AMSIMP
An open-source solution which leverages machine learning to improve numerical weather prediction
AMSIMP Developer Blog is LIVE!
get the latest news on the developments with AMSIMP

Powerful Neural Networks
Fast and accurate, AMSIMP’s neural networks provide high quality weather forecasts and predictions.

Operational Model
AMSIMP offers a pretrained operational model architecture. It is trained on a dataset from the past decade, ranging from the year 2009 to the year 2016.

Near Real-Time NWP
AMSIMP offers near real-time numerical weather prediction through initialisation conditions provided by the Global Data Assimilation System.

Performance
The core of AMSIMP is well-optimized Python code. A performance increase of 6.18 times can be expected in comparison against a physics-based model of a similar resolution.

Easy to use
AMSIMP’s high level and intuitive syntax makes it accessible for programmers and atmospheric scientists of any experience level.

Open source
Distributed under the GNU General Public License v3.0, AMSIMP is developed publicly on GitHub.

Demo
A quick look at the software's high level and intuitive syntax

PROGNOSTIC VARIABLES

2 metre temperature
Example 2 metre temperature prediction from the AMSIMP GFM.
The 2 metre temperature field gives forecasters a general idea of what the temperature is now and what it will be in the future at the Earth’s surface.
850 hPa air temperature
Example 850 hPa air temperature prediction from the AMSIMP GFM.
The second air temperature parameter is at a pressure surface of 850 hPa. The 850 hPa pressure surface is usually above the planetary boundary layer and therefore not affected by diurnal variations but provides information about broader temperature trends.
500 hPa geopotential
Example 500 hPa air temperature prediction from the AMSIMP GFM.
Geopotential at 500 hPa pressure is a commonly used variable that encodes the synoptic-scale pressure distribution. It is the standard verification variable for most medium-range numerical weather prediction models.