Southwestern US with MISR Mark Chopping, Lihong Su, Libertad Urena - - PowerPoint PPT Presentation

Mapping Forest Crown Cover, Canopy Height, and Biomass in the Southwestern US with MISR

Mark Chopping, Lihong Su, Libertad Urena (Montclair State University), Gretchen Moisen (USFS); Andrea Laliberte, Albert Rango, Debra P. C. Peters (USDA, ARS); and John V. Martonchik (NASA, JPL) NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6 2007 http://csam.montclair.edu/~chopping/wood

Quick Tim e™ and a TIFF (Uncom pressed) decompressor are needed to see this picture.

QuickTime ™ and a TIFF (Unco mp res sed) de com pr essor ar e needed to see th is pictu re.

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Acknowledgments

This research was supported by NASA Earth Observing System grant NNG04GK91G to MC. The MISR data used were obtained from the NASA Langley Research Center Atmospheric Science Data Center. US-FIA Maps for the Interior West were provided by the US Forest

• Service. Rocky Mountain Research Station, Ogden, UT.

Thanks to Matt Smith and the Global Land Cover Facility, University of Maryland, College Park, MD, for providing the VCF , SRTM, and SDB data. The Jornada Experimental Range is administered by the USDA Agricultural Research Service and is a Long Term Ecological Research site supported by the National Science Foundation.

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Large area mapping of southwestern forest crown cover, canopy height, and biomass using MISR: Introduction

Data from the Multiangle Imaging SpectroRadiometer (MISR) on the NASA Earth Observing System Terra satellite can be interpreted through a simple geometric-optical (GO) model (SGM) to retrieve estimates of crown cover, mean canopy height, and woody biomass. These are important parameters in western forests that are increasingly vulnerable to wildfire with earlier melting of the snow

• pack. Method: MISR Level 1B2 Terrain radiance data from overpasses

in May and June 2002 over S.E. Arizona and S. New Mexico were atmospherically corrected using smoothed MISR aerosol optical depth data @558 nm and bidirectional reflectance factors (BRF) were mapped to a 250 m grid. The background angular response in the MISR viewing plane was estimated prior to model inversion using the isotropic, geometric, and volume scattering weights of a LiSparse- RossThin kernel-driven model, plus nadir camera blue, green and near- infrared reflectance factors. Calibration of these relationships was effected using woody plant cover estimates obtained from Ikonos 1 m panchromatic imagery in the USDA, ARS Jornada Experimental Range. The SGM was adjusted against red MISR BRFs in 9 views using the Praxis algorithm to minimize the Root Mean Square Error (RMSE) between model and data, producing maps of crown cover, mean canopy height and woody biomass.

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

GO models predict BRF at specified illumination and viewing directions as a function of stand-scale canopy structure parameters of practical and ecological significance:

• mean crown shape (b/r)
• mean crown radius (r)
• plant number density ()
• mean canopy height (h)
• background brightness

and anisotropy (functions

• f understory density and

soil brightness)

sunlit crown shadowed crown sunlit ground shadowed ground

R = G . kG + C . kC + T . kT + Z . kZ G, C, T, Z can be assumed Lambertian or may have defined reflectance anisotropies

Courtesy: Wenge Ni-Meister

Number density = # plants distributed within a unit area

Large area mapping of southwestern forest crown cover, canopy height, and biomass using MISR: Introduction

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

inv ert the LiSparse-RossThin k ernel-driven m

• del against

MISR data

• btain relationships between the

k ernel weights and the back ground BRDF in the MISR plane by fixing shrub statistics and adjusting the Walthall model param eters invert the SGM m

• del

using the estim ated back ground and fix ing all parameters ex cept mean shrub radius; adjust the model against MISR data

• btain estimates of woody

shrub mean radiius and num ber density v ia thresholding 1 m IKONOS panchromatic imagery calculate fractional shrub cov er from mean shrub density MISR Red Band BRFs in Nine Cameras

Quic kTime™ and a TIFF (Unc ompresse d) de compress or are nee ded to see th is pic ture. Mesquite shrubs Black grama grass Exposed soil

Figure 1. SGM example protocol for retrieval of shrub cover

The Walthall model (W) is used to represent the background and a Ross function allows for within- crown volume scattering; G is replaced with W and C with Ross, so that

R = G W.kG +C Ross.kC

T and Z are considered black

Red band in 9 views

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Large area mapping of southwestern forest crown cover, canopy height, and biomass using MISR: Method

