Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Anàlisi de les afectacions de la processionària amb UAV i LANDSAT

66 views

Published on

Presentació realitzada per Kaori Otsu, Magda Pla i Lluís Brotons (CREAF-CTFC-CSIC) a la jornada "Observació de la Terra i espai forestal, eines de diagnòstic" (08/11/2018)

Published in: Environment
  • Be the first to comment

  • Be the first to like this

Anàlisi de les afectacions de la processionària amb UAV i LANDSAT

  1. 1. ESTIMATING THE SEVERITY OF DEFOLIATION CAUSED BY PINE PROCESSIONARY MOTH USING LANDSAT AND UAV IMAGERY Kaori Otsu1, Magda Pla2, Lluís Brotons1,2,3 1 UAB, Centre for Ecological Research and Forestry Applications 2 InForest JRU 3 Spanish National Research Council J O R N A D A " O B S E R V A C I Ó D E L A T E R R A I E S P A I F O R E S T A L , E I N E S D E D I A G N Ò S T I C I C G C , 8 N O V E M B E R 2 0 1 8
  2. 2. Background 08/11/2018 ICGC Barcelona K. OTSU 2 Defoliator, pine processionary moth (PPM), since 1990s in the Mediterranean region Expansion of host and insect distribution Severely infested stands by PPM outbreaks Increased demands for forest monitoring
  3. 3. PPM Monitoring Annual field survey and mapping in Catalonia  From 2010 by Rural Agents (Generalitat de Catalunya) Data entry  Severity levels 1-4  Tree species  Elevation and orientation Outbreak in winter over 2015-2016 08/11/2018 ICGC Barcelona K. OTSU 3
  4. 4. Study Area (2015) 08/11/2018 ICGC Barcelona K. OTSU 4
  5. 5. Study Area (2016) Severely affected areas with level 4  6800 ha near Solsona, Catalonia  Elevation at 600-1100 m  Mediterranean continental climate  Pinus nigra, P. sylvestris Sketch mapping concerns  Qualitative classification  Coarse spatial resolution  Inclusion of non-forest stands 08/11/2018 ICGC Barcelona K. OTSU 5
  6. 6. Study Area (2017) 08/11/2018 ICGC Barcelona K. OTSU 6
  7. 7. Objectives  To quantify the severity of defoliation by the recent PPM outbreak with Landsat-based vegetation indices (VIs)  To calibrate the VIs with defoliation degrees interpreted by unmanned aerial vehicle (UAV) imagery 08/11/2018 ICGC Barcelona K. OTSU 7
  8. 8. Methodology Workflow 08/11/2018 ICGC Barcelona K. OTSU 8 Landsat 8 Imagery Pine Stands Extraction Change Detection (VIs) Predicted Variable (X) UAV Imagery Photo Interpretation Visual Defoliation (%) Observed Variable (Y)
  9. 9. Methodology Workflow 08/11/2018 ICGC Barcelona K. OTSU 9 X Predicted Vegetation Indices Y Observed Defoliation Degrees Regression Analysis 50
  10. 10. X = d(Vegetation Index) Multispectral Bands (OLI) Index Acronym Formula Middle Infrared Wavelengths MID b6 + b7 Moisture Stress Index MSI b6 / b5 Normalized Difference Moisture Index NDMI (b5 – b6) / (b5 + b6) Normalized Difference Vegetation Index NDVI (b5 – b4) / (b5 + b4) Normalized Burn Ratio NBR (b5 – b7) / (b5 + b7) Change detection in VI dVI VI (2015) - VI (2016) Landsat 8 Vegetation Indices 08/11/2018 ICGC Barcelona K. OTSU 10 (b4 = Red, b5 = Near Infrared, b6 = Shortwave Infrared 1, b7 = Shortwave Infrared 2 )
