S. Kim - SMAP
S. Kim - SMAP
S. Kim - SMAP
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
National Aeronautics and Space Administration<br />
UAVSAR data,<br />
soil moisture,<br />
crop structure data<br />
3 rd Canadian<br />
<strong>SMAP</strong> workshop<br />
March 20-21, 2013<br />
Seungbum <strong>Kim</strong>*, Xiaolan Xu*,<br />
Andreas Colliander*, Thomas Jackson**,<br />
Heather McNairn***, Eni Njoku*<br />
(also thanks to Narendra Das, Grant Wiseman,<br />
& <strong>SMAP</strong>VEX12 team)<br />
* Jet Propulsion Laboratory, Caltech<br />
** Hydrol. Remote Sensing Lab, USDA<br />
*** AAFC<br />
Copyright 2013. All rights reserved<br />
- 1 -
UAVSAR postprocessing<br />
<br />
<br />
<br />
Incidence angle normalization to 40degs<br />
– The original incidence angle ranges from 25 to 60 degs<br />
Removing heterogeneous (man-made) structures<br />
Matchup with in situ data<br />
Strip ID: 31606<br />
Copyright 2013. All rights reserved - 2 -
UAVSAR Availability<br />
flight # 31603 31604 31605 31606 09002 27002 PALS<br />
6/17 Y Y Y Y Y Y Y<br />
6/19 Y Y Y Y<br />
6/22 Y Y Y Y Y Y Y<br />
6/23 Y Y Y Y Y Y<br />
6/25 Y Y Y Y Y Y Y<br />
6/27 Y Y Y Y Y Y Y<br />
6/29 Y Y Y Y Y Y Y<br />
7/3 Y Y<br />
7/5 Y Y Y Y Y Y Y<br />
7/8 Y Y Y Y Y Y Y<br />
7/10 Y Y Y Y Y Y Y<br />
7/13 Y Y Y Y Y Y Y<br />
7/14 Y Y Y Y Y Y<br />
7/17 Y Y Y Y Y Y Y<br />
PALS also flew on 6/7, 6/12, 6/15, 7/19<br />
Copyright 2013. All rights reserved - 3 -
Correction Flow Chart *<br />
Raw Data<br />
4 Diagonal<br />
passes<br />
Or<br />
1 Horizontal<br />
passes<br />
Calculate<br />
Statistics<br />
mean(inc,veg)<br />
std(inc,veg)<br />
inc = 21 to 65°<br />
veg: 24 types<br />
Correction<br />
HIST-method<br />
image 1<br />
image 2<br />
image 3<br />
image 4<br />
Final<br />
image at<br />
40 degree<br />
For each pixel<br />
σ norm,VC<br />
= σ 40,VC<br />
+ ˆσ 40,VC<br />
⋅ σ raw _ inc,VC −σ raw _ inc,VC<br />
ˆσ raw _ inc,VC<br />
*Mladenova, I. E.; Jackson, T. J.; Bindlish, R.; Hensley, S.; , "Incidence Angle Normalization of Radar Backscatter Data,"<br />
Geoscience and Remote Sensing, IEEE Transactions on , vol.PP, no.99, pp.1-14, 0<br />
Copyright 2013. All rights reserved - 4 -
Incidence angle normalization<br />
Raw Data Corrected Data<br />
0<br />
Due to no<br />
landcover data<br />
-25<br />
Copyright 2013. All rights reserved - 5 -
Filtering heterogeneous objects<br />
<br />
<br />
Green and red polygons are the field boundaries before and after the filtering,<br />
respectively. The number is the field ID.<br />
The normalization accuracy is assessed with overlapping passes using the 40deg strip<br />
(40 +/-0.5).<br />
40deg strip<br />
Copyright 2013. All rights reserved - 6 -
Normalization: QA - quantitative<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
Norm minus Raw (dB): HH<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
VV<br />
<br />
to<br />
0<br />
the class.<br />
HV<br />
QA was performed per each<br />
field, 5 then compiled over each<br />
crop class where there are<br />
more than one fields belonging<br />
-0.5<br />
-0.5<br />
-1.0<br />
-1.5<br />
bean<br />
pasture<br />
corn<br />
wheat<br />
w.wheat<br />
canola<br />
forage<br />
soybean<br />
-1.0<br />
-1.5<br />
bean<br />
pasture<br />
corn<br />
wheat<br />
w.wheat<br />
canola<br />
forage<br />
soybean<br />
<br />
The -5 error is < 1dB rmse (bean,<br />
corn, wheat, canola)<br />
– Forage, winter wheat,<br />
soybean: too few samples<br />
bean<br />
pasture<br />
corn<br />
wheat<br />
w.wheat<br />
canola<br />
forage<br />
soybean<br />
50<br />
45<br />
40<br />
35<br />
Inc. Angle<br />
20-30<br />
30-50<br />
50-65<br />
10<br />
8<br />
6<br />
4<br />
2<br />
# fields/day<br />
30-50deg strip offers at least 1<br />
field per crop class per day à<br />
enabling soil moisture analysis<br />
with 30-50deg strip only.<br />
<br />
HV needs further work<br />
30<br />
0<br />
Tue Mar 12 00:47:24 2013 sm_vex12_norm_eval130312.pro<br />
Copyright 2013. All rights reserved - 7 -
Soil moisture retrieval: method<br />
Land cover<br />
Forward model: corn,<br />
bean, pasture (wheat).<br />
Trained with past<br />
campaign data<br />
Burgin et al. TGRS 2011<br />
Van Zyl, book, 2011<br />
Huang et al. TGRS 2012<br />
<strong>Kim</strong> et al. TGRS in press<br />
Collect in situ data: Mv,<br />
roughness, VWC, texture,<br />
land cover<br />
In situ Mv,<br />
in situ roughness<br />
VWC,<br />
texture<br />
Retrieve Mv & roughness<br />
<strong>Kim</strong> et al. TGRS 2012, in press<br />
