jinyang du, lucas a. jones, and john s. kimball

Development of Consistent Long-term Global Land Parameter Data Record based on AMSR-

E, AMSR2 and MWRI observations

Jinyang Du, Lucas A. Jones, and John S. KimballNumerical Terradynamic Simulation Group and Flathead Lake Biological Station, Division of Biological

Sciences, The University of Montana

AMSR Joint Science Team Meeting23-24 September 2014Huntsville, AL

AMSR Joint Science Team Meeting 22-23 September 2014, Huntsville, AL

OVERVIEW

Consistent long-term land parameter records including retrievals of vegetation optical depth, surface temperature & moisture, landscape freeze/thaw dynamics, open water inundation & Atm. water vapor changes are desirable for Ecological studies & applications

In this study, satellite Tb inter-calibrations were carried out between similar sensors including AMSR-E (2002-2011), AMSR2 (from Jun-2012) and FY3B-MWRI (from Dec., 2010);

Further developments and re-calibration of UMT Global Land Parameter algorithms were also made for L1R AMSR2 swath data and then implemented for the reprocessed ( RSS V7) AMSR-E data;

AMSR-E algorithm has been applied to the calibrated Tb datasets to produce long-term land parameter data records from 2002 to 2014.

2002 AMSR-E

MWRI

AMSR2 NOW

Algorithm Calibration

• WMO Stations Temp.• AIRS Water Vapor• MODIS Land Cover• DEM

Subset Brightness Temperature and AIRS products for WMO Stations

Screening Datasets for RFI, Snow, Precipitation and High DEM variations

Adjust Algorithm Parameters based on WMO measurements and AIRS products

Calibrated brightness Temperature

AMSR-E / AMSR/MWRISwath Brightness Temperature

Gridded brightness Temperature

Data Gridding & Tb Inter-Calibration

Run Land Parameter Retrieval Algorithms

General Data Record Production Procedures


Towards Consistent Land Surface Parameter Records --Part I

Inter-calibration of AMSR-E, AMSR2 and MWRI Observations


Instruments Configurations

AMSR-2 AMSR-E MWRISatellite Platform GCOM-W1 AQUA FY3B

Altitude 700 km 705 km 836 km

Equator Crossing Time1:30 PM Ascending 1:30 PM Ascending 1:40 PM Ascending

Antenna Size 2 m 1.6 m 0.977 m x 0.897 m

Incident Angle 55 55 53

Spatial Resolution [km x km]Band[GHz] AMSR-2 AMSR-E MWRI

6.93 62 x 35 75 x 43 N/A7.3 62 x 35 N/A N/A

10.65 42 x 24 51 x 29 85x 5118.7 22 x 14 27 x 16 50x 3023.8 26 x 15 32 x 18 45x 2736.5 12 x 7 14 x 8 30x 18 89.0 5 x 3 6 x 4 15x 9


Sensor Inter-calibration – Estimating Observation Biases

Estimating Sensor Biases using Double Difference Method

_ 2 _

_ _

( 2, - ) ( 2, ) ( - , )

( 2, )

( - , )

b AMSR b MWRI

b AMSR E b MWRI

DD AMSR AMSR E SD AMSR MWRI SD AMSR E MWRI

SD AMSR MWRI T T

SD AMSR E MWRI T T

Example global distributions of estimated ascending orbit biases between uncalibrated AMSR2 and AMSR-E baseline observations for the H-Polarized (a) 23GHz, and (b) 18GHz channels, respectively (areas with correlation coefficient R<0.95 are marked in grey)

(a)23 GHz

(b)18 GHz

Source: Du et al. 2014. Remote Sensing 6


H-Polarized 18 GHz Tb comparisons from overlapping MWRI and AMSR ascending orbit observations for the selected Amazon tropical reference area

Sensor Inter-Calibration– Linear Calibration

_ 1 1 _b AMSR E b MWRIT a b T _ 2 2 2 _b AMSR b MWRIT a b T

_ _ 2b AMSR E b AMSRT a b T

Linear Calibration of AMSR2 and MWRI against AMSR-E observations


Towards Consistent Land Surface Parameter Records --Part II

Precipitable Water Vapor (PWV) & Surface Air Temperature Retrieval over Land from AMSR2


Development of PWV Retrieval Algorithm for AMSR2

PWV=𝑎0+𝑎1𝑇 𝑠+𝐴𝑣

𝑎𝑣 23−𝑎𝑣 18

(𝑎2+𝑎3 ex p(−𝐻))+𝑎4 log(∆𝑇 𝑏 (89.0 )∆𝑇 𝑏 (36.0 ) )

