integration of sar multi-frequency and optical data for the … · 2018-09-13 · integration of...

36
Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content A. Padovano 1,2 , F. Greifeneder 1 , R. Colombo 2 , G. Cuozzo 1 , C. Notarnicola 1 1 - Eurac Research Institute for Earth Observation, Bolzano, Italy 2 - Department of Environmental Science, University of Milano Bicocca, Milano, Italy

Upload: others

Post on 24-Jun-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content A. Padovano1,2, F. Greifeneder1, R. Colombo2, G. Cuozzo1, C. Notarnicola1

1 - Eurac Research Institute for Earth Observation, Bolzano, Italy2 - Department of Environmental Science, University of Milano Bicocca, Milano, Italy

Page 2: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 219/09/17

➢ The soil moisture retrieval from SAR data is a challenging topic, especially in vegetated areas.

➢ The radar signal can be of difficult interpretation as the total radar backscatter is a complex sum of the backscatter fromvegetation and soil, making it complicated to determine which contribution comes from soil and which contributioncomes from vegetation.

➢ Optical and SAR images integration by the use of data driven techniques, such as Support Vector Regression (SVR), canlead to the separation of soil and vegetation contribution.

➢ Separation of soil and vegetation contribution offers the opportunity to compute both soil moisture and vegetationwater content.

The main aim of this work is to:

➢ Understand the contributions of different bands (optical and radar) for the retrieval of Soil Moisture Content (SMC) andVegetation Water Content (VWC).

➢ Develop an algorithm that integrates multi-frequency SAR data and optical data to estimate SMC and VWC in vegetatedareas

INTRODUCTION

Page 3: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 319/09/17

OUTLINE

➢ Test sites and data set

➢ Algorithm description

➢ 1.SVR SMC – Soil Moisture retrieval

➢ 2.Compute Bare Soil Backscattering with physical based models

➢ 3.SVR VWC – Vegetation water content retrieval

➢ Results and discussion

➢ Conclusions and future steps

Page 4: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 419/09/17

Test site & Data setSMEX 02 - IOWA

To develop the algorithm a controlled data set will be used, the data-set that was chosen is the SOIL MOISTURE EXPERIMENTS IN 2002 (SMEX02) IOWA, USA.

Data set for the algorithm development

• AIRSAR data have been geocoded and projected

in ground-range

Ground Soil Moisture

SMEX02 Iowa Regional Ground Soil Moisture Data (on Corn and Soybean

fields)

Aircraft Remote Sensing

SMEX02 Airborne Synthetic Aperture Radar Data (AIRSAR, Band P,L and

C spatial resolution of 7.5 m) Data

Satellite Remote Sensing

SMEX02 Iowa Satellite Vegetation and Water Index (NDVI and NDWI) Data

Derived from Landsat 5 and 7 Thematic Mapper Imagery (resolution of

30m)

Vegetation

SMEX02 Vegetation Water Content (on Corn and Soybean fields), Iowa

Regional and Walnut Creek Watershed

Page 5: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 519/09/17

Test site & Data setMazia ValleyThe study area is Mazia Valley, it is a small valley located at Nord-West of Alto Adige/Südtirol (Italy). It covers an area of about 100 Km2 with an height ranging between 920m and 3738m and it is mainly composed by meadows and pastures.Field campaigns have been performed during 2016 Summer and new field campaigns have been planned for 2017and 2018.The final version of the algorithm will be run by using this data-set

Sentinel 1C band(20m)

ALOS 2L band(10m)

Sentinel 2(20m)

in situ Soil Moisture

16/06/2016 15/06/2016 16/06/2016

22/07/2016 21/06/2016 22/07/2016

27/08/2016 29/08/2016 27/08/2016 29/08/2016

20/09/2016 21/09/2016 19/09/2016 21/09/2016

Page 6: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 619/09/17

Algorithm for soil moisture and vegetation water content retrieval

Feature Extraction

Satellite SAR data

Satellite Optical data

DEM data

SVR_SMC Model

SVR_SMC Model

Application

Electromagnetic

Model ApplicationComputed Bare Soil

Backscattering

Feature Extraction

SVR_VWC Model

ApplicationVegetation Water

Content Map

Satellite Optical Data

SVR_SMC Training

and

Test

Computed SMC Map

SVR_VWC Training

and TestSVR_VWC Model

Computed Soil

Backscattering Maps

Satellite SAR data

AVG of RoughnessBest Fit Roughness

Parameters

SAR backscattering

coefficient Maps

Optical Surface

Reflectance Maps

DEM Maps

SAR backscattering

coefficient Maps

Optical Surface

Reflectance Maps

• The measured backscattering from SAR images contains both soil and vegetation contributions