  11. 11. Y = UAV Images RGB camera  DJI Phantom 2 Vision FC200 UAV flight  Altitude 50-100 m  7 surveys in winter 2016 (post-outbreak)  Image processing for orthomosaic  Ground resolution 2.0-3.5 cm 08/11/2018 ICGC Barcelona K. OTSU 11
  12. 12. Y = UAV Images 3D model by PhotoScan 08/11/2018 ICGC Barcelona K. OTSU 12
  13. 13. Y = UAV Images 3D model by PhotoScan 08/11/2018 ICGC Barcelona K. OTSU 13
  14. 14. Y = UAV Images Orthomosaic Visual interpretation 08/11/2018 ICGC Barcelona K. OTSU 14 Severity Defoliation (%) Samples Nil 0 - 5 10 Low 10 - 30 23 Medium 35 - 65 8 High 70 - 100 9
  15. 15. Regression Analysis 08/11/2018 ICGC Barcelona K. OTSU 15
  16. 16. Regression Analysis Logistic Regression Models Threshold Classification 08/11/2018 ICGC Barcelona K. OTSU 16 Index Equation R2 (McFadden’s) dMID 𝑌 = 1 1 + 𝑒−(−3.1299111−0.0041928𝑋) 0.740 dMSI 𝑌 = 1 1 + 𝑒−(−3.3570352−0.0092755𝑋) 0.815 dNDMI 𝑌 = 1 1 + 𝑒−(−3.5552389+0.0014107𝑋) 0.749 dNDVI 𝑌 = 1 1 + 𝑒−(−3.509468+0.001767𝑋) 0.787 dNBR 𝑌 = 1 1 + 𝑒−(−3.6323329−0.0013874𝑋) 0.776 X Y = Defoliation (%) Low (10) Medium (35) High (70) dMID -222 -599 -949 dMSI -125 -295 -453 dNDMI 963 2081 3121 dNDVI 743 1636 2466 dNBR 1034 2172 3229 𝑌 = 1 1 + 𝑒 )−(𝑎+𝑏𝑋 𝑋 = 𝑙𝑛 𝑌 1 − 𝑌 − 𝑎 𝑏
  17. 17. Predicted Defoliation Map 08/11/2018 ICGC Barcelona K. OTSU 17
  18. 18. Classification Accuracy Confusion Matrix 08/11/2018 ICGC Barcelona K. OTSU 18 Class Predicted (Landsat 8) Nil Low Medium High Total Producer’s Accuracy Observed (UAV) Nil 9 1 0 0 10 0.90 Low 2 17 4 0 23 0.74 Medium 0 3 4 1 8 0.50 High 0 0 3 6 9 0.67 Total 11 21 11 7 50 User’s Accuracy 0.82 0.81 0.36 0.86 0.72
  19. 19. Discussions Robust VI for defoliation  Moisture Stress Index  Normalized Difference Vegetation Index Classification accuracy  Sample size  Non-parametric algorithms  Spectral bands of dVIs Data resolution trade-off = spatial + temporal + spectral  Ground & aerial sketch mapping  Spaceborne Landsat  Airborne UAV 08/11/2018 ICGC Barcelona K. OTSU 19
  20. 20. Conclusions and Future Study 08/11/2018 ICGC Barcelona K. OTSU 20 Additional UAV images in a new study area may increase the robustness of VI models. The UAV technology holds great potential for cost-effectively monitoring forest health. Combining satellite and UAV data may serve as a tool to provision pest distribution and forest production. Validation Ground Truth Ecosystem Service
  21. 21. Publication Otsu, K.; Pla, M.; Vayreda, J.; Brotons, L. Calibrating the Severity of Forest Defoliation by Pine Processionary Moth with Landsat and UAV Imagery. Sensors 2018, 18, 3278. https://www.mdpi.com/1424-8220/18/10/3278/htm 08/11/2018 ICGC Barcelona K. OTSU 21
  22. 22. supported by http://www.alertaforestal.com/alertas/procesionaria k.otsu@creaf.uab.cat
  23. 23. PPM Life Cycle (Thaumetopoea pityocampa) 08/11/2018 ICGC Barcelona K. OTSU 23 Fall-Winter • larval feeding Spring • pupation in soil Summer • adult flight

×