Wagner et al. 199x, 200x<br />
UAVSAR time-series data<br />
Accuracy Analysis<br />
HH HV VV<br />
Copyright 2013. All rights reserved - 8 -
Soil moisture retrieval (preliminary)<br />
dB<br />
Mv, cm 3 /cm 3<br />
-10<br />
-12<br />
-14<br />
-16<br />
-18<br />
-20<br />
-22<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
170 175 180 185 190 195 200<br />
DAY<br />
Pasture field, VWC ~ 0.3 kg/m2<br />
Mv RMSE: 0.086, mean Δ: -0.084<br />
r (time-series), r(Wagner): 0.81, 0.71<br />
dB<br />
1.0<br />
0.8<br />
Mv, cm 3 /cm 3<br />
relative change index<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
-10<br />
-12<br />
-14<br />
-16<br />
VV-cube -18<br />
HH-cube -20<br />
-22<br />
0.7<br />
0.6<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
VV-Obs.<br />
HH-Obs.<br />
in situ<br />
time-series<br />
snapshot<br />
dubois<br />
yunjin<br />
wagner<br />
170 175 180 185 190<br />
DAY<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
Wheat field: VWC = 1.7 to 2.5 kg/m2<br />
Mv RMSE: 0.086, mean Δ: -0.058<br />
r (time-series), r(Wagner): 0.94, 0.94<br />
relative change index<br />
VV-cube<br />
HH-cube<br />
VV-Obs.<br />
HH-Obs.<br />
in situ<br />
time-series<br />
snapshot<br />
dubois<br />
yunjin<br />
wagner<br />
Copyright 2013. All rights reserved - 9 -
Soil moisture retrieval (cont)<br />
dB<br />
Mv, cm 3 /cm 3<br />
-10<br />
-12<br />
-14<br />
-16<br />
-18<br />
-20<br />
-22<br />
0.30<br />
0.25<br />
0.20<br />
0.15<br />
0.10<br />
0.05<br />
0.00<br />
170 175 180 185 190 195 200<br />
DAY<br />
Bean field: VWC=0.1 to 1.9 kg/m2<br />
Mv RMSE: 0.053, mean Δ: -0.013<br />
r (time-series) r(Wagner): 0.45, 0.66<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0.0<br />
relative change index<br />
VV-cube<br />
HH-cube<br />
VV-Obs.<br />
HH-Obs.<br />
in situ<br />
time-series<br />
snapshot<br />
dubois<br />
yunjin<br />
wagner<br />
<br />
<br />
Corn’s Mv retrieval needs<br />
more work (mostly on forward<br />
model side)<br />
Forward model quality<br />
– Wheat: immature<br />
– Pasture: mature<br />
– Soybean, corn: medium<br />
– Canola: future<br />
[McNairn et al. in prep]<br />
Copyright 2013. All rights reserved - 10 -
Summary and plans<br />
<br />
<br />
<br />
UAVSAR<br />
– Normalization error is smaller than 1dB rmse (co-pol)<br />
– Removed heterogeneous objects<br />
– More than one passes were very useful<br />
– HV normalization needs work:<br />
<br />
when to put into archive HH/VV only or HH/VV/HV<br />
Soil moisture retrieval<br />
– Reasonable retrievals with an rmse of ~0.05 cm3/cm3 (individual field; need to<br />
improve mean error)<br />
– Keep improving forward models<br />
Crop structure data<br />
– In progress<br />
– Will be used to improve forward models<br />
Thanks for hosting the workshop!<br />
Copyright 2013. All rights reserved - 11 -
ackup<br />
Copyright 2013. All rights reserved - 12 -
Crop structure measurements<br />
<br />
<br />
<br />
<br />
Objectives<br />
– Help understand radar response<br />
Measreuments<br />
– Complete characterization of geometry of crops<br />
<br />
<br />
<br />
<br />
Length: plant, stalk, branch, leaf<br />
Angle: stalk, branch, leaf<br />
Diameter or thickness: stalk, branch, leaf<br />
Density or number: stalk, branch, leaf<br />
– Collocated & coincident VWC measurements<br />
– Fast-growing crops were sampled more frequently<br />
Status<br />
– Geometry data are complete<br />
– Some glitches in dry weight records (missing records and mislabels)<br />
Plan<br />
– Cross-compare between vegetation teams’ VWC and structure team’s VWC, as a QC<br />
Copyright 2013. All rights reserved - 13 -
Crop Structure Sampling Dates<br />
70Jun012<br />
90Jun012<br />
110Jun012<br />
120Jun012<br />
130Jun012<br />
150Jun012<br />
160Jun012<br />
170Jun012<br />
180Jun012<br />
190Jun012<br />
210Jun012<br />
220Jun012<br />
230Jun012<br />
240Jun012<br />
250Jun012<br />
270Jun012<br />
280Jun012<br />
290Jun012<br />
300Jun012<br />
50Jul012<br />
70Jul012<br />
80Jul012<br />
90Jul012<br />
100Jul012<br />
130Jul012<br />
140Jul012<br />
160Jul012<br />
170Jul012<br />
180Jul012<br />
Pasture Corn Beans Soybean W0Wheat Wheat Forage Canola VegeTeam<br />
Copyright 2013. All rights reserved - 14 -<br />
<br />
<br />
<br />
Some missing VWC<br />
values for structure<br />
samples.<br />
Plan to calibrate the<br />
structure data using ‘Vege<br />
Team’ VWC<br />
Augment week-long gaps<br />
in Vege Team VWC<br />
week-long gaps<br />
week-long gaps