Pang-Wei Liu 1, Roger De Roo 2, Anthony England 2,3, Jasmeet Judge 1 1. Center for Remote Sensing, Agri. and Bio. Engineering, U. of Florida 2. Atmosphere,

1

Understanding Soil Moisture Transport In Sandy Soils Using Multi-Frequency

Microwave Observations

Pang-Wei Liu1, Roger De Roo2, Anthony England2,3,

Jasmeet Judge1

1. Center for Remote Sensing, Agri. and Bio. Engineering, U. of Florida

2. Atmosphere, Oceanic, and Space Sciences, U. of Michigan

3. Electrical Engineering and Computer Science, U. of Michigan

UFUNIVERSITY of

FLORIDA

Outline

Introduction & MotivationMicroWEX-5MB ModelMethodologyResultsConclusions

2

Introduction & MotivationSoil moisture (SM) is an important factor

In hydrology: evapotranspiration, infiltration, surface runoff, and groundwater recharge.

In agriculture: crop growth and yield.

Satellite missions for SM: AMSR-E, NASA and JAXA, 2002

– V- & H-pol passive at C-band.– Spatial resolution at 6.25-57km and repeat coverage in 1-2 days.

SMOS, ESA, Nov. 2009.– V- & H-pol passive at ~1.4GHz (L-band).– Spatial resolution at 40-50km and repeat coverage in 2-3 days

SMAP, NASA, Oct. 2014.– Active at 1.26 GHz and passive at 1.41GHz.– Spatial resolution of active at 1-3 km and of passive at ~40km and

repeat coverage in 2-3 days.

Provide TB for assimilation and soil moisture retrieval.

3

Introduction & Motivation

Problem: The near-surface SM is highly dynamic, particularly in sandy soils. Current forward microwave algorithms typically use SM averaged

over 0-5cm may result in unrealistic TB.

Objectives: To determine the vertical resolution of the soil moisture necessary

to provide realistic TB at L-band for bare soils. To utilize combined C- & L- band observations to determine the

surface roughness and moisture, and the vertical resolution in the soil.

4

Microwave Water and Energy Balance Experiments (MicroWEXs)

Series of season-long experiments conducted at a 9-acre field in NC Florida.

Fifth MicroWEX (MicroWEX-5): growing season of sweet corn from March 9 (DoY 68) through May 26 (DoY 146) in 2006

The bare soil period: from DoY 68 to 95; LAI < 0.3

Soil moisture and temperature values were observed every 15 minutes at the depths of 2, 4, 8, 16, 32, 64, and 120cm.

V- & H-pol. TB at C-band and H-pol. TB at L-band every 15 minutes.

Soil Texture Parameters

Porosity (m3/m3) 0.37

Sand (% by vol.) 89.4

Clay (% by vol.) 7.1

Silt (% by vol.) 3.5

5

Mesh board for soil roughnessLiDAR for soil roughness

Mesh Board

Correlation Length (cm)

rms Height (cm)

1 15.5 0.8

2 8.1 0.7

3 5.3 0.4LiDAR

Correlation Length (cm)

rms Height (cm)

1 11.3 0.7

2 9.1 0.7

3 3.0 0.46

MB ModelTypical Approaches

Radiative Transfer Equation: zero order approximation

TBsoil, p = Teff ∙ ep

– Teff Soil temperatures at surface (TIR) and deep layer (~50cm).

– ep= (1 - rp) rp (εr, roughness)

– εr (SM, soil texture) dielectric models: Dobson et al., 1996 and Mironov et al., 2009

Rough surface models– Semi-empirical model: Q-h model Wang & Choudhury, 1981 rp (εr, rmsh,

f, θ).

– Empirical model Wegmüller & Mätzler, 1999 rp (εr, rmsh, f, θ); 1-100GHz.

– Physically-based model: IEM (Fung et al., 1992) ep (εr, rmsh, cl, f, θ); applicable for wide range of surfaces.

