Hyperspectral Remote Sensing of Forests - (Evaluation and Validation of EO-1 for Sustainable Development (EVEOSD) Project)

Transcription

1 Hyperspectral Remote Sensing of Forests - (Evaluation and Validation of EO-1 for Sustainable Development (EVEOSD) Project) Presenter: Jay Pearlman (TRW, Co-I) PI: David G. Goodenough ( ) Co-Is: A. Hollinger (Canadian Space Agency) NRCan CFS: R. Hall, J. Iisaka, D. Leckie, A. S. Bhogal, A. Dyk J. Miller (York University), Olaf Niemann (University of Victoria) Karl Staenz (NRCan CCRS), Harold Zwick (MDA) plus Tian Han, Justin Love, Chris West (UVic, PFC) Contact : dgoodeno@nrcan.gc.ca canadien 2001 November 7

2 Introduction and Activities contains 10% of the world s forests. Multitemporal Landsat data will be a primary source for measuring indicators of sustainable development, such as forest area, forest type, biomass, disturbance, above-ground carbon, reforestation, afforestation, and deforestation, etc. We will use EO-1 data to validate that products obtainable from earlier Landsats can be obtained from ALI and Hyperion. New products not possible with earlier Landsats will also be developed for ALI, Hyperion, and LAC. Comparisons will be made of reflectances obtained from EO-1 and the Landsat series. Atmospheric correction models will be used to obtain the highest accuracies with EO-1. The efficacy of LAC and Hyperion for atmospheric correction will be assessed. canadien

3 Activities #2 The preferred bands and algorithms for forestry from Hyperion will be identified. The 4 primary test sites are Greater Victoria Watershed (BC), Clayoquot Sound (BC), Algoma (ON), and Hoquiam (WA). The 4 secondary test sites are: BOREAS SSA (SK), BOREAS NSA (MB), Hinton (AB), and Georgia Strait (BC). The EO-1 data will be evaluated for the measurement of forest structure, canopy chemistry, LAI, proportional estimation, and land cover change. Comparisons will be made between AVIRIS, casi, Hyperion, CHRIS, and ground spectral measurements. Sensor performance will be assessed in terms of radiometric and geometric correction, temporal stability, and accuracy of calibrations. canadien

4 EVEOSD Test Site Distribution canadien

5 EVEOSD Vegetation Sampling canadien

6 Sampling Locations in GVWD Tree tops 605 Douglas Fir from 55 plots +50 of five other species Salal 251 samples from 43 plots Chemical Analysis Organic and inorganic Two dates of collection: September 2000 and July-August 2001 canadien

7 Example: Plot 13 5 Salal samples, Center N E S W. Ten tree top foliage samples. Point of Commencement and bearings noted. Plots correspond to existing CFS initiatives. canadien

8 EVEOSD Database: Opening Screen canadien

9 Table Name Conifer Chemistry Conifer Inorganic Chemistry ICP- ES Different Species Conifer Inorganic Chemistry ICP- ES Samples Conifer Inorganic Chemistry ICP- MS Different Species Conifer Inorganic Chemistry ICP- MS Samples Current Conifer Chemistry Field Spectral Plot Data Climate Data Non-Current Conifer Chemistry Plot Data Plot Pictures Point of Commencement Salal Chemistry Salal Inorganic Chemistry Spectra Radiance Data Spectra Reflectance Data Spectra Site Picture Data Spectra Directory Data Weather Station Location Statistics Conifer Inorganic Statistics Conifer Organic Current Statistics Conifer Organic Noncurrent Statistics Overall Salal Inorganic Statistics Salal Inorganic Statistics Salal Organic Primary Key(s) Plot Number, Sample Number ESDifferntSpeciesID Conifer_Inorganic_ICPESID VarioustreesMSID ConiferICPMSID Current_Growth_ID Plot Number, Date Climate_ID Non_Current_Growth_ID Plot_Number Plot_Picture_ID PoCID Plot Number, Sample Number SalalInorganicID RadianceID ReflectanceID Picture_DataID Excel_File_Name Location_Name Plot_Number Plot_Number Plot_Number ElementID Plot_Number Plot_Number EVEOSD Chemical and Spectral Database relational tables and primary keys.

