Observations of Physical Properties and Microwave Emission of Snow at CREST Station in

Caribou, ME

Peter Romanov1,2, Tarendra Lakhankar1, Carlos Perez Diaz1, Reza Khanbilvardi1

1NOAA-CREST, City University of New York, New York, NY

2Center for Satellite Applications and Research, NESDIS/ NOAA

Why and What For ?

• Continuous monitoring of physical and radiative properties of snow

- Snow pack physical and microwave emission modeling

- Satellite products/algorithms cal/val

• Field work training for students

• Testing new instrumentation for snow research

Motivation

Lack of synchronous observations of snow MW emission and physical properties

- Hampers development of satellite snow retrieval algorithms

Objectives

Mean seasonal snowfall: 2.8 m Seasonal max snow depth: 60-70 cm Snow cover duration: 4 months

CREST Station

- Established in 2010 - Located in Caribou, ME - Premises of regional airport - Next to NWS RFC

CREST-SAFE Snow Field Research Station: Location

CREST Snow Station: Growing Each Year

37 GHz Radiometer 89 GHz Radiometer 2010

2011

2012

2013

Snow Pillow (SWE) Rain/Snow Gauge Soil Moisture (2.5, 5, 10cm) Soil Temp (2.5, 5, 10cm) Manual Snow Pit measurements Network Cameras (2)

Infrared Thermometer (Skin Temperature) Snow Depth Sensor Radiation Sensors Wind Speed/Direction, Humidity/air temperature

19 and 10.65 GHz Radiometers CIMEL Sun Photometer (BRDF)

Snow Temperature Profiler

2014 Snow Wetness profiler GMON3 SWE Sensor

Weather sensor (precipitation, visibility)

2015

Oct 2012

Currently: over 15 sensors operating 24/7 Snow pit measurements 2-3 time a week Data from most sensors is available online Funding: NOAA, CCNY, DoD

Where we are now…

- Over 15 instruments running 24/7 - Instrumental observations are automated - Snow pit measurements 2-3 times a week - Web cameras

Provide

- Snow pack physical properties - Depth, SWE, temperature, density, grain size, profiles

- Snow pack radiative properties (MW, IR, VIS) - Soil moisture/temperature - Standard meteorology & radiation - Site Imagery (weather, clouds)

Most data are transmitted to CREST, processed and posted online within 1-2h

CREST Microwave Radiometers CREST-SAFE SSMI AMSR-E/2 TMI WindSAT GMI (GPM)

6.93 VH 6.8 VH 10.65 VH 10.65 VH 10.7 VH 10.7 VH 10.65 VH

19 VH 19 VH 18.7 VH 19.4 VH 18.7 VH 18.7 VH 22 V 23.8 VH 21.3 V 23.8 VH 23.8 V

37 VH 37 VH 36.7 VH 37 VH 37.0 VH 36.5 VH 89 VH 87 VH 89 VH 85.5 VH 89.0 VH

CREST Microwave Observations

- Spectral bands same as at satellites

- Polarized (V,H)

- Collected at 30 sec interval 24/7

- Resampled to 5 min interval

- Calibration 3-4 times a year

- Less than 3% data loss in the last two winter seasons

Microwave spectra: Snow melt and refreeze

Dec 27, 2015 Dec 28, 2015 Dec 29, 2015

Microwave spectra every 4 hours, red: V-pol, blue: H-pol

Air and skin temperature

Snow Depth

Wind speed

Gamma-Ray Snow Water Equivalent Sensor

• CS725 (GMON3) Sensor: SWE estimates using observed gamma-ray emission • Installed in spring 2014 • Data have yet to be calibrated

0 20 40 60 80 100 120Day of Year 2014

0

200

400

600

Gam

ma-

ray,

cou

nts

0

10

20

30

40

50

SW

E, c

m

Potassium (gamma-ray)Thalium (gamma-ray)NOHRSC modelCREST Snow Tube

CS725 SWE, non-calibrated

Snow pack and soil temperature profile - Temperature at 16 levels - Down to 10 cm into the soil - Up to 90 cm in the snow pack - Reported every 5 min

Observed snow depth and snow temperature profiles in January to March 2011

Red color indicates temperature of 00 C and over

Snowpack Temperature Profiler

Started in December 2011 Include: • Snow pit measurements, every 2-3- days

−Vertical profiles of snow density, grain size, hardness, temperature − Snow depth, SWE − Performed by local students, trained by CREST personnel

Manual Observations of Snow Pack

Webcam Imagery

Helps to better characterize the weather type and to identify clear and cloudy periods of the day

Snowpack Phyisical Modeling (SNTHERM)

Observed

Simulated (SNTHERM)

Observed

Simulated (SNTHERM)

Snow Density (2011-2012) Snow Temperature (2010-2011)

Model vs in situ: good agreement on the temperature profile, not so good on the snow density and grain size.

