Requirements

Data

• High resolution Planet imagery
• Annual land cover and land use maps from MapBiomas Venezuela using Landsat, Sentinel -2, VRSS-1/2, ZY -1/2/3, GF-1, SPOT-5 images

Software

• IDRISI software
• QGIS
• Google Earth Engine (GEE)

Outline steps for a solution

1. Leveraging existing land cover maps for mapping land cover change (optional step) - a quick change assessment 2020-2025 (in development)
2. High resolution land cover land use map of 2020, and 2025 (in development)
3. Post-classification Change Assessment (2020-2025) (in development)

Steps to a solution

Step 1. Leveraging existing land cover maps for mapping land cover change (in development)

• Mapbiomas Venezuela
• Living Atlas Sentinel 2 maps
• Algorithm to combine the multiannual land cover maps into a land cover change map (Python, QGIS)
• Each team member performs step 1. for one basin
• Result: map indicating change hotspots/clusters of change and time of change

Step 2. High resolution land cover land use map of 2020, and 2025 (in development)

• For every basin and year mosaic the images (into a single radiometrically calibrated raster) *potential difficulties: complex radiometric calibration due viewing angle and illumination differences*
• Decision on land cover classes to map
• Select areas of interest for the classes
• Incorporation of digital elevation model (DEM) for classification
• Supervised classification (random forest and support vector machines)
• Accuracy assessment based on areas of interest
• Result: one land cover class map per year per basin in high resolution with desired classes

Step 3. Post-classification Change Assessment (2020-2025) (in development)

• Subtract one map over the other
• Result: map indicating change hotspots/clusters of change and time of change in high resolution and with our classes

Solution follow-on: Impact projection

• QGIS or IDRISI software for predicting land use change for the next 5-10 years
• Hydrological simulation of soil erosion and sediment discharge into reservoir
  • Hydrological simulation models: Soil and water assessment tool (SWAT), IBER, HEC-RAS 2D
• Considering driving factors of land cover change (agricultural land expansion, deforestation, mining)

Solution – 2. Steps to the solution

1. Radiometric calibration and mosaicking of high-res images: per basin and per year (in development).
2. Calculation of spectral indices (NDVI, water, soil, etc.) for each mosaic (in development).
3. Layer stack of mosaics, so we create images with bands representing spectral index for 1 year (in development).
4. Algorithm to spot per-pixel relatively large increments of spectral indices, indicative of change. Change-threshold to be fine-tuned (in development).
5. Visual analysis of changes and classification (in development).

Result: Raster highlighting potential land cover changes. These will be visually assessed and potentially, classified (supervised classification).