10 Chemistry Measurements Organic and inorganic data analyses were performed on salal and tree top samples. The organic chemical analysis conducted on the tree top samples was partitioned into current growth and non-current growth. Current growth is defined as any foliar growth that occurred within the last year. All other foliar growth is considered to be non-current growth. The organic analysis contains data on chlorophyll a and b, total chlorophyll, moisture percentage of dry weight, and nitrogen percentage for each sample. Significant conifer concentrations found were: Cd, Cu, Mn, Mo, Ni, Pb, Zn, Al, As, B, Ba, Ca, Co, Cr, Fe, Ga, K, La, Mg, Na, P, Sr, Va, and S. canadien

11 Current Chlorophyll (µg/mg) vs Current Nitrogen (%) for Douglas Fir y = 1.102x R 2 = Nitrogen (%) Chlorophyll (µg/mg) Current Chlorophyll canadien vs Current Nitrogen Linear (Current Chlorophyll vs Current Nitrogen)

12 Current Nitrogen vs Non-Current Nitrogen for Douglas Fir y = x R 2 = Non-Current Nitrogen (%) Current Nitrogen (%) Current Nitrogen vs canadien Non-Current Nitrogen Linear (Current Nitrogen vs Non-Current Nitrogen)

13 The spectral response of D. Fir and Salal stacks using a field spectrometer. The D. Fir spectra area is based on a footprint size of 19.2 centimeters. The salal spectra are based on a footprint size of 13.4 centimeters Salal Reflectance (%) D. Fir Wavelength (nm) canadien

14 Reflectance Calibration for Leaf Stack Target TRW Leaf Reflectance AFC Leaf Reflectance AFC Leaf Reflectance Converted Reflectance (%) Wavelength (nm) canadien

15 Salal Angle Sensitivity Salal Reflectance Compare 100% 90% 80% 70% Reflectance(%) 60% 50% 40% 30% 20% 10% 0% Wavelength (nm) Salal 90deg Reflectance ASD-TRW Salal Stack PFC Parking Lot 8/9/2001 2:58:51 PM canadien Salal -15deg Reflectance ASD-TRW Salal Stack PFC Parking Lot 8/9/2001 3:05:34 PM Salal 30deg Reflectance ASD-TRW Salal Stack PFC Parking Lot 8/9/2001 3:01:04 PM Salal -30deg Reflectance ASD-TRW Salal Stack PFC Parking Lot 8/9/2001 3:07:25 PM Salal 15deg Reflectance ASD-TRW Salal Stack PFC Parking Lot 8/9/2001 3:03:27 PM

16 IS AVIR Hy p erion AVIRIS / Hyperion Comparison GVWD - Index TM canadien

17 AVIRIS / Hyperion Comparison Con t Spectral Comparison of Plot 53 canadien

18 Hoquiam Hyperion Image Area of focus Plots with varying nitrogen. canadien

19 Hyperion Bad Pixel Correction Hyperion Level 1A Band 94 Before canadien After

20 Eigenvalue vs. Eigenvalue number (GVWD) 153 Bands Selected from Hyperion Level 1A Image give 15 good MNF bands canadien

21 EVEOSD activities Field work completed for 2001 at primary sites. Chemical organic and inorganic measurements completed; LAI and tree cores completed for GVWD. All data distributed to Co-investigators Copies of 8mm tapes distributed. FTP site of all data set up for Co-investigators. Includes all levels of clear EO-1 data for every site, plus LANDSAT, AVIRIS and relevant documents. Hyperion bad pixel correction software created Calibration measurements made at sites by CSA. Atmospheric correction experiments at MDA canadien

22 CHRIS The Project for On-Board Autonomy (PROBA) small satellite mission was launched by ESA on October 22, This satellite carries the Compact High Resolution Imaging Spectrometer (CHRIS). CHRIS will collect data over BOREAS SSA, Greater Victoria Watershed, Clayoquot Sound, and Sudbury. CHRIS provides a spectral coverage from 400 to 1050 nm with a minimum spectral sampling interval ranging between 1.25 and 11nm and a ground sampling interval of 25m at nadir. The PROBA satellite platform provides along- and across-track pointing. Number of spectral bands is 25m and 50m in an image area of 18.6 km x 18.6 km. CHRIS will be compared with EO-1 Hyperion. canadien

23 Next Steps Conduct field experiments in 2002 in support of AVIRIS and CASI overflights including canopy chemistry sampling from helicopters. Generate EO-1 simulated products from AVIRIS data. Conduct field and airborne experiments in 2002 in support of EO-1 and CHRIS missions. Complete analysis of 2001 EO-1 and AVIRIS data. Provide reports to NASA. Complete all reports and analyses of EO-1 and CHRIS data by March 2003 for CSA. canadien

24 Acknowledgements We acknowledge the support of NASA, Natural Resources, the Canadian Space Agency, TRW, MDA, and the University of Victoria and York University. We acknowledge the support of Weyerhaeuser and the Governments of BC, AB, and ON. We thank Martin Bergeron for his calibration contributions to this research. canadien