About hight temporal resolution
- 1. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion About High Temporal Resolution Jordi Inglada CNES/CESBIO 09-11-2010 Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 1 / 36
- 2. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Outline Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 2 / 36
- 3. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion New sensors Venus Sentinel (1,2) LDCM New applications . . . . . . which require to closely monitor the temporal trajectory of the characteristics of land surfaces. real time classification evolving nomenclatures Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 3 / 36
- 4. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Challenges Global coverage every few days Expectations for land cover change monitoring Real-time: update the land-cover maps for every new acquisition Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 4 / 36
- 5. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Approaches It's not about methods but about needs/applications spatio-temporal trajectories of clusters in a kernelized feature space are cool . . . but a hard threshold on NDVI can sometimes work Many scientists have developed models for the physical processes Some are easy to use; some are complex Some can be spatialized; some can't Many are Open Source (more on this later) Expert knowledge i.e. agricultural practices Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 5 / 36
- 6. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Outline Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 6 / 36
- 7. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Essential Climate Variables For climate change assessment, mitigation and adaptation: River discharge, Water use, Groundwater, Lakes, Snow cover, Glaciers and ice caps, Permafrost, Albedo, Land cover (including vegetation type), Fraction of absorbed photosynthetically active radiation (FAPAR), Leaf area index (LAI), Above-ground biomass, Fire disturbance Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 7 / 36
- 8. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Models Areas of interest: hydrology, agriculture, forestry, Media: Aerial, terrestrial, aquatic, mixed How to find the good balance complexity, number of input parameters and variables, computational cost Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 8 / 36
- 9. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Models They describe the physical reality Their assumptions/simplifications are clear Naturally use/need ancillary data (meteo, ground measures) They can be multi-sensor or better . . . . . . Sensor Agnostic benefit from the synergy between sensors increase temporal sampling! Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 9 / 36
- 10. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Open source models - some examples Prospect: optical model for estimating leaf-level reflectance and transmittance Sail: canopy reflectance model Daisy: mechanistic simulation model of the physical and biological processes in an agricultural field 6s: a basic RT code used for calculation of look-up tables in the MODIS atmospheric correction algorithm Arts: radiative transfer model for the millimeter and sub-millimeter spectral range. etc. have a look at ecobas.org Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 10 / 36
- 11. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Outline Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 11 / 36
- 12. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Purpose Which is the best sensor to recognize these: 60 40 20 0 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 sol nu sec végétation eau visible proche infrarouge moyen infrarouge Longueur d'onde (µm) Réflectance(%) Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 12 / 36
- 13. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Purpose Or these Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 13 / 36
- 14. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Principle Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 14 / 36
- 15. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Architecture Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 15 / 36
- 16. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Results 0 0.2 0.4 0.6 0.8 1 Vegetation Soils M an-m ade M inerals Accuracy Spot 5 Quickbird Pleiades Landsat TM Ikonos Formosat Meris Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 16 / 36
- 17. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Results 0 0.2 0.4 0.6 0.8 1 R oad C oncretesC onstructions R oof Igneous M etam orphic Sedim entary Alfisol Aridisol Entisol InceptisolM ollisol Accuracy Spot 5 Quickbird Pleiades Landsat TM Ikonos Formosat Meris Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 17 / 36
- 18. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Results 0 0.2 0.4 0.6 0.8 1 brow nc om ps hr f dkg rayc om ps hr f ltg rayc om ps hr f orangg rayc om ps hr f grayg ravelr fredg ravelr fbrow nm etalr f ltg raym etalr fltg raya sphaltlr f redtiler f brow ng raytiler f tarr f w oods hingler f greenv egnpv bares oil openw atersw imp oollta sphaltr ddka sphaltr dconcreter dgravelr d parkinglotrailroadtracktennisc ourtreds porttartan Accuracy Spot 5 Quickbird Pleiades Landsat TM Ikonos Formosat Meris Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 18 / 36
- 19. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Results 0 0.2 0.4 0.6 0.8 1 M an-m ade Igneous M etam orphic Sedim entary Soils Accuracy Spot 5 Pleiades Pleiades+MIR Spot5-MIR Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 19 / 36
- 20. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion But we said HTR . . . How to simulate a multi-t mission? Venus, Sentinel-2 Realistic temporal evolutions Use existing image time series Formosat-2 8 m., 4 bands (B,V,R,NIR), 3 days Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 20 / 36
- 21. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Spectral bands 500 1000 1500 2000 wavelength 0.0 0.2 0.4 0.6 0.8 1.0 Formosat-2 Relative Spectral Responses 500 1000 1500 2000 wavelength 0.0 0.2 0.4 0.6 0.8 1.0 Venus 500 1000 1500 2000 wavelength 0.0 0.2 0.4 0.6 0.8 1.0 Sentinel-2 Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 21 / 36
- 22. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Example of series March 14, 2006 Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 22 / 36
- 23. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Example of series July 17, 2006 Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 23 / 36
- 24. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Example of series November 2, 2006 Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 24 / 36
- 25. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Available data 49 images in 2006 Orthorectification OK Radiometric corrections OK TOC and aerosol corrections Cloud screening Land-cover map available Leaf pigments data base for several vegetation types (LOPEX'93) Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 25 / 36
- 26. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Simulator architecture Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 26 / 36
- 27. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Example of application Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 27 / 36
- 28. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Outline Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 28 / 36
- 29. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Soil work Main goal: improve real-time crop classification; soil work can give hints on the type of crop Soil map: is also interesting in itself as a product Inter-crop Stubble disking Deep ploughing Harrowing Sowing preparation Emergence Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 29 / 36
- 30. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Approach Radiometry only: only the reflectances and combinations of them (indexes) are used; no texture, statistics, nor object-based features. Index Formula NDVI NIR−R NIR+R Color R−B R Brightness √ G2 + R2 + NIR2 Shape 2R−G−B G−B Redness R−V R+V Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 30 / 36
- 31. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Approach Statistical analysis: the temporal evolution of the reflectances and the indexes (globally and per class) are studied. 2 kinds of analysis: Identification of the soil state: classification Identification of the transitions between states: change detection SVM classification: both used as separability measure and as classification tool Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 31 / 36
- 32. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Results for change detection Transition D→H H→SP H→E SP→E Accuracy (%) 97.0 88.74 87.91 96.76 The number of transitions is very low for some cases (between 12 and 50 plots; or between 1k and 10k pixels) Many transitions between states can't be detected accurately However, some changes are well detected (about 90% and more) Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 32 / 36
- 33. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Outline Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 33 / 36
- 34. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion What we've got Source code available for many simulators Ongoing work for Prospect, Sail & Daisy integration new hyper/multi- spectral/temporal algorithm integration Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 34 / 36
- 35. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion What we need Engineering - Development Improve image simulation: MTF, realistic landscapes Hide physical models under common interfaces Research Learn to select the best model set for a given problem Incorporate domain expert knowledge Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 35 / 36
- 36. Introduction Physical models Multi-temporal simulator Land-cover change Conclusion Creative Commons Attribution-ShareAlike 3.0 Unported License Jordi Inglada (CNES/CESBIO) About High Temporal Resolution 09-11-2010 36 / 36