A New Optical Trapezoid Model for Remote Sensing of Soil Moisture - - PowerPoint PPT Presentation

Morteza Sadeghi Scott B. Jones

• Dept. Plants, Soils, and Climate,

Utah State University

A New Optical Trapezoid Model for Remote Sensing of Soil Moisture

1

Ebrahim Babaeian Markus Tuller

• Dept. Soil, Water & Environmental Science,

The University of Arizona

2

 EM radiation in various wavelengths is correlated to soil moisture.

Optical [0.4-2.5 μm] Thermal [3.5-14 μm] Microwave [0.5-100 cm]

 High penetration depth. ⨯ Low spatial resolution.  High spatial resolution. ⨯ Low penetration depth.

Downscale

3

Thermal-Optical Trapezoid Model (TOTRAM)

d d w d d w

LST LST W LST LST θ θ θ θ − − = = − −

 Linear LST-θ relationship:  Linear dry and wet edges:

d d d

LST i s NDVI = +

w w w

LST i s NDVI = +

( )

d d d w d w

i s NDVI LST W i i s s NDVI + − = − + −

 TOTRAM:

4

Two main limitations of TOTRAM:

1) TOTRAM cannot be used for satellites with no thermal band (e.g. Sentinel-2). 2) Beside soil moisture, LST depends on ambient environmental factors (e.g. air temperature, wind speed). TOTRAM needs to be parameterized for each individual image.

5

Can we resolve both limitations by proposing an “Optical” Trapezoid model?

Core idea?

Reflectance-soil moisture relationship is not significantly affected by environmental factors. So, a universal parameterization is feasible.

Soil water content (฀)

0.0 0.2 0.4 0.6 0.8

STR

1 2 3 4 5 6

Aridisol Andisol Mollisol Entisol

6

OPTRAM is based

• n

a linear physically-based model:

d d w d w d

STR STR W STR STR θ θ θ θ − − = = − −

( )

2

1 2

SWIR SWIR

R STR R − =

where: RSWIR: Reflectance at SWIR STR: Transformed reflectance at SWIR

Sadeghi et al. 2015. A linear physically- based model for remote sensing of soil moisture using short wave infrared bands. Remote Sensing of Environment. 164:66-76.

Optical Trapezoid Model (OPTRAM)

7

Optical Trapezoid Model (OPTRAM)

 Linear STR-θ relationship at a given NDVI:  Linear dry and wet edges:

d d d

STR i s NDVI = +

w w w

STR i s NDVI = +

( )

d d d w d w

i s NDVI STR W i i s s NDVI + − = − + −

 OPTRAM:

d d w d w d

STR STR W STR STR θ θ θ θ − − = = − −

8

New Model Traditional Model

( )

d d d w d w

i s NDVI LST W i i s s NDVI + − = − + −

( )

d d d w d w

i s NDVI STR W i i s s NDVI + − = − + −

9

Study Area

1 SCAN site; 15 rain-gauge stations 17 USDA-ARS micro-net stations

Arizona Oklahoma

10

Landsat-8

NASA (11 February 2013) 9 Optical and 2 thermal bands Spatial resolution: 30-100 m Temporal resolution: 16 days

Satellite Imagery

Sentinel-2

ESA (23 June 2015) 13 optical bands Spatial resolution: 10 to 60 m Temporal resolution: ~10 days

12 images in WG 5 images in LW 2015-2016 17 images in WG 4 images in LW 2015-2016

11

Model Parameterization

 Feasibility of universal parameterization was tested incorporating all images.  Two scenarios were considered: 1) Local calibration: Edges were determined visually. W was calibrated with θ data. 2) No local calibration: Edges were determined by fitting. W was converted to θ using measured min and max θ.

d w d

W θ θ θ θ − = −

( )

d d d w d w

i s NDVI LST W i i s s NDVI + − = − + −

( )

d d d w d w

i s NDVI STR W i i s s NDVI + − = − + −

TOTRAM: OPTRAM: Normalized soil moisture:

12

 A nearly trapezoidal shape is formed:

LST is sensitive to θ in a broad range of fractional vegetation covers.

Traditional Trapezoid

 Integrated trapezoid consists of several separate smaller trapezoids:

LST depends on ambient environmental factors besides soil moisture.

13

NDVI NDVI

 A nearly trapezoidal shape is formed:

STR is sensitive to θ even in densely vegetated soils.

 Trapezoids are visually similar:

Universal calibration is feasible.