Microwave Emission: Observed vs Simulated (HUT)

HUT with input from SNTHERM, single snow layer : red (V) and purple (H) Observed BTs: blue (V) and green (H)

Model vs observed

Differences up to 20K-30K, larger disagreement after the first snowmelt (Day 50) Model underestimates polarization difference at 89 GHz Small improvement with MEMLS and DMRT

January-February 2011 37 GHz

89 GHz

10 km

91 GHz

37 GHz

19 GHz

Caribou, ME

Range of forest cover fraction (%) within sensor FOV 10 GHz 19 GHz 37 GHz 89/91 GHz

SSMIS - 44-51 27-47 18-46 AMSR2 26-46 19-45 16-44 3-50

Range of forest cover fraction was calculated assuming 5 km maximum displacement of the center of the satellite sensor FOV from the station location

Comparing Satellite and in Situ: Consider Forest Cover

SSMIS FOV

Each SSMIS and AMSR2 pixel is ~ 20-50% covered by forest

AMSR2 spectral difference

In situ spectral difference

Snow cover on the ground

Brightness temperature at 19 and 37 GHz, V-polarization

Snow Depth

- Larger satellite BTs and smaller spectral gradient are due to forest cover - Largest differences in the beginning and in the end of snow season - Close similarity over snow free land

Microwave emission at 19 and 37 GHz: In situ vs AMSR2

Pol

ariz

atio

nDif,

10G

Hz

Pol

ariz

atio

nDif,

19G

Hz

Pol

ariz

atio

nDif,

37G

Hz

Pol

ariz

atio

nDif,

89G

Hz

10v

10h

19v

19h

37v

37h

89v

89h

1 0-1

9v10

-19h

19-3

7v1 9

-37h

37-8

9v37

-89h

10-3

7v10

-37h

19-8

9v19

-89h

0

0.2

0.4

0.6

0.8

1

Cor

rela

tion

SSMIS vs In SituAMSR2 vs in situ

January 1 to April 20, Fully snow-covered land surface

Correlation of Satellite (AMSR2, SSMIS) and in Situ Observations

High correlation indicates that - Both sensors “see” the same snow processes - Station obs. are representative for wider area

NASA SnowEx experiment, Feb 2017, Grand Mesa, Colorado - Surface + airborne measurements

Proposal pending at DoD to acquire

• Microwave radiometer 1.4 GHz - Soil moisture, state of soil surface

• Microwave radiometer 150 GHz - Snow emission, atmospheric remote sensing

• Scanning lidar - Evaluate terrestrial laser scanning (TLS) technique for areal snow studies

What’s next

Info/Data Access

18

Project web site: crest.ccny.cuny.edu/snow/data.html Data: www.star.nesdis.noaa.gov/smcd/emb/snow/caribou/caribou_microwave.html

We are open for collaboration

- Will provide data for analysis, model development, joint projects

- Other instrumentation ? We have some space at the station.

- Specific experiments with available instrumentation ?

Questions, comment to: peter.romanov@noaa.gov, tlakhankar@ccny.cuny.edu

Thank you !

Gamma-ray: can be also used in soil moisture studies

• Gamma ray sensor data are strongly correlated with rainfall

• Sensitive to monitor soil moisture.

140 160 180 200 220 240 260 280 300Day of Year 2014

0

20

40

60

80

100

Gam

ma-

ray,

cou

nts

0

20

40

60

80

100

120

Pre

cipi

tatio

n, m

m

PotassiumThaliumDaily rainfall

Snow Wetness Profiler

CS650 Water Content Reflectometer

Dielectric Permittivity

Vertical Snow Wetness

Profile

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1/31/2014 2/10/2014 2/20/2014 3/2/2014 3/12/2014 3/22/2014 4/1/2014 4/11/2014 4/21/2014 5/1/2014

Perm

ittiv

ity 80 cm 60 cm 40 cm 20 cm

-40

-30

-20

-10

0

10

20

1/31/2014 2/10/2014 2/20/2014 3/2/2014 3/12/2014 3/22/2014 4/1/2014 4/11/2014 4/21/2014 5/1/2014

Sens

or T

empe

ratu

re (C

)

80 cm 60 cm 40 cm 20 cm

Instrument consists of 16 CS625 water content reflectometers

Liquid water content can derived from measured permittivity

Relationship between snow permittivity and liquid water content has yet to be established.

Applications: Snowpack modeling

Observed Simulated (SNTHERM)

Ice layer at the bottom of the snow pack is not reported

Reporting protocol should be improved to allow for detailed characterization of ice layers

Satellite: Much smaller polarization difference over snow-covered land

AMSR2 polarization difference

In situ polarization difference

Snow cover on the ground

Brightness temperature at 37 GHz, V and H polarization

Snow Depth

Microwave emission at 37 GHz, V&H: In situ vs AMSR2

Standard Deviation, AMSR2 and in Situ

Pol

ariz

atio

nDif,

10G

Hz

Pol

ariz

atio

nDif,

19G

Hz

Pol

ariz

atio

nDif,

37G

Hz

Pol

ariz

atio

nDif,

89G

Hz

10v

10h

19v

19h

37v

37h

89v

89h

10-1

9v10

-19h

19-3

7v19

-37h

37-8

9v37

-89h

10-3

7v10

-37h

1 9-8

9v19

-89h

0

10

20

30

40

Sta

ndar

d D

evia

tion,

KIn SituAMSR2

Jan 1 – April 20, 2015, Fully snow-covered land surface

-Much larger variability of in situ temperatures

- Variability of satellite temperatures may be dampened by forest cover

AMSR2 In Situ

Mean Bias, Satellite vs In Situ

• Larger in situ-satellite differences in winter

• Larger differences in H-pol

• Close agreement to AMSR2 with no snow on the ground

• Better in situ agreement to AMSR2 than to SSMIS

10v

10h

19v

19h

37v

37h

89v

89h

Spectral Band, GHz

-20

0

20

40

60

80

Bias

, K

AMSR vs In SituSSMIS vs In Situ

10v

10h

19v

19h

3 7v

3 7h

8 9v

8 9h

Spectral Band, GHz

-20

0

20

40

60

80

Bias

, K

AMSR vs In SituSSMIS vs In Situ

Snow on the ground (Jan 1 – April 20) No snow on the ground (May 1 – May 31)