Michael Saha

Github Page

I have written a few python projects in support of my research:

---

blahb

A tool for working with spatial coordinates on regular grids.

This library provides:

• IndexSet - A numba.jitclass representing a set of N-dimensional pixels
• Easy integration with numpy and numba. Many methods can be called from within nopython mode.
• Performant set operations on IndexSets
  • union
  • intersection
  • asymmetric_difference
  • symmetric_difference
• Connected component labeling.
• Easy parallelization gains on common operations (just set n_workers)
• Storing data associated with each pixel and propagating data through set operations.

---

pyrseid

Helps with downloading and organizing remote sensing data.

Downloading MODIS data on the exceptional Alaska 2015 fire season is as simple as:

import pyrseid as ps
import pyrseid.platforms.modis as modis

# Collection 6 Burned Area Product
burned_area_product = ps.Product('MCD45A1.006')

# Granule Range (Alaska)
tiles = list(ps.tile.TileRange(modis.Tile(h=9, v=2), modis.Tile(h=12, v=3)))

# Times (inclusive)
dates = ps.Range(datetime.datetime(2015, 4, 1), datetime.datetime(2015, 10, 1))

# Top level MODIS directory, specific products and dates will be organized within
modis_folder = "/where/to/download/MODIS/"

for file in modis.available_files(burned_area_product, dates=dates, tiles=tiles):
    file.grab(top_dir=modis_folder)

# Then query local files for reading in/processing
files = modis.have_files(modis_folder, burned_area_product, dates, tiles)

files[0].path # Do something with a local file...

---

geotree

Defines a single class, GeoTree, that allows easy and efficient nearest-neighbor resampling using geographic coordinates (latitude/longitude) using kdtrees. This is done by first converting lat/lon to Euclidean points on a geoid, then resampling in 3d space.

---

This website

The source css/js/html for this website is available on my github page

Link to Drossel & Schwabl forest fire simulation code: fire.js