Fractional crown cover was calculated by adjusting crown radius (r) with fixed tree number density (), exploiting sensitivity to brightness. Canopy height was calculated by adjusting b/r with fixed h/b, where b is vertical crown radius and h is crown center height above the reference plane, exploiting sensitivity to BRF shape (Figure 2). The starting point for the inversions was r = 0.25, b/r = 0.2, with fixed parameters set to 0.012 (), 2.0 (h/b), 0.09 (leaf reflectance) and 2.08 (crown foliage density). Model fits were good (Figure 3), with a RMSE mean (standard deviation) of 0.012 (0.025) and a mode of 0.004

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Figure 2. The effects of changing (a) fractional crown cover with plant number density ( fixed at 0.012 and maintaining canopy height at 3.0 m (b) crown shape (b/r), maintaining h/b fixed at a typical value of 2.00, where r, b and h are crown horizontal radius, vertical radius and center height. If we have r and b/r, and h/b is

fixed, we can obtain h. We only have an estimate of r -- but this does not appear to be critical.

0.00 0.05 0.10 0.15 0.20 0.25 0.30

• 80
• 60
• 40
• 20

20 40 60 80

View Zenith Angle (°)

fcov=0.1 fcov=0.5

(a)

0 .1 0 0 .1 5 0 .2 0 0 .2 5 0 .3 0

• 8 0 -6 0 -4 0 -2 0

2 0 4 0 6 0 8 0 V iew Z enith A ngle (

b/r=0.5 b/r=3.0

(b)

Adjustment of the Model for r and b/r with Praxis:

Fractional crown cover is from r with fixed  (sensitive to brightness) Canopy height is from b/r with fixed h/b (sensitive to BRF shape)

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Model fits to MISR data were good with a RMSE mean (standard deviation) of 0.012 (0.025) and a mode

• f 0.004 for the

composited maps. Figure 3 shows typical fits to MISR data for a range of cover values.

Model Fits to MISR Data

Figure 3. Model fits to MISR data

0.00 0.05 0.10 0.15 0.20 0.25

• 80.0
• 60.0
• 40.0
• 20.0

0.0 20.0 40.0 60.0 80.0

cover: 0.20 height: 5.8 m

0.00 0.05 0.10 0.15 0.20 0.25

• 80.0
• 60.0
• 40.0
• 20.0

0.0 20.0 40.0 60.0 80.0

cover: 0.47 height: 11.2 m

0.00 0.05 0.10 0.15 0.20 0.25

• 80.0
• 60.0
• 40.0
• 20.0

0.0 20.0 40.0 60.0 80.0

cover: 0.18 height: 4.9 m

0.00 0.05 0.10 0.15 0.20 0.25

• 80.0
• 60.0
• 40.0
• 20.0

0.0 20.0 40.0 60.0 80.0

cover: 0.34 height: 6.7 m

View Zenith Angle (°) View Zenith Angle (°) – – –

0.00 0.05 0.10 0.15 0.20 0.25

• 80.0
• 60.0
• 40.0
• 20.0

0.0 20.0 40.0 60.0 80.0

cover: 0.88 height: 6.2 m

0.00 0.05 0.10 0.15 0.20 0.25

• 80.0
• 60.0
• 40.0
• 20.0

0.0 20.0 40.0 60.0 80.0

cover: 0.65 height: 13.4 m

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Large area mapping of southwestern forest crown cover, canopy height, and biomass using MISR: Results

The results from nine Terra overpasses were merged using model fitting error as the compositing criterion, producing almost cloud- and cloud shadow-free maps. 1063 random locations in forest were used to extract MISR/SGM and corresponding reference data from US Forest Service maps for the Interior West, based on Forest Inventory Analysis (FIA) survey data, elevation, slope, aspect, soils, existing eco-region and land cover maps, MODIS data, and climate variables (Figure 4). Topographic effects were visible in the extracted cover data as an isolated cluster (Figure 5 (a)). Filtering on high RMSE, outliers, and screening for topographic shading reduced N to 576 (54%).

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Distribution of Reference Data Points

Quick Tim e™ and a TIFF (Uncom pressed) decompressor are needed to see this picture.

Arizona New Mexico

Tucson, A Z • lava flow Rio Grande

• White

Sands

Figure 4. MISR/GO crown cover map in grayscale showing illuminated reference locations (red points) and those impacted by topography (blue points).

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

MISR/GO Forest Structure Mapping: First-order Validation

Figure 5. (a)-(c) Retrievals vs. US Forest Service Map Data. The upper cluster of data points in (a) corresponds to locations affected by severe topography (d)-(f) Retrievals with screening for topographic effects using a Digital Elevation Model from the Shuttle Radar Topography Mission. Points with RMSE >= 0.01 and outliers ±2 st. devs. from the mean of crown cover were discarded, retaining 576 points (54%).