σ image = f(Vegetation, soil parameters) ;

The main concept is to develop an approach able to disentangle these two contributions:

• Optical and SAR data integration, by the mean of machine learning techniques (such as SVR), is used to compute SMC maps;

• By using the retrieved SMC maps and a physical model, bare soil backscattering is simulated

σ soil = f(soil parameters) ;

• A second SVR is trained providing as input reflectance of optical bands, SAR satellite backscattering and also bare soil simulated backscattering, to compute VWC maps .

Page 7: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 719/09/17

SVR SMC – Soil Moisture retrieval

Feature Extraction

Satellite SAR data

Satellite Optical data

DEM data

SVR_SMC Model

SVR_SMC Model

Application

SVR_SMC Training

and

Test

Computed SMC Map

SAR backscattering

coefficient Maps

Optical Surface

Reflectance Maps

DEM Maps

• SVR is a data driven machine learning technique.

• The regressor is trained to create a model that is able to derive soil moisture values measured in field by providing as input reflectance and backscattering values (in correspondence of the measured data).

• Other additional parameters are tested to checked whether the accuracy of the result can be improved (Digital Elevation Model, Local Incidence Angle, Topographic Wetness Index, etc.).

• The obtained model can finally be applied providing as input optical and SAR images to get SMC maps.

Page 8: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 819/09/17

Electromagnetic

Model ApplicationComputed Bare Soil

Backscattering

Computed SMC Map

AVG of RoughnessBest Fit Roughness

Parameters

Compute Bare Soil Backscattering

• Dielectric constant computation from soil moisture maps has been done by using the empirical relation

of Hallikainen, Ulaby (1985)

• The bare-soil simulated backscattering has been computed using three different models:

➢ Integral Equation Model (Fung et al. 1992)

➢ Semi-empirical Oh method (Oh et al. 1992, Oh 2002)

➢ Semi-empirical Dubois method (Dubois et al. 1995)

All the methods provide HH polarized back-scattering and VV polarized backscattering.

The enhanced semi-empirical Oh method (2002) provide also HV polarized backscattering.

Page 9: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 919/09/17

SVR VWC – Vegetation Water Content retrieval

Computed Bare Soil

Backscattering

Feature Extraction

SVR_VWC Model

ApplicationVegetation Water

Content Map

Satellite Optical Data

SVR_VWC Training

and TestSVR_VWC Model

Computed Soil

Backscattering Maps

Satellite SAR data

SAR backscattering

coefficient Maps

Optical Surface

Reflectance Maps

• A second SVR is then applied;

• The regressor is trained to create a model that is able to derive VWC values by providing as input reflectance and real and simulated backscattering values (in correspondence of the measured data);

• The obtained model can finally be applied providing in input optical and SAR real and simulated maps to obtain VWC maps.

Page 10: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1019/09/17

SVR SMC – Soil Moisture retrieval - Results

input: L HH, L VV, L HV, L VH, C HH, C VV, C HV and C VH

SMEX 02 SVR_SMC

• The SVR performances prove that providing only SAR backscattering information in a vegetated area is not enough to model SMC;

RMSE = 0.07R= 0.28

Page 11: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1119/09/17

SVR SMC – Soil Moisture retrieval - Results

input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV LVH

SMEX 02 SVR_SMC

• The SVR performances prove that providing only SAR backscattering information in a vegetated area is not enough to model SMC;

• L band backscattering and optical reflectance combined information leads to a stable and reliable model (0.7<R<0.9 – 0.04[m3 /m3]<RMSE<0.05[m3 /m3]);

RMSE = 0.04R= 0.89

Page 12: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1219/09/17

SVR SMC – Soil Moisture retrieval - Results

input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV LVH DEM

SMEX 02 SVR_SMC

• The SVR performances prove that providing only SAR backscattering information in a vegetated area is not enough to model SMC;

