3D Data from Interferometric SAR and Stereo Radargrammetry …earth.esa.int/dragon-2015-programme/karila-3d_data_from... · 3D Data from Interferometric SAR and Stereo Radargrammetry

3D Data from Interferometric SAR and Stereo Radargrammetry in Mapping of Forest Resources

Kirsi Karila1, Mika Karjalainen1, Xiaowei Yu1, Mikko Vastaranta2 1Finnish Geospatial Research Institute, Finland; 2University of Helsinki, Finland

Purpose of this study • 3D digital elevation models (DEM) are created from

satellite SAR data using two alternative approaches: – 1) SAR interferometry (InSAR) – 2) radargrammetry

• These DEMs are related to the forest canopy height models using accurate digital terrain model based on airborne lidar surveys.

• The study area is located in Evo, Southern Finland, from which plot- and stand-wise reference data of standard forest resources related parameters are available.

3D SAR techniques INSAR Stereo-radargrammetry

• Phase difference

• Baseline of hundreds of meters

• Simultaneous imaging required

• Stereoscopic measurement (automatic image matching)

• Baseline of hundreds of kilometers

• Similar conditions required

[0, 2π]

Study area: Evo

Aerial image © NLS

Evo campaign 2014 • Field inventory, TLS • MLS with ATV and PLS using backpack • Satellite and airborne data acquisition

Mobile Laser Scanning

Reference measurements • 91 test plots, 32m x 32m

• Pine 45 %, spruce 33 % and deciduous trees 22% of total volume

BA, m2/ha

Dg, mm Hg, dm V,

m3/ha

Total biomass, tons/ha

mean 26.8 257.8 210.9 270.1 134.5

min 6.6 139.2 100.2 34.5 19.1

max 43.2 464.2 310.9 518.4 230.6

std 7.8 74.6 43.9 109.4 48.1

SAR data Date Incidence

angle Time (UTC) Weather

9.7.2014 (D1) 26° 4:57 19°C, Clear

4.7.2014 (D2) 36° 4:48 14°C, Clear

29.6.2014 (D3) 44° 4:40 14°C, Rain showers

9.7.2014 (A1) 30° 15:45 27°C, Clear

3.7.2014 (A2) 40° 15:54 18°C, Clear

8.7.2014 (A3) 47° 16:03 25°C, Possible rain showers

5.6.2014 (TDX InSAR pair) 48° 16:03 26°C, Clear

6 Terrasar-X images, ascending and descending orbits ( SpotLight 1 m ) 6 stereo radargrammetric

pairs: point clouds combined after radargrammetric processing

1 TanDEM-X InSAR pair (stripmap ~3 m resolution) Baseline 190 m, HoA 45 m

HH polarization

Digital surface models TDX InSAR:

average (min-max) m

TSX stereo: average (min-

max) m

TSX - TDX average

(m)

max_H 19.1 (6.3-38.7) 22.7 (12.4-35.8) 3.6

mean_H 14.4 (4.3-31.9) 13.4 (7.1-24.0) -1.0

std_H 2.5 (0.8-6.9) 4.7 (2.4-7.7) 2.2

Stereo

InSAR

Aerial image and InSAR CHM

Aerial image © NLS

Forest attribute prediction • Attributes studied: Basal area, stem volume, biomass, mean

DBH and Lorey’s height (basal area weighted mean height) • From the SAR CHMs mean, max, min, std height, CV,

10,20,30,40,50,60,70,80,90% percentiles of height distribution for the plots were extracted for the 32x32 m areas

• Area based approach and random forests (RF) technique were used in the prediction of the forest plot attributes

• Forest variables measured in the field were used as target observations and plot-specific metrics derived from SAR data were used as predictors.

• The predicted were compared to the reference values

Predicted vs. observed plots

InSAR Stereo

Mean height (m) Basal Area (m2/ha) Biomass (Mg/ha)

Mean DBH (cm) Volume (m3/ha)

Predicted

obse

rved

Results InSAR Stereo

Bias Bias

(%) RMSE RMSE (%) R

Mean height (m) 0.03 0.14 2.02 9.56 0.89

Mean DBH (cm) -0.05 -0.20 4.87 18.88 0.76

Basal area (m2/ha) -0.01 -0.04 5.51 20.57 0.71

Volume (m3/ha) 0.08 0.03 57.53 21.30 0.85

Biomass (Mg/ha) -0.80 -0.60 26.96 20.05 0.83

Bias Bias

(%) RMSE RMSE (%) R

Mean height (m) -0.01 -0.07 2.97 14.09 0.74

Mean DBH (cm) 0.02 0.09 5.18 20.11 0.72

Basal area (m2/ha) -0.04 -0.14 6.87 25.64 0.50

Volume (m3/ha) 0.55 0.20 84.55 31.30 0.65

Biomass (Mg/ha) 0.26 0.19 38.40 28.55 0.61

Conclusions and future • InSAR results (1 pair) slightly better than stereo

(6 pairs) • Forest inventories of wide areas are possible

using SAR data combined with an accurate terrain model

• Forest change detection using SAR 3D data • Comparison to other 3D and 2D remote sensing

techniques

Thank you!

More information www.fgi.fi