services

services Anton Van Rompaey 1 , Veerle Vanacker 2 , Vincent Balthazar - PowerPoint PPT Presentation

Forest transition in mountain areas: topographic correction, large scale mapping and modeling of ecosystem services Anton Van Rompaey 1 , Veerle Vanacker 2 , Vincent Balthazar 2 , Eric Lambin 2 , Jaclyn Hall 2 , Patrick Hostert 3 , Patrick


  1. Forest transition in mountain area’s: topographic correction, large scale mapping and modeling of ecosystem services Anton Van Rompaey 1 , Veerle Vanacker 2 , Vincent Balthazar 2 , Eric Lambin 2 , Jaclyn Hall 2 , Patrick Hostert 3 , Patrick Griffiths 3 , Steven Vanonckelen 1 , Armando Molina 1 , Derek Bruggeman 2 , Marie Guns 2 1: Geography Research Group, Dep. Earth and Environmental Sciences, KULeuven 2: Department of Geography, Earth and Life Institute , UCLouvain 3: Geographisches Institut, Humbolt-Universität zu Berlin Belgian Earth Observation Day 2012 September 5 – 2012 - Bruges

  2. W HAT IS THIS RESEARCH PROJECT ABOUT ?  Forest transition  Mountain areas  Topographic correction  Large scale mapping  Ecosystem service assessment

  3. F OREST TRANSITION  Transition from a phase of net deforestation to a phase of net afforestation?  For a region: pathway through time driven by  Economic development -> off-farm jobs -> land abandonment (+ forest plantations)  Land degradation  lowering of yields -> land abandonment (+ forest plantations) Regional patterns of FT Mather, 1995: FT in France

  4. M OUNTAIN AREAS  Forest transitions typically occur in mountain areas Drome (SE-France): 1985

  5. D ETECTION OF FOREST TRANSITIONS  Remote sensing seems privileged tool but many difficulties  Transition are of subtle : gradual forest degradation in stead of clear cuts  Typically patchy landscape with various phases of degradation, natural regrowth , plantations, …  Understanding of mechanisms only possible with large scale mapping  Mountain environment: atmospheric conditions, illumination effects  Impact on ecosystem services is important (= both cause and consequence of FT)

  6. R ESEARCH QUESTIONS  To what extent is it possible to detect forest transitions and their effect on ecosystem services?  Atmospheric and topographic correction procedures exist but was is their added value for FT-mapping  Most complex procedure not always appropriate because of difficulty to automate  Development of automated preprocessing chains and large scale mapping procedures  Monitoring ecosystem services in FT- landscapes  3 study sites  Romanian Carpathians  Buthan Himalayas  Ecuadorian Andes

  7. A DDED VALUE OF TOPOGRAPHIC AND ATMOSPHERIC CORRECTIONS ? TOPOGRAPHIC CORRECTIONS No Corr Band Cosine C-corr Minnaert ratioing corr corr No Corr ATMOS- Dark object subtraction PHERIC CORREC Physical TIONS transfer function

  8. SOME FORMULAS • DN values to at-satellite radiances • Calculation of corrected path radiance L t, λ : L s,λ is the uncorrected radiance value of the image and L p represents the minimum radiance value of the image, calculated as the 1 th percentile. T r,λ is the Rayleigh scattering transmittance function, including the sea-level atmospheric pressure ( P 0 ; in mbar), the ambient atmospheric pressure ( P ; in mbar) and the band wavelength ( λ ). M is the relative air mass and θ s is the solar zenith angle (in degrees). T w,λ is the water-vapor transmittance function, calculated with the precipitable water vapor ( W ; in cm), relative air mass ( M ) and water-vapor absorption coefficients ( a w ).

  9. S OME FORMULAS • Conversion to observed reflectance on an inclined terrain ρ t, λ • Conversion to normalized reflectance of a horizontal surface ρ H, λ θ s is the solar zenith angle and β is the incident solar angle, cos β = cos θ s cos θ n + sin θ s sin θ n cos ( ϕ t – ϕ a ), θ n is the slope angle of the terrain and k ,λ is the slope of the regression between x = log(cos θ n cos β ) and y = log( ρ T,λ cos θ n ). Parameter C λ is the quotient of intercept ( b λ ) and slope ( m λ ) of the regression line between x and y , the h -factor represents a topographic parameter derived from the SRTM ( h = 1- θ n / π ) and the h 0 -factor an empirical parameter derived from the regression line between reflectance and cos β ( h 0 = ( π +2 θ s )/2 π ).

  10. APPLICATION AND EVALUATION OF EFFICIENCY

  11. S TATISTICAL ANALYSIS  Efficient methods reduce the reflectance values between shaded and illuminated slope segments? Band 4 DOSwithout TC no AC or TC (b) (a) 50 50 y = 14.639x + 16.12 y = 12.6x + 19.4 P = 1.78E-37 45 P = 5.08E-42 45 R² = 0.0813 R² = 0.0938 40 40 Reflectance (%) 35 Reflectance (%) 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 TF with cosine DOS with band ratio (d) (c) 45 5 50 y = -2.7x + 19.6 y = -2.7x + 19.6 y = -8.7x + 40.7 P = 2.67E-11 P = 0.001 P = 0.001 4.5 40 45 ( a) no AC or TC; R² = 0.0055 R² = 0.0055 R² = 0.0282 4 40 35 (b) DOS without TC; Reflectance (%) Reflectance (%) 3.5 35 30 3 30 (c) DOS with band ratio; 25 2.5 25 20 (d) TF with cosine; 2 20 15 15 1.5 (e) TF with PBM; 10 1 10 5 (f) TF with PBC. 0.5 5 0 0 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 TF with PBM TF with PBC 50 50 (f) (e) y = 0.7x + 17.7 y = 1.3x + 37.4 P = 0.483 P = 0.465 45 45 R² = 0.0003 R² = 0.0003 40 40 35 35 Reflectance (%) Reflectance (%) 30 30 25 25 20 20 15 15 10 10 5 5 0 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 cos β cos β