• L band backscattering and optical reflectance combined information leads to a stable and reliable model (0.7<R<0.9 – 0.04[m3 /m3]<RMSE<0.05[m3 /m3]);

• In a flat area information about elevation, aspect and slope does not improve SVR performances;

RMSE = 0.04R= 0.81

Page 13: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1319/09/17

SVR SMC – Soil Moisture retrieval - Results

input: B1 B2 B3 B4 B5 B7 NDVI NDWI L HH L VV L HVC HH C VV C HV

SMEX 02 SVR_SMC

• The SVR performances prove that providing only SAR backscattering information in a vegetated area is not enough to model SMC;

• L band backscattering and optical reflectance combined information leads to a stable and reliable model (0.7<R<0.9 – 0.04[m3 /m3]<RMSE<0.05[m3 /m3]);

• In a flat area information about elevation, aspect and slope doesn’t improve SVR performances;

• Providing information about C band to the SVR worsen the performances of the model, this is mainly due to the fact that especially in corn fields the radar signal is not able to reach the ground.

RMSE = 0.04R= 0.67

Page 14: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1419/09/17

1 July 2002

SVR SMCSoil Moisture retrieval

Results

SMEX 02

input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV

Page 15: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1519/09/17

5 July 2002

SMEX 02

SVR SMCSoil Moisture retrieval

Results input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV

Page 16: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1619/09/17

7 July 2002

SMEX 02

SVR SMCSoil Moisture retrieval

Results input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV

Page 17: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1719/09/17

8 July 2002

SMEX 02

SVR SMCSoil Moisture retrieval

Results input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV

Page 18: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1819/09/17

9 July 2002

SMEX 02

SVR SMCSoil Moisture retrieval

Results input: B1 B2 B3 B4 B5 B7 NDVI NDWI LHH LVV LHV

Page 19: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 1919/09/17

SVR SMC – Soil Moisture retrieval - Results

input: Sentinel 1, Alos 2

Mazia ValleySVR_SMC

• The SVR performances prove that providing only SAR backscattering gives good results to model SMC if we provide also information about Local Incidence angle (R= 0.76 – RMSE= 0.06[m3 /m3]);

RMSE = 0.06R= 0.76

Page 20: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2019/09/17

SVR SMC – Soil Moisture retrieval - Results

Mazia ValleySVR_SMC

• The SVR performances prove that providing only SAR backscattering gives good results to model SMC if we provide also information about Local Incidence angle (R= 0.76 – RMSE= 0.06[m3 /m3]);

• In an area with a complex topography elevation, aspect and slope information are relevant to improve SVR performances;

input: Sentinel 1, Alos 2 + DEM

RMSE = 0.04R= 0.87

Page 21: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2119/09/17

SVR SMC – Soil Moisture retrieval - Results

input: Sentinel 1, Sentinel 2, Alos 2, DEM

Mazia ValleySVR_SMC

• The SVR performances prove that providing only SAR backscattering gives good results to model SMC if we provide also information about Local Incidence angle (R= 0.76 – RMSE= 0.06[m3 /m3]);

• In an area with a complex topography elevation, aspect and slope information are relevant to improve SVR performances;

• L and C band backscattering plus optical reflectance information leads to a stable and reliable model (R=0.9 –RMSE=0.04[m3 /m3]);

• L and C band combined information, in an area mainly composed by pastures and meadows, improve the SVR performances.

RMSE = 0.04R= 0.91

Page 22: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2219/09/17

SVR SMCSoil Moisture retrieval

Results

MaziaValley [0-1]

21 September 2016

input: Sentinel 1, Sentinel 2, Alos 2, DEM

Page 23: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2319/09/17

SVR VWC – Vegetation water content retrieval

SMEX 02 SVR_VWC Corn and Soybean

Results

• VWC values for soybean and corn fall into two different clusters, this brings to a bias, then both corn and all fields analysis have been performed (there were not enough data about soybean to perform a specific SVR);

Page 24: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2419/09/17

SVR VWC – Vegetation water content retrievalInput:

LHH LVV LHH bare-soil LVV bare-soil

SMEX 02 SVR_VWC

Corn andSoybean

Corn

Results

• VWC values for soybean and corn fall into two different clusters, this brings to a bias, then both corn and all fields analysis have been performed (there were not enough data about soybean to perform a specific SVR);