R2 = 0.74 0.0 20.0 40.0 60.0 80.0 0.0 20.0 40.0 60.0 80.0 USFS Biomass (tons/acre) R2 = 0.81 0.0 20.0 40.0 60.0 80.0 0.0 20.0 40.0 60.0 80.0 USFS Biomass (tons/acre) R2 = 0.68 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 USFS Crown Cover R2 = 0.33 0.0 10.0 20.0 30.0 0.0 10.0 20.0 30.0 USFS Weighted Height (m) R2 = 0.69 0.0 10.0 20.0 30.0 0.0 10.0 20.0 30.0 USFS Weighted Height (m)

(e) (f) (a) (b) (c)

R2 = 0.78 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 USFS Crown Cover

(d)

AFFECTED BY TOPOGRAPHIC SHADING (must be screened out)

OK

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

The maps also include woody shrubs in desert grasslands as the approach relies mainly on surface physical characteristics (Figures 6-8). The mean absolute error in estimates of fractional crown cover, mean canopy height, and woody biomass were 0.10, 2.2 m, and 4.5 tons acre-1 (10.1 Mg ha-1), with RMSE errors of 0.12, 3.3 and 6.2 (14.0), respectively (Table 1).

Large area mapping of southwestern forest crown cover, canopy height, and biomass using MISR: Results

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

• Fig. 6: Regional Aboveground Biomass

0.00 >50.0

Tons/acre

Aboveground Biomass

0.00 >110.0

Mg ha-1

15.0 6.70

100 km

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

• Fig. 7: Regional Mean Canopy Height

1.00 >20.00

meters

Mean Canopy Height

< 1.00

Quick Tim e™ and a TIFF (Uncom pressed) decompressor are needed to see this picture.

USFS-Interior West Map

AZ NM TX Mexico 100 km

Reference

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

• Fig. 8: Regional Forest Crown Cover

0.00 0.29 Shrubs

Crown Cover

Bright white areas show anomalies (high error on model fitting) 0.30 0.90 Forest 100 km

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

• Fig. 9: RMSE on Model Fitting

0.019 >0.140

Rejected (mostly desert grassland)

RMSE

<0.003 0.016

Accepted (includes almost all forest)

The large area with high RMSE is White Sands National Monument (gypsum dunes and alkali flats) 100 km

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Q ui ckT i m e™ and a TI FF (U ncom pr essed) decom pressor ar e needed t o see t hi s pi ct ur e.

50 km

< 6 > 60

Estimated Biomass (tons/acre)

Source: USFS-IWFIA Map

Arizona New Mexico

Jornada Sevilleta Fig 10: Reference: USDA Forest Service map showing estimated forest biomass for the Interior West on a hill-shaded background; the Arizona / New Mexico border; and the locations of the Jornada Experimental Range and the Sevilleta National Wildlife refuge (NM). Light lines indicate county boundaries.

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

0.0 10.0 20.0 30.0 40.0 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162 169 176 183 190 1 176 183 190 197 204 211 218 225 232 239 246 253 260 267 274 281 288 295 302 309 316 323 330 337 344 351 358 365 372 379 386

Sample #

5 372 379 386 393 400 407 414 421 428 435 442 449 456 463 470 477 484 491 498 505 512 519 526 533 540 547 554 561 568 575

Figure 11. Plot of Retrieved Height vs USFS IW-FIA Map Height

Sample #

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

Table I: Error Analysis

Fractional crown cover (dimensionless) Mean canopy height (meters) Woody Biomass (tons acre-1) (Mg ha-1) Mean Relative Error (%) 30 28 28

• Mean Absolute Error

0.10 2.2 4.5 10.1 Mean (MISR) 0.48 10.3 21.8 49.0 Mean (USFS) 0.38 8.7 21.8 0.0 Root Mean Square Error 0.12 3.3 6.2 14.0 R2 0.78 0.69 0.81

• Large area mapping of southwestern forest crown cover,

canopy height, and biomass using MISR: Results

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

These results show that MISR data can be interpreted through a simple GO model to provide maps of canopy crown cover, canopy height, and biomass

• ver large areas that are highly compatible with US

Forest Service data. A more robust validation is being pursued using FIA field survey data. To our knowledge, these results demonstrate the first use of passive moderate resolution EOS data to map forest, cover, canopy height, and woody biomass over large areas.

Large area mapping of southwestern forest crown cover, canopy height, and biomass using MISR: Conclusions

NASA Land Cover Land Use Change Science Team Meeting, Adelphi, MD, April 4-6, 2007

http://www-misr.jpl.nasa.gov/ http://csam.montclair.edu/~chopping/wood