  12. STATISTICAL ANALYSIS  Efficient preprocessing results in better land cover maps? (a) (b) (c) (a) no AC or TC; (b) TF with cosine; (c) TF with PBC

  13. STATISTICAL ANALYSIS  Efficient preprocessing results in better land cover maps? 0.35 90 NoTC_DOS 0.3 85 NoTC_TF 80 0.25 75 Bandratio_NoAC 70 0.2 Bandratio_DOS 65 0.15 Bandratio_TF 60 55 0.1 Cosine_NoAC 50 Cosine_DOS 0.05 45 40 Cosine_TF 0 No TC Band ratio Cosine PBC PBM No TC Band ratio Cosine PBC PBM No TC Band ratio Cosine PBC PBM BS BL CF MX GRASS WT PBC_NoAC -0.05 PBC_NoAC -0.1 PBC_TF No AC DOS TF -0.15

  14. AUTOMATION POTENTIAL OF PREPROCESSING TECHNIQUES

  15. L ARGE SCALE MAPPING  Improved data policy, image products and IT makes image compositing an appealing option for moderate resolution data analysis  Key features:  pixel based perspective!  “increase” observation frequency  Objectives: Develop algorithm to produce regional, cloud free, 1. radiometrically consistent image datasets 2. Use image composites for pan Carpathian forest change mapping

  16. Image compositing principle Use:  All images..  Multiple years  Multiple seasons Decision rule set: select best observation per pixel  Create best observation composite image  Make use of all unclouded observations of a pixel 16

  17. 17 17

  18. Output examples Flag image (RGB = Composite image Score image PathRow/Year/DOY) (RGB=4/5/3) (low=blue, high = red) Mean image (RGB=4/5/3) STDV image (RGB=7/5/4) Clear observation count (low=blue, high=red) 18 18

  19. 19

  20. Disturbance rate [%] per country  Disturbed forest area [ha]  20

  21. E COSYSTEM SERVICES ASSESSMENT AT REGIONAL SCALE The significance of forest transitions in creating more sustainable societies depends on the effects of these transitions on the environmental services that forests provide” ( Rudel et al., 2005)  Natural hazard regulation (mass movements)  Water and erosion regulation (agricultural production, water quality&quantity)  Carbon storage and biodiversity (REDD/REDD+)

  22. STUDY SITE: RIO PANGOR ( ECUADOR )

  23. IMPACT OF LAND COVER CHANGE?

  24. IMPACT ANALYSIS  LINK WITH CHANGES IN LANDSLIDE FREQUENCY?  LINK WITH CHANGES IN DISCHARGE REGIMES  LINK WITH CHANGES IN CARBON SEQUESTRATION AND BIODIVERSITY?

  25.  Objective of creating landslide inventories for different years based on HR imagery  Selection and validation of the method for a recent time-period (covered by VHR data)  Application of shape and spectral characteristics back in time with HR imagery  Combination with morphometric characteristics and use of semi-extraction techniques  Statistical relation between landslide occurrences and land-use changes

  26. WP 4: E COSYSTEM S ERVICES ASSESSMENT 4. Ecosystem services assessment 4.1. Natural hazard: 4.1.2. Use of HR data (VB) LANDSAT ETM+: RGB 453 (30 m) LANDSAT ETM+: RGB 453 (15 m PAN-fusion) LANDSAT ETM+ (P-S): RGB (PCA1, PCA2, PCA3) ASTER VNIR + SWIR: RGB (PCA1,PCA2, PCA3)

  27. Long-term trends corrected for intra and interannual varation Monthly rainfall data Monthly streamflow data Land Use Monthly rainfall (mm)

  28. C ARBON AND BIODIVERSITY  a. Northern Costa Rica Tempisque watershed  5414 km 2  sea level - 1964 m  Sub-landscape 25 km 2   b. Northern Vietnam 91,600 km 2  ~100 - 3100 m  Sub-landscape 400 km 2  ¯ a. b.  c. Valdivia Chile 11,775 km 2  Sea level - 1222 m  Sub-landscape 25 km 2   d. Highland Ecuador Pangor watershed  282 km 2  ranging from 1431 to 4333 m  Sub-landscape 2.25 km 2  c. d.

  29. P OTENTIAL TO SUPPORT NATIVE FLORAL BIODIVERSITY

  30. IMPACT OF FOREST TRANSITION ON BIODIVERSITY • a. Northern Costa Rica – Natural forest regeneration • b. Northern Vietnam – Natural forest regeneration • c. Valdivia Chile – Natural forest to plantation • d. Highland Ecuador – Ag. to plantation – Páramo to plantation

Recommend


More recommend