Generate flood map by return period from a given reporting point locations

[lucernae]

Based on these recommendation documents:

https://github.com/kartoza/fbf-project/blob/develop/Datasets/Reports/Visit%20Indo%20InaSAFE%203-8%20nov%20AHA.docx

Hassan provided some sample curve.

We should be able to generate specific flood map return period using this curve.
These is what I imagine the steps would be.

1. Select return period (we probably need to generate 10 or 20 year return period)
2. from the curve, find water elevation w.r.t to river bank. We will have height (m)
3. Choose specific a reporting point in the map. This point is the location that usually reports water elevation. At the moment, we probably just pick random points in the big river around Karawang.
4. Load DEM dataset for the area, or datasets that have elevation/height.
5. The point will have height, we use this as pivot for height 0 in river.
6. We should be able to get relative height. For example, if the flood elevation wrt riverbank is 3 m, all the area under 3 m relative height from this point can be marked as flood. Put the pixel with relative height information (this would be the relative flood height).

[lucernae]

Possible hydro reporting points:

http://tech4water.com/map

[lucernae]

Sample generated flood with the classification

https://github.com/kartoza/fbf-project/tree/develop/Datasets/Flood/flood_classes

[gubuntu]

shouldn't we just be choosing an historical flood map that matches the reported level, not generating a new one?

[AndreLester]

Steps needed to create a flood level:

1. Receive the point in the XY format and the return period in years.
2. Calculate the flood depth from the return period with the formula above. flood_depth = 1.1435*ln(return_period) - 0.1385
3. Determine the closest river line from the river centerline data.
4. Generate big cross-section lines on the river which preferably do not cross. (challenging)
5. Retrieve the DEM(SRTM) data for the cross-sections.
6. Determine the river bed depth of each cross-section. (Elevation where the cross-section and the river centerline cross)
7. Add an elevation to each cross-section. riverbed elevation + flood_depth
8. Generate an elevation grid from the cross-sections (Gridding)
9. Subtract the SRTM DEM from the cross-section elevation grid (values less than zero -> no value)
10. Vectorize the resulting grid

It might be easier for me to code in Python than to create a QGIS workflow. I don't know the impact of this on the deployment.

[lucernae]

@gubuntu

shouldn't we just be choosing an historical flood map that matches the reported level, not generating a new one?

We don't have any historical flood now. We generate this for demo purposes only to populate the database with corresponding flood map with matching return period.

[lucernae]

@AndreLester do whatever the quickest one, but it has to be reproducible (so we can generate for different locations/return period).
We can migrate it to QGIS workflow later.

[AndreLester]

Dependencies so far:

1. fiona
2. rtree
3. shapely