• Providing original L band SAR backscattering and simulated SAR backscattering to train the SVR leads to a good model for computing VWC maps (R=0.93 - RMSE=0.65[Kg/m2]; Corn R=0.74 - RMSE=0.57[Kg/m2]);

RMSE = 0.65R= 0.93

RMSE = 0.57R= 0.74

Page 25: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2519/09/17

SVR VWC – Vegetation water content retrievalInput:

CHH CVV LHH LVV LHH bare-soil LVV bare-soil

SMEX 02 SVR_VWC

Corn andSoybean

Corn

Results

• VWC values for soybean and corn fall into two different clusters, this brings to a bias, then both corn and all fields analysis have been performed (there were not enough data about soybean to perform a specific SVR);

• Providing original L band SAR backscattering and simulated SAR backscattering to train the SVR leads to a good model for computing VWC maps (R=0.93 - RMSE=0.65[Kg/m2]; Corn R=0.74 - RMSE=0.57[Kg/m2]);

• Adding information about C band brings to a worsening of the model (R=0.87 - RMSE=0.86[Kg/m2]; Corn R=0.70 - RMSE=0.6[Kg/m2]);

RMSE = 0.86R= 0.87

RMSE = 0.6R= 0.7

Page 26: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2619/09/17

SVR VWC – Vegetation water content retrievalInput:

NDWI LHH LVV LHH bare-soil LVV bare-soil

SMEX 02 SVR_VWC

Corn andSoybean

Corn

Results

• VWC values for soybean and corn fall into two different clusters, this brings to a bias, then both corn and all fields analysis have been performed (there were not enough data about soybean to perform a specific SVR);

• Providing original L band SAR backscattering and simulated SAR backscattering to train the SVR leads to a good model for computing VWC maps (R=0.93 - RMSE=0.65[Kg/m2]; Corn R=0.74 - RMSE=0.57[Kg/m2]);

• Adding information about C band brings to a worsening of the model (R=0.87 - RMSE=0.86[Kg/m2]; Corn R=0.70 - RMSE=0.6[Kg/m2]);

• Providing optical information by the mean of NDWI index, improves SVR performances (R=0.93 - RMSE=0.63[Kg/m2]; Corn R=0.83 - RMSE=0.48[Kg/m2]).

RMSE = 0.63R= 0.93

RMSE = 0.48R= 0.83

Page 27: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2719/09/17

1 July 2002

SVR VWCVegetation Water Content retrieval

SMEX 02

ResultsInput:

NDWI LHH LVV LHH bare-soil LVV bare-soil

Page 28: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2819/09/17

5 July 2002

SMEX 02

SVR VWCVegetation Water Content retrieval

ResultsInput:

NDWI LHH LVV LHH bare-soil LVV bare-soil

Page 29: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 2919/09/17

7 July 2002

SMEX 02

SVR VWCVegetation Water Content retrieval

ResultsInput:

NDWI LHH LVV LHH bare-soil LVV bare-soil

Page 30: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 3019/09/17

8 July 2002

SMEX 02

SVR VWCVegetation Water Content retrieval

ResultsInput:

NDWI LHH LVV LHH bare-soil LVV bare-soil

Page 31: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 3119/09/17

9 July 2002

SMEX 02

SVR VWCVegetation Water Content retrieval

ResultsInput:

NDWI LHH LVV LHH bare-soil LVV bare-soil

Page 32: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 3219/09/17

SVR VWC – Soil Moisture retrieval

Input:LHH LHV LHH bare-soil LHV bare-soil

Mazia Valley SVR_VWC

Results

• Alos 2 L band backscattering combined with L band simulated backscattering, leads to a reliable model to compute VWC, without any optical information.

RMSE = 0.07R= 0.81

Page 33: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 3319/09/17

Mazia Valley

(preliminary results*)

SVR VWCVegetation Water Content retrieval

21 September 2016

Kg/m2

Results

Input:LHH LHV LHH bare-soil LHV bare-soil

Page 34: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 3419/09/17

Conclusions and next steps➢The integration of optical and SAR data by the mean of machine learning techniques lead to an accurate retrieval of soil

moisture (<RMSE> = 0.04);

➢ In mountain areas, it is important to provide additional information about topography and SAR Satellite local incidenceangle in order to obtain accurate results;

➢The simulations of theoretical bare soil backscattering can help in providing information for a discrimination of watercontent in the soil and in the vegetation;

➢Results about Mazia Valley are still preliminary results; further validation is needed with ground data.

