Landslide EarlyDetection

Machine Learning Research Project: Landslide Early Detection from Satellite Data

Data

Our dataset can be found at the following link

products = ReMasFrame.get_products()

Nombre	Categoría	ID	¿Funcionando?	Link a Notebook	Resolución	Por Hacer
goes17:fulldisk:v1	weather	products['weather']['goes']	NO	Notebook Link	NA	NA
GSOD Daily Interpolation Weather Product	weather	products['weather']['gsod']	SI	Notebook Link	0.1	NA
CHIRPS Daily Precipitation Weather	weather	products['weather']['chirps']	SI	Notebook Link	0.05	create stack
NCEP CFS-v2 Derived Daily Weather Product	weather	products['weather']['cfs']	SI	Notebook Link	0.20	TODO
smap:SMPL3SM_E	soil_moist	products['soil_moist']['smap']	SI	Notebook Link	0.1	NA
aster:gdem3:v0	elevation	products['eleveation']['asger']	SI	Notebook Link	0.001	NA
GPW_Population_Density_V4_0¹	population	products['population']['population']	SI	Notebook Link	NA	NA
VisMet Data	weather	products['weather']['vismet']	NO	Notebook Link	NA	to product

¹ : Actual id is `d15c019579fa0985f7006094bba7c7288f830e1f:GPW_Population_Density_V4_0`

Recordatorios:

products = ReMasFrame.get_products()

buffer_size = 0.1 $\iff$ 0.1 deg $\approx$ 10km $\implies$ box de 10kmx10km aprox.

Google Compute Engine

Instance

If you have been given access to the GCP project projectx-uch, you can find the compute instances.

Bucket

url : https://console.cloud.google.com/storage/browser/data-projectx/processed_data

!pip install fsspec
!pip install gcsfs

import fsspec

fs = fsspec.get_mapper('gcs://data-projectx/')

# show files
print(list(fs))

# access files
with fs.open(path, mode='rb', cache_type='readahead') as f:
  use_for_something(f)

Tutorial access see notebook example (TODO)
ffsspec docs

DescartesLab

url : Workbench Gives access to satellite data and Workbench (4vCPUs, 13GB~RAM, optional GPU)

Credentials

If you need to access use your given email (usually an academic one)

Competencia-de-Climate-Change/Landslide_EarlyDetection