kirchhoff ,SPM,IEM
(2) Semi-empirical model & Empirical
model： Oh Model, Dubois Model, Shi
Model
(3) The model of the high vegetation
coverage areas ：Water-Cloud Model,
MIMICS Model
It has the quality of all-day and all-
weather resistance to mist and weather,
and high spatial resolution relative to
passive microwave sensor , making the
advantage on Soil moisture detection.
Soil Moisture Retrieving by Remote Sensing

HyMap image from July 30th 2004 (RGB from
bands 13/7/2). The red line depicts the
boundaries of soil moisture measurements in
the field
Haubrock, S. et al. 2005. Proceedings of 4th EARSeL Workshop
Surface soil moisture from July 30th 2004 determined
from (a) FDR（频域反射仪） field measurements
and (b) HyMap‐based ratio calculation (1288 vs.
1515 nm) based on the regression function for
quaternary sand. The dashed line shows the part of
the area being covered with quaternary sand
Hyper-spectral aerial remote sensing
高光谱航空遥感
Hymap
0.45‐2.5 μm
128 bands

Supported by NSFC(40871100), 2009‐2011
Multi-temporal SAR images - False color compostie
R:2009-2-15 PALSAR HH G:2009-5-14 PALSAR HH
B: 2009-8-18 PALSAR HH (Polarization)
Hangzhou Bay
杭州湾
曹
娥
江
Hydrology
station
Weather
station
Active microwave remote sensing
主动微波遥感
Detecting soil moisture based
on ALOS/PALSAR

Equipments:
TDR300 (USA, ST)
W.E.T Sensor (UK, Delta‐T)
MiniTrase (USA, SEC)
eKo PRO (USA, XBOW)
Ground measurement in situ
Synchronous field
experiment
Soil moisture(%)
Nov 21th 2010
South Area
Soil moisture(%)
Nov 21th 2010
North Area

Oh model
Inversion model 反演模型
Measured soil moisture(%)
Measured soil moisture(%)
HVinversionofsoilmoisture(%)HHinversionofsoilmoisture(%)
Soil moisture(%)
High 80.00
Low 0.00

杭州湾
曹
娥
江
Thermal infrared remote sensing
热红外遥感
LST_EVI scatter diagram
Eigenspace
散点图、特征空间
Very wet
Wet
Normal
Drought
Very drought
Time-series maps of soil moisture covered the whole
Zhejiang province land in 2007
浙江省2007年1月、2月土壤相对含水量分布图
Detecting soil moisture
based on MODIS data

Detecting soil moisture based on multi-resource remote sensing

Reverse of Soil Moisture for Summer Drought Period in Hangzhou by Using TVDI Method
Using Temperature—Vegetation Dryness Index (TVDI) to predict the soil
surface moisture. The remote sensing data include LST and EVI from MODIS
data , also containing measured soil moisture over the same period.
Application I：Monitoring drought

Soil remote sensing and information technology http://rs.zju.edu.cn
Application II：Flooding forecastThe study of the integration
system of flood forecasting

Application III：DSM
Raphael
combine soil spectral libraries, and
other geo-referenced information, such
as from digital terrain models and field
observations, with information from
multi- and hyper-spectral remote
sensing imagery

Zhou Y and Shi Z. Digital soil mapping technology. 2010
County Level Digital Soil Mapping (1:50,000) Using Decision Tree Model
DT algorithmDT algorithm
ID3
CART
C4.5
SLIO
SPRINT
C4.
5
C4.
5
Easy to build Rule system
Low efficiency in computation
Easy to build Rule system
Low efficiency in computation
SCOPAN approach
(McBratney et al., 2003)
Study area – Fuyang county

County Level Digital Soil Mapping (1:50,000) Using Decision Tree Model

County Level Digital Soil Mapping Using Decision Tree Model
soil type map (1:50,000)
Predictor variables
Soil (s)
Soil classification map (categorical)
Landsat ETM band 1,3,6,7
Soil spectral data
Climate (c)
Mean annual rainfall (mm)
Mean annual temperature (°C)
Organisms(o)
Landsat ETM band 2,4,5
Vegetation cover, NDVI, Wetness
Relief (r)
DEM, slope, relief, aspect
Parent materials (p) and age(a)
Geology
Land use
Geology map

County Level Digital Soil Mapping
Using Decision Tree Model
aspect slope
elevation
curvatures
relief
Predictors
data from
DEM

Normalized difference vegetation index(NDVI)
Predictors data from Landsat
image (optical remote sensing)
Normalized difference water index (NDWI)
Soil color index (SCI)

Soil samples (n=3682) and spectral
measurement in Lab (n=347)
Referenced mapReferenced map
Predicated mapPredicated map
Predictors data from soil sampling
and spectral measurement

Summary and discussion
Soil remote sensing and information technology http://rs.zju.edu.cn
The increasing number of new satellite or sensors now available and the
higher spatial/spectral resolution along with shorter revisit intervals offers
greater potential than ever improve the digital mapping of soil type and
some functional soil properties.
But, the knowledge of classical mapping is not replaced by the new
digital technologies including remote sensing(RS) and proximal
sensors(PS) which mostly provide the auxiliary information for DM
nowadays.
In spite of recent advances, development of robust methods for
estimating soil properties using RS/PS has to be extremely complicated.
We should further to research the theory behind of new technologies /
modeling approaches, and do more extensive field calibration working in
different soils under different agricultural practice.

Acknowledgement
National scientific foundation of China(NSFC)
（1）土壤空间变异与景观模型支持下的土壤遥感解译技术研究（40001008）, 2001-2003
（2）滨海盐土高光谱特性和二向反射模型研究（40571066）, 2006-2008
（3）基于高精雷达卫星的滨海盐土水盐含量遥感监测建模研究（40871100）, 2009-2011
Ministry of education of China, National scientific foundation of
Zhejiang Province
教育部新世纪人才计划、浙江省自然科学杰出青年基金、浙江省科技计划重大项目等
（1）区域典型土壤光谱标准数据库和有机质预测模型研究. 教育部新世纪优秀人才. 2010－2012年.
（2）星陆双基遥感土壤表层含水量同化反演方法研究.浙江省自然科学杰出青年基金.2011-2013.
（3）流域汛旱情遥感监测与防减灾辅助分析系统研究.浙江省重大科技计划.2011-2013年
土壤遥感与信息技术 http://rs.zju.edu.cn

Thanks and questions?Thanks and questions?
Institute of Agricultural Remote Sensing and Information Sysytem,
Zhejiang University, Hangzhou, China e-mail: shizhou@zju.edu.cn

Predicting and mapping soil properties using proximal/remote sensing by Zhou Shi, Ji W J, Cheng J L, Li X, Zhou L Q, Huang Q T, Huang M X, Yang HQ

More Related Content

What's hot (20)

Viewers also liked (14)

Similar to Predicting and mapping soil properties using proximal/remote sensing by Zhou Shi, Ji W J, Cheng J L, Li X, Zhou L Q, Huang Q T, Huang M X, Yang HQ (20)

More from FAO (20)

Recently uploaded (20)

Predicting and mapping soil properties using proximal/remote sensing by Zhou Shi, Ji W J, Cheng J L, Li X, Zhou L Q, Huang Q T, Huang M X, Yang HQ