7

Comparison with observations

VSM0-5 from MicroWEX-5 Soil porosity = 0.37 Rms height = 0.616 cm Correlation length = 8.4 cm Looking angle = 50o

8

MethodologyModifications in the MB model:

Soil:– Discrete layers with non-uniform temperature and SM.– Rough surface – Semi-infinite lower boundary

Sandy soils are more porous at the surface.– Top 1.5 cm divided into 7 layers.– 1.5 – 32.5 cm divided into 1cm thick layers.– > 32.5 cm layer thickness increases with depth

1st order RTE– Single reflection considered at each layer interface.– IEM model is applied at layer 1 - rough surface

– TB contributions from each layer combine to obtain the total TB

TB

9

Methodology

Refractive mixing model for ε– Modified Mironov’s model (2010)

Use C-band (6.7 GHz) TB observations to estimate

– Surface roughness rms height and correlation length– Soil porosity in top 1mm– SM in top 1mm

These parameters are used with the SM observation from lower layers to estimate H-pol. TB at L-band.

10

Results

Estimation of rms height, correlation length, and porosity in top 1mm

Provide the best estimate during the dry (SM1mm = 0.01) and the wet (SM1mm = 0.29) periods

The SM from 0-2.5cm linearly interpolated

-Rms height = 0.41cm -Correlation length = 8.4cm -Soil porosity = 0.55

SM at > 2.5cm from MicroWEX-5

11

Results

Estimation of SM in top 1mm.

SM in the top 1mm b/w breaking points linearly interpolated

Rms height = 0.41cm Correlation length = 8.4cm Soil porosity = 0.55

0.29 0.25 0.16 0.18 0.18 0.02 0.01 0.10 0.10 0.01

MicroWEX-5Best estimation

12

Results

Comparison of SM in the top 1mm with 0-5 cm SM during MicroWEX-5

Soil porosity: 1mm = 0.55; rest layers =0.37SM profiles at wet, medium, and dry points

13

MicroWEX-5

ResultsComparison of:

TB from MicroWEX-5

Case1: TB using SM 0-5 cm from MicroWEX-5.

Case2: TB using best estimate of SM, porosity, and roughness in the top 1mm from C-band; SM from 1mm-2.5cm linearly interpolated; SM > 2.5cm from MicroWEX-5.

Case3: TB using average of the best estimate in the top 1mm from C-band and SM at 2.5cm from MicroWEX-5; SM > 2.5 cm from MicroWEX-5; SM in top 1mm at the time of event from C-band for up to 30minutes.

14

Results

Extension of methodology to the another drydown period from DoY 87.5-90.5

Estimation of SM in top 1mm.

0.32 0.28 0.19 0.19 0.01 0.10 0.01

MicroWEX-5Best estimation

15

SM in the top 1mm b/w breaking points linearly interpolated

Rms height = 0.41cm Correlation length = 8.4cm Soil porosity = 0.55

Comparison of SM in the top 1mm with 0-5 cm SM during MicroWEX-5

Soil porosity: 1mm = 0.55; rest layers =0.37SM profiles at wet, medium, and dry points

Results

16

MicroWEX-5

Results

17

Comparison of: TB from MicroWEX-5

Case1: TB using SM 0-5 cm from MicroWEX-5.

Case2: TB using best estimate of SM, porosity, and roughness in the top 1mm from C-band; SM from 1mm-2.5cm linearly interpolated; SM > 2.5cm from MicroWEX-5.

Case3: TB using average of the best estimate in the top 1mm from C-band and SM at 2.5cm from MicroWEX-5; SM > 2.5 cm from MicroWEX-5; SM in top 1mm at the time of event from C-band for up to 30minutes.

Conclusions

SM 0-5cm is not adequate for estimating realistic TB at L-band in sandy soils, particularly during and immediately following precipitation/irrigation events.

TB at C-band may be used to derive soil surface characteristics such as roughness, porosity, and SM.

TB at L-band may be obtained using the derived properties and the observations at 2cm.

Future work: Extending/generalizing the methodology for larger applicability.

18

Acknowledgment

NASA Terrestrial Hydrology Program (NASA-THP-NNX09AK29G)

MicroWEX-5 was supported by the NSF Earth Science Division (EAR-0337277) and the NASA New Investigator Program (NASA-NIP-00050655).

19

Thank You For Attention

Questions??

20

While the soil saturated

21

The VSM at 1mm layer was set at 1% in dry period. - rmsh=0.616cm, cl=8.4cm - soil porosity = 0.5

22

The VSM at 1mm layer was set at 29% in the wet period. -rmsh=0.41cm, cl=8.4cm -Porosity = 0.5

23

ResultsComparison of radiative emission models.

1. Overall, 484 pairs of soil moisture and temperature profiles were applied.

2. The average difference is within 3K at L-band.

3. 1st order model was applied for further work.

24