𝐴𝑣=−[ 12

log(MAWVI𝛽 )cos (𝜃 )+𝑎𝑜23−𝑎𝑜18 ]MAWVI=𝛽 ∙𝛤 (23.8 )𝛤 (18.7 )

Ascending Descending

Comparisons between AMSR2 & AIRS PWV retrievals over 200 global WMO sites

Source: Du et al. 2014. TGARS (In-review)


Comparisons of PWV Seasonal Distribution Patterns from Three Products

AMSR2

MODIS

NVAP-M

AMSR2

MODIS

NVAP-M

Winter (DJF) Summer (JJA)

Source: Du et al. 2014. TGARS (In-review)


Updated AMSR2 Surface Air Temperature Retrieval Algorithm

Winter (DJF) Summer (JJA)

AMSR2 mean daily maximum air temperature over Winter (left) and Summer (Right)

Tmn RMSE Tmn(Kelvin)

Tmx RMSE Tmx(Kelvin)

0.81 3.43 0.88 3.40

𝑇𝑚𝑥=3.90+0.90 (𝑇 𝑠 )−16.23𝑇 𝑐+18.54𝑇 𝑐2−0.14 (|(𝐿𝑎𝑡 )|)+3.42𝛶 cos (𝑡)

𝑇𝑚𝑛=1.12+0.89 (𝑇 𝑠 )−2.02𝑇𝑐+4.87𝑇 𝑐2−0.12 (|(𝐿𝑎𝑡)|)+3.61𝛶 cos(𝑡 )


Towards Consistent Land Surface Parameter Records-- Part III

Generating Long-term Land Surface Parameters from 2002-2014 with Updated

UMT Land Parameter Algorithm


Evaluation of the Extended Daily Maximum/Minimum Temperature Records based on WMO site measurements

AMSR-E (2010)

AMSR2 & MWRI (2012)

AMSR2(2013)


Statistical Summary of the Validation Results

Retrieved Air Temperature vs WMO measurements

Tmin Tmax

RMSE(Celcius)

R^2 RMSE(Celcius)

R^2

AMSR-E (year 2010) 3.5 0.79 3.4 0.86AMSR2&MWRI (year 2012) 3.5 0.78 3.5 0.86

AMSR2 (year 2013) 3.5 0.77 3.4 0.87

Retrieved PWV vs AIRS products


RMSE (mm)

R^2 RMSE (mm)

R^2

AMSR-E (year 2010) 4.6 0.83 5.9 0.74AMSR2 &MWRI (year 2012) 5.1 0.76 6.9 0.63

AMSR2 (year 2013) 4.7 0.80 6.2 0.72

Surface Air Temperature Validation Results

Water Vapor Validation Results


Recent Ecological Application Studies


Spring Hydrology Determines Summer Net Carbon Uptake in Northern Ecosystems

Major Findings:• Wetter springs promote summer net carbon

uptake independent of temperature effects;• Warming still promotes widespread greening

(as observed by NDVI), but with less net carbon uptake in warmer, drier years;

• Stronger coupling of northern carbon & water cycles with continued climate warming.

Surface soil moisture anomaly for June, 2009 from satellite observations (1AMSR-E); positive values denote wetter-than-normal conditions from the mean (2002-2011).

A synthesis of atmospheric CO2 and satellite and regional climate data reveals the major role of spring hydrology in determining summer net carbon uptake (NCU) for northern (≥50°N) ecosystems. Spring wetting inhibits fire emissions and promotes NCU, independent of temperature effects.

Summer (JJA) Net Ecosystem CO2 Exchange (NEE) anomaly for 2009 from CarbonTracker; NEE +/- sign denotes ecosystem carbon gain/loss, where NCU ≈ NEE + fire emissions.

Source: Yi et al. 2014. ERL 9, 064003


Surface Water Inundation in the boreal-Arctic: Potential Impacts on Regional Methane Emissions

Key Findings:• Strong temporal variability in boreal-Arctic Fw, with

sensitivity to regional temp. & precip. patterns; • Longer-term (2003-2011) drying across boreal

ecosystems, with substantial Fw wetting in Arctic tundra & continuous permafrost landscapes (Above);

• Accounting for dynamic changes in high-latitude wetland extent (e.g. Fw inputs) can significantly reduce regional CH4 emission estimates.

