Newsletter Sep 2018
Birdhouse at FOSS4G 2018
For the first time an international FOSS4G conference took place in an African country. FOSS4G was hosted in Dar-Es-Salaam in Tanzania, and was a great success. OSGeo and the Humanitarian OpenStreetMap Team (HOT) organised the conference. The importance of data for sustainable development was the primary issue reinforced with the conference motto ‘leaving no one behind’. Several projects using the birdhouse framework were presented. This presentation showcased birdhouse’s server-side solutions for spatial data infrastructures supporting sustainable development.
Dar-Es-Salaam Release
We have a new release of the Birdhouse components called Dar-es-Salaam. See the release notes.
Congratulation to the first African regional data cubes
The birdhouse team sends its congratulations to the UN Partnerships for the launch of the first African regional data cubes. We see it as a great contribution to the targets of the Agenda 2030' sustainability goals. While birdhouse grew out of the climate community and has focused so far on climate model data, the regional data cubes are designed to process Earth Observation (EO) data. Despite these differences, both projects share similar principles and technologies, and we believe there are potential synergies between them. We’re hoping birdhouse can grow to interact with data cubes and help create climate services mixing climate model projections with EO data.
Xclim: a library of climate indices using xarray
Climate services are often build from derived climate variables. That is, we are not working with the raw model output (think 3-hourly total precipitation), but rather with climate indices, such as the maximum annual daily precipitation. Climate analysts either write their own code for these climate indices, or rely on existing libraries such as ICCLIM or CLIMDEX. These libraries are often written in C or Fortran for performance, with wrappers that facilitate their use with scripted languages such as R or Python. Parallelism is usually provided by MPI (Message Passing Interface), which is not trivial to get to work inside a Web Processing Service.
To enable developer-friendly forms of parallelism (dask) and simplify the implementation of new indices, we’ve started to work on a library called xclim where indices are defined using the xarray library. Beyond the standard indices found in ICCLIM and CLIMDEX, we’ll include indices relevant to other disciplines such as hydrology, forestry, sea ice, agriculture, etc. Although this may seem ambitious, xarray makes it so easy to define and implement a broad set of indices. We are planing to release an alpha version this Fall (contributions welcome).
In parallel, we are working on xarray support in OCGIS. This will allow us to create a Web Processing Service dedicated to climate indices that gracefully handles spatio-temporal subsets of CF-compliant netCDF files.
