1 / 10

Land cover mapping at global scale: some lessons learnt from the GLC 2000 project

Land cover mapping at global scale: some lessons learnt from the GLC 2000 project. E. Bartholomé JRC-Ispra. Some observations on the following issues. Properties of input data Pre-processing Classification procedures and cover type identification The way forward. Input data.

glynis
Download Presentation

Land cover mapping at global scale: some lessons learnt from the GLC 2000 project

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Land cover mapping at global scale: some lessons learnt from the GLC 2000 project E. Bartholomé JRC-Ispra

  2. Some observations on the following issues • Properties of input data • Pre-processing • Classification procedures and cover type identification • The way forward

  3. Input data • Number of channels (more channels would be better: cloud mask, LC discrimination) • Radiometry (OK) • Atmospheric correction (Cloud mask, aerosols) • Geometry (Sub-pixel accuracy: the real difference) • Data format (Global coverage in one single image)

  4. Pre-processing • How to build syntheses • 3rd min albedo (Cabral & al) • Average (Vancutsem & al) • BRDF correction (Champeaux & al) • Cloud mask (Viaux & al., Cherlet & al.) • How to clean temporal profiles • Low vegetation coverage (Defourny & al.) • Removal of atmospheric noise (Bartholomé & al) • Good strategic choice not to wait for improved methods, but to directly start map production (last data received March 2001, this workshop March 2003) • But results obtained through labour intensive techniques difficulty to repeat regularly

  5. Classification • Most people have used from-the-shelf classification software • Classical approach: 1) divide space according to spectral properties, 2) label each area with reference information • Ideally, the structural properties of each land cover class should have to be identified form space capacity to measure a number of parameters: • Vegetation height for each strata • Vegetation cover for each strata (forest cover ¹ fcover) • Vegetation type (woody/grassy/…, broadleaf / needle leaf) • seasonality

  6. Classification • At least two cases of ad hoc classification algorithm • Small water bodies (VGT only, semi arid regions only) • Urban areas (VGT + DMSP)

  7. Next steps – medium term • Combination of existing LC map with new data to identify possible changes • Multi-sensor database to complete the spectrum of measurements • Intensification on priority land cover classes to respond to specific needs (i.a. forest, agriculture, protected areas,…) – use of medium resolution

  8. Next steps – medium term • We should consider LC mapping & monitoring at global scale as a component of environmental monitoring: • Technical commonality between several applications (e. g.Forest change & FCCC/KP, Agriculture & food security, protected areas & CBD, wetlands & Ramsar conv.) • Incorporate indicators of disturbances possibly leading to land cover conversion / degradation (e. g. fire, burn scars, drought index, surface water, etc…)

  9. Next steps – long run • Need to incorporate specifications related to land cover monitoring into future space missions • Choice of instruments / wavebands / repeat cycle, etc. • Quality of output data (e. g. geometry) • Need for research to develop new approaches to retrieve significant LC-related parameters (e. g. BRDF parameters as indicators of vegetation structure)

  10. The end !

More Related