Introduction

Monkey Wrench provides tools to run pre-trained neural network (NN) models for precipitation intensity using satellite data, based on the Chalmers/CSU Integrated Multi-Satellite Precipitation Retrieval Program, also referred to as CHIMP for short. CHIMP is a general framework for training, evaluating, and running NN-based retrievals of physical properties from remote-sensing observations. However, the focus of Monkey Wrench has so far been NN-based retrievals of precipitation intensity using satellite data, and CHIMP will be used here synonymously with such already pre-trained models.

In particular, Monkey Wrench includes utilities to fetch and process CHIMP input files, analyze and visualize its output, and much more! It is an attempt to avoid reinventing the wheel by consolidating all Python scripts and code that we use at SMHI alongside CHIMP.

Example use cases of Monkey Wrench:

  • ☑ Querying and fetching SEVIRI data from the EUMETSAT API via either single queries or batches

  • ☑ Resampling fetched files from the API

  • ☑ Using the provided CHIMP extensions to allow for reading files which are not natively supported by CHIMP

  • ☑ Performing queries on local files, datetime objects, and conversion between filenames and datetime stamps

  • ☑ Running CHIMP for different groups of input files in a queue

  • ☑ Running tasks from YAML files, alleviating the need for coding

  • ☐ Performing post-processing of CHIMP output as well as some visualization utilities

Note

Monkey Wrench has been developed at the Swedish Meteorological and Hydrological Institute (SMHI) as part of the Skyfall project, partly funded by the Swedish Civil Contingencies Agency (MSB).

Note

Monkey Wrench uses the eumdac package for API calls. This allows for customized downloading of files from EUMETSAT Data Store tailored to our needs. However, our wrapper does not utilize or even expose all the functionalities of the eumdac, at least not yet!

Warning

Monkey Wrench is a work in progress and under active development. As a result, the API and the scheme of task (YAML) files might and will change!

License, Disclaimer of Warranty, and Limitation of Liability

This program is distributed under the GNU General Public License Version 3. You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/gpl-3.0.html.