New Trapezoid

14

0.0 0.1 0.2 0.3

Estimated Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

OPTRAM, Sentinel-2, WG

MAE = 0.033 RMSE = 0.042 R2 = 0.500 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

OPTRAM, Sentinel-2, LW

MAE = 0.024 RMSE = 0.031 R2 = 0.886 0.0 0.1 0.2 0.3

Estimated Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

OPTRAM, Landsat-8, WG

MAE = 0.026 RMSE = 0.033 R2 = 0.608 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

OPTRAM, Landsat-8, LW

MAE = 0.027 RMSE = 0.037 R2 = 0.785

Measured Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

Estimated Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

TOTRAM, Landsat-8, WG

MAE = 0.033 RMSE = 0.045 R2 = 0.543

Measured Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

TOTRAM, Landsat-8, LW

MAE = 0.018 RMSE = 0.026 R2 = 0.897

Overall Accuracy (with local calibration)

 TOTRAM and OPTRAM showed similar accuracy.  Both models, when calibrated, yield reasonable estimates (error < 4%)

15

Overall Accuracy (No local calibration)

 Without local calibration, both models still yield reasonable estimates (error ~ 4-5%)  Scattering is due to approximations:

1) Linear LST-θ relationship at a given NDVI. 2) Linear STR-θ relationship at a given NDVI. 3) Linear LST-NDVI relationship at a given θ. 4) Linear STR-NDVI relationship at a given θ.

0.0 0.1 0.2 0.3

Estimated Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

OPTRAM, Sentinel-2, WG

MAE = 0.036 RMSE = 0.045 R2 = 0.316 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

OPTRAM, Sentinel-2, LW

MAE = 0.048 RMSE = 0.059 R2 = 0.596 0.0 0.1 0.2 0.3

Estimated Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

OPTRAM, Landsat-8, WG

MAE = 0.032 RMSE = 0.042 R2 = 0.296 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

OPTRAM, Landsat-8, LW

MAE = 0.040 RMSE = 0.051 R2 = 0.569

Measured Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

Estimated Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3

TOTRAM, Landsat-8, WG

MAE = 0.030 RMSE = 0.041 R2 = 0.489

Measured Soil Moisture (cm3 cm-3)

0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

TOTRAM, Landsat-8, LW

MAE = 0.036 RMSE = 0.045 R2 = 0.677

16

Soil Moisture Maps

 TOTRAM yielded W in a narrow range.  OPTRAM maps better match the

• DEM. They show river network.

17

WG (17-Nov-15) RMSE: OPTRAM = 0.12 TOTRAM = 0.12 WG (3-Dec-15) RMSE: OPTRAM = 0.13 TOTRAM = 0.15 WG (20-Jan-16) RMSE: OPTRAM = 0.19 TOTRAM = 0.15

Estimated W

0.0 0.5 1.0 WG (5-Feb-16) RMSE: OPTRAM = 0.21 TOTRAM = 0.14 WG (21-Feb-16) RMSE: OPTRAM = 0.12 TOTRAM = 0.11 WG (24-Mar-16) RMSE: OPTRAM = 0.10 TOTRAM = 0.10 WG (9-Apr-16) RMSE: OPTRAM = 0.10 TOTRAM = 0.07

Estimated W

0.0 0.5 1.0 WG (25-Apr-16) RMSE: OPTRAM = 0.12 TOTRAM = 0.08 WG (11-May-16) RMSE: OPTRAM = 0.07 TOTRAM = 0.04 WG (27-May-16) RMSE: OPTRAM = 0.08 TOTRAM = 0.04 LW (2-Dec-15) RMSE: OPTRAM = 0.19 TOTRAM = 0.16

Measured W

0.0 0.5 1.0

Estimated W

0.0 0.5 1.0 LW (18-Dec-15) RMSE: OPTRAM = 0.19 TOTRAM = 0.16

Measured W

0.0 0.5 1.0 LW (4-Feb-16) RMSE: OPTRAM = 0.18 TOTRAM = 0.17

Measured W

0.0 0.5 1.0 LW (23-Mar-16) RMSE: OPTRAM = 0.16 TOTRAM = 0.15

Measured W

0.0 0.5 1.0 LW (11-May-16) RMSE: OPTRAM = 0.20 TOTRAM = 0.17

Estimated W

0.0 0.5 1.0 WG (1-Nov-15) RMSE: OPTRAM = 0.16 TOTRAM = 0.14

Date-by-Date Comparison

 TOTRAM failed in predicting spatial variability of soil moisture: Universal parameterization is not feasible.  OPTRAM successfully captured spatial variability of soil moisture: Universal parameterization is feasible.

18

Conclusions:

 OPTRAM resolves two limitations of TOTRAM.  OPTRAM and TOTRAM overall accuracy is comparable.

Future Work:

 More extensive evaluations.  Improving model accuracy and parameterization.

19

Reference: Sadeghi, M., E. Babaeian, M. Tuller, S. B. Jones. 2017. The Optical Trapezoid Model: A Novel Approach to Remote Sensing of Soil Moisture Applied to Sentinel-2 and Landsat-8 Observations. Remote Sensing of Environment, Accepted. Acknowledgement: Funding from National Science Foundation awarded to USU and UofA.