➢New ground data acquisition for soil moisture and vegetation water content are planned for 2017-2018 for the area ofVal Mazia;

➢Next step is to evaluate the propagation of the error in the algorithm to check the accuracy on the retrieved parameters

Page 35: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation water content 3519/09/17

References

- J. Wang and T. Schmugge, “An empirical model for the complex dielectric permittivity of soils as a function of water content.,” IEEE Trans. Geosci. Remote Sens., 1980.

- R. Bindlish and A. Barros, “Multifrequency soil moisture inversion from sar measurements with the use of iem.,” Remote Sens. Environ., 2000.

- H. S. Srivastava, P. Patel, Y. Sharma, and R. R. Navalgund, “Large-area soil moisture estimation using multi-incidence-angle RADARSAT-1 SAR data,” IEEE Trans. Geosci. Remote Sens., vol. 47, no. 8, pp. 2528–2535, 2009.- T. P. Anguela, M. Zribi, N. Baghdadi, and C. Loumagne, “Analysis of local variation of soil surface parameters with TerraSAR-X radar data over bare agriculture fields,” IEEE Trans. Geosci. Remote Sens., vol. 48, no. 2, pp. 874–881, 2010- Y. Oh, K. Sarabandi, and F. T. Ulaby, “An empirical modal and an inversion technique for radar scattering from bare soil surfaces,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 2, pp. 370–381, 1992.- L. Pasolli, C. Notarnicola, G. Bertoldi, L. Bruzzone, R. Remelgado, F. Greifeneder, G. Niedrist, S. Della Chiesa, U. Tappeiner and M. Zebisch, Estimation of umidità del suolo in mountain areas using SVR technique applied to multiscale

active radar images at C band, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 01/2015; 8(1):262-283.

- R. Prakash et al.: A fusion approach to retrieve soil moisture with SAR and Optical data, “IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 5, NO. 1, FEBRUARY 2012”, pp 196-206.

- Hallikainen, M.T.F.T.Ulaby, M.C. Dobson, M. A. El-Rayes, and L. Wu (1985)“Microwave Dielectric Behavior of Wet Soil – Part I: Empirical Models and Experimental Observations from 1.4 to 18 GHZ.” IEEE Transactions on Geoscience

and Remote Sensing, Vol. 23, Jan, 25-34.

- Fung, A.K., Li, Z., and Chen, K.S. 1992. Backscattering from a randomly rough dielectric surface. IEEE Transactions on Geoscience and Remote Sensing, Vol. 30, No. 2, 356369. doi: 10.1109/36.134085.

- Oh, Y., Sarabandi, K. and Ulaby, F.T., 1992, An empirical model and an inversion technique for radar scattering from bare soil surfaces. IEEE Transactions on Geoscience and Remote Sensing, 30, pp. 370–382.

- Oh, Y., Sarabandi, K. and Ulaby, F.T., 2002, Semi-empirical model of the ensemble-averaged differential Mueller matrix for microwave backscattering from bare soil surfaces. IEEE Transactions on Geoscience and Remote Sensing,

40, pp. 1348–1355.

- Dubois, P.C., van Zyl, J., and Engman, T. 1995. Measuring soil moisture with imaging radar. IEEE Transactions on Geoscience

and Remote Sensing, Vol. 33, No. 6, pp. 915-926. doi: 10.1109/TGRS.1995.477194.

-Y. Oh, K. Sarabandi, and F. T. Ulaby, “An empirical model and an inversion technique for radar scattering from bare soil surfaces,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 2, pp. 370–381, 1992.

-Greifeneder, F., C. Notarnicola, G. Cuozzo, G. Bertoldi, S. Della Chiesa, G. Niedrist, J. Stamenkovic, and W. Wagner. 2014. Estimation of surface soil moisture in alpine areas based on medium spatial resolution SAR time-series and

upscaled in-situ measurements. p. 92430G–92430G–10. In Proceeding of SPIE Vol. 9243. SPIE.

Page 36: Integration of SAR multi-frequency and optical data for the … · 2018-09-13 · Integration of SAR multi-frequency and optical data for the retrieval of soil moisture and vegetation

Thanks for your attention!!