Sponsors: NASA Earth Science program

Above: AMSR-E Fw retrievals indicate that ~5% (8.4 x 105 km2) of northern tundra & peatland landscapes are inundated during non-frozen summer months.

1ERL Video Abstract: http://bcove.me/qcjjracp

A satellite data-driven model study of surface temperature and AMSR-E daily fractional open water (Fw) inundation controls on high-latitude wetland CH4 emissions reveals a strong regulating influence by contrasting regional wetting and drying patterns.

Source: Watts et al. 2014. ERL 9, 075001

http://bcove.me/qcjjracp


Recent Publications

Journal papers:

Du, J.; Kimball, J.S.; Shi, J.; Jones, L.A.; Wu, S.; Sun, R.; Yang, H. Inter-Calibration of Satellite Passive Microwave Land Observations from AMSR-E and AMSR2 Using Overlapping FY3B-MWRI Sensor Measurements. Remote Sens. 2014, 6, 8594-8616.

Du, J., J.S. Kimball, and L.A. Jones, 2014. Satellite microwave retrieval of total precipitable water vapor and surface air temperature over land from AMSR2. TGARS (In-review).

Jang, K., S. Kang, J.S. Kimball, et al. 2014. Retrievals of all-weather daily air temperature using MODIS and AMSR-E data. Remote sensing, 6, 9, 8387-8404.

Watts, J.D., J.S. Kimball, A. Bartsch, and K.C. McDonald, 2014. Surface water inundation in the boreal-Arctic: potential impacts on regional methane emissions. ERL, 9, 075001.

Yi, Y., J.S. Kimball, and R.H. Reichle, 2014. Spring hydrology determines summer net carbon uptake in northern ecosystems. ERL 9, 064003.


Summary

• Detectable biases found between AMSR-E, AMSR2 and MWRI1 observations. Inter-calibrations based on swath Tb data records significantly decreased sensor biases and improved Tb correlations.

• UMT Global Land Parameter algorithms successfully adapted to AMSR2, with favorable accuracy in surface air temperature and water vapor retrievals. The updated algorithms applied to calibrated AMSR2, MWRI & AMSR-E (V7) Tb data.

• Based on AMSR-E and calibrated AMSR2/MWRI Tb observations and recent algorithm updates, long-term UMT land surface parameter records have been produced. In general, similar retrieval accuracy has been found for the AMSR-E and post AMSR-E periods, except that MWRI water vapor retrieval is slightly lower than the other sensor products.

• Continuing calibration & extension of UMT record planned in support of several global ecosystem studies.

1AMSR-E V7 reprocessed Tb record provided by Remote Sensing Systems; AMSR2 L1R data are from JAXA; MWRI data are from China National Satellite Meteorological Centre

Thank You!

Funding for this study provided from NASA Terra and Aqua Science, and MEaSUREs programs.

NTSG Project Team: John Kimball, Jinyang Du, Lucas Jones, Youngwook Kim, Joe Glassy, Matt Jones, Jennifer Watts,

Yonghong Yi

Project Data Archives (Updates coming soon!):

http://nsidc.org/data/nsidc-0451

http://freezethaw.ntsg.umt.edu


Backup Slides


Algorithm Flowchart

Temperature Algorithm

Tb 6.9 or 10.7 V & H pol. Tb 18.7, 23.8 V & H pol.

60-day running smoother

60-day running smoother

Estimate slope parameter:

Estimate emissivity

Invert for VOD(assume dry baseline soil conditions)

Invert for SM

Selection of the WMO stations & GPS sites for PWV & Air Temperature Retrieval Study

PWV/Temperature Algorithm Development: WMO Training Sites (black Square) and Validation Sites (black triangle) over the MODIS IGBP

global land cover map; PWV Algorithm Validation: SuomiNet North American GPS stations

(white circles with black outlines)


AMSR2 PWV Validation using GPS retrievals


AMSR2 PWV vs GPS PWV

Histogram of the absolute AMSR2 PWV estimation errors for the ascending (left) and descending (right) orbit retrievals relative to independent PWV

observations from 350 SuomiNet North American GPS sites


Extended Land Parameter Records – Global Vegetation Optical Thickness Distributions

X-band Vegetation Optical Thickness (May 30, 2010,2011,2013 / May 31,2012)

AMSR-E (2010)

AMSR-E (2011)

MWRI(2012) AMSR2 (2013)

jinyang du, lucas a. jones, and john s. kimball

Documents