Clouds, precipitation and wind have a significant and direct impact on all levels of traffic. Reuniwatt and DWD, the German Meteorological Service, are cooperating in an innovative project, ICamCloudOps (Intelligent Camera Cloud Operators for NWP), to research and promote Artificial Intelligence methods to improve wind and cloud analyses for numerical weather prediction (NWP) models and thus forecasts of cloudiness, precipitation, and wind.
AI methods for cloud observations
The atmospheric analysis of clouds using new techniques of artificial intelligence is an important and active topic in international research. The German Meteorological Service develops and operates numerical weather prediction models. With its Sky InSight™ camera, Reuniwatt is a leader in the development and operation of all-sky infrared imagers. The German Meteorological Service, based in Offenbach, and Reuniwatt have been cooperating since 2015 in the field of data assimilation. In their latest project, ICamCloudOps, they’re focusing on the development of Artificial Intelligence methods to improve the methods of wind and cloud analysis. The project is funded by the “modernity fund” (mFUND) of the German Federal Ministry of Transport and Digital Infrastructure (BMVI). The development of intelligent forward operators for data assimilation is an important step for the further scientific development of atmospheric and climate science making use of numerical weather prediction models. The development of intelligent forward operators for sky camera observations will significantly increase in realism in the analysis of the cloudiness and wind situation, and is therefore expected to improve the forecast quality for a wide range of users of weather forecasts in the area of mobility (air, rail, water and road) and renewable energy (PV and wind).
Taking up innovative work at Reuniwatt
In order to predict socially critical weather phenomena, especially wind and cloud developments in the lower troposphere, a significantly improved initialization of the forecast models is necessary. The use of new types of observations based on techniques of artificial intelligence in the form of cameras in the infrared and visible range breaks new ground: The project takes up innovative AI preparatory work and integrates it at a crucial point in the initialization of the predictions. The use of deep learning methods, which break down a complex relationship between input and output variables into a sequence of associative neuronal maps, will ultimately result in a fast radiation simulation in the visible and infrared spectrum to determine atmospheric state. With the rapid development of ground-based camera images in the form of IR sky cameras / VIS cameras and webcams, a wealth of such new techniques and observations are available. This is where the the proficiency of Reuniwatt lies:
The integration of these observations into the prediction systems of the numerical weather forecast should take place through the development of intelligent camera observation operators.
Reuniwatt is known for its contiuous efforts in the development of better forecasting techniques. With its Sky InSight™ camera, Reuniwatt is a leader in the development and evaluation of all-sky infrared cameras, and contributes with its expertise to the use of artificial intelligence for camera images. Our spokesperson for the project is Dr. Frederik Kurzrock, an expert in the assimilation of cloud-influenced observations into high-resolution numerical models at Reuniwatt.
The intelligent cloud forward operators developed in the ICamCloudOps project will be used to develop new tools for creating links between the “big data” data sets of numerical weather forecast and camera observations in the infrared and visible wavelength range.
Creating new methods for the Mobility and Energy sectors
After the basic development phase of intelligent forward operators using Reuniwatt’s cloud observations in the infrared range, it is planned to develop interfaces, i.e. the integration of an experimental version of the operators in the numerical weather prediction systems. The developed algorithms will be based on modern techniques of deep learning. After quality controls, both project partners will conduct case studies in their respective NWP systems. The significantly improved analysis and forecasting of clouds should visibly improve the forecast quality for a wide range of users of DWD’s weather forecasting systems in the area of mobility (air, rail, water and road). Modern transport infrastructures are characterized by high vulnerability in relation to weather events such as storms, thunderstorms, extreme rainfall etc. As the dependency of traffic on the weather will increase with the growing conurbations in the coming years, the seamless interlocking of transport systems from local and long-distance public transport (and other ways of transport) is of crucial importance for further social and economic development in Europe. The analysis and prediction of weather phenomena that are critical for traffic dynamics is an essential step to improve the control of passenger and freight transport and enable its further development.
Besides transport and logistics, the energy sector shall also benefit from the results of ICamCloudOps. European power grids have no national borders, and with the increasing share of renewable energy, the stability of the electricity grids is increasingly dependent on the forecast of wind and clouds. A high quality analysis and prediction of cloud development and cloud dynamics in combination with the prediction of wind speeds and gusts is becoming a critical data source for planning the energy supply from variable natural resources. The further development and gradual improvement of the forecast systems for wind and PV is a challenging and important task for the public sector and the basis for the energy infrastructure of the future. The improved cloud analyses, achieved by ICamCloudOps, will lead to a direct improvement of Reuniwatt’s high-resolution and cloud-resolving NWP model forecasts, serving customers worldwide with high-precision day-scale forecasts of PV and Wind power production.