Using MATLAB to Empower Modern Numerical Weather Forecasts Dr. - PowerPoint PPT Presentation

Using MATLAB to Empower Modern Numerical Weather Forecasts Dr. Martin Fengler CEO

World Class Talent in Meteorology, Data Science, Drone Development and Service Delivery We are proud of Meteomatics’ fair, hardworking, ‘can -do' culture and a highly skilled multi-disciplinary team who rise to the challenge with our customers in a positive fashion. Creativity is a core skill whether it be in thinking, design, architecture or science. 2

Why Does Weather Matter? Better understanding of It affects our the weather helps daily life. reducing business costs. Better understanding of the It affects our weather improves business. predictive maintenance. Better understanding of the It is highly weather reduces the impacts of natural hazards. variable. 3

Key Takeaways Weath ather er API PI & M MATLAB AB enable us to: ... model gathered drone data … simulate new measurement techniques … implement physical parametrizations … visualize meteorological data … carry out statistical analyses … enrich training of machine & AI learning with weather data … give deeper insights into your weather related business 4

Key Challenges • Inaccuracy of Forecasts • Access to Historical Data • Huge Amount of Data • Inconsistent Data Formats 5

Current Data Situation Trigger for Storms Satellite 1000 km Aircraft Radar 100 km Balloons Low Stratus 10 km PBL* up to 1.5 km Radar 1 km Fog Limited Data 100 m 10 m * PBL = Planetary Boundary Layer 6 Weather Station Laser Sound/Microwave

Improving Data Situation Satellite 1000 km Aircraft Radar 100 km Balloons 10 km PBL* up to 1.5 km 1 km Meteodrone 100 m 10 m 7 * PBL = Planetary Boundary Layer Weather Station Laser Sound/Microwave

Our Milestones EVLOS Approval First Commercial First Projects with Flight Campaign NSSL/NOAA Passed Total Hazard & BVLOS Approval Risk Analysis 2015 2014 Start with AM2S 2012 2017 Roll-Out Switzerland Proof of Concept 2016 Full System Test 2013 Product 8 Readiness First Prototype

9

Meteodrone Sensors & Flight Profile v Pressure Dew Point Accuracy: 0.2 °C Accuracy: 0.1 hPa α Response Time: 250 ms Response Time: < 4 s Temperature Wind Speed & Direction Accuracy: 0.1 °C Accuracy: < 1 m/s Response Time: 1 s Response Time: 250 ms The aircraft automatically compensates wind drag: Relative Humidity • Compute wind speed and direction Accuracy: < 2 % from roll & nick angle Response Time: < 4 s Vertical flight profile up to 3’000 m • • Currently working on increasing flight altitude to 6’000 m Prototyping done in MATLAB Modelling & Simulation Sensors are radiation-shielded and mounted in the rotor downwash. 10

Modelling & Simulation of Meteodrone Input Drone Model Share Results • Roll and Pitch angle • Physics based • Send data in real-time to ground • Power Consumption • Automatic wind drag compensation station • Comparison to wind tunnel and • Post-processing / WRF model-input outside conditions • Weather API • Postprocessing and calibration • MATLAB / C++ v • Deployed on ARM Processor α 11

Amlikon 21. – 22.09.17 Temperature White dots indicate the drone flight track. Visualization done in MATLAB 12

Amlikon 21. – 22.09.17 Relative Humidity Visualization done in MATLAB 13

Amlikon 21. – 22.09.17 Wind Speed & Direction Visualization done in MATLAB 14

Amlikon 05. – 06.06.17 Temperature Ground Inversion Relative Humidity Shallow Fog: Up to 150 m 100% RH 15

16

Morning Fog at Lake Constance 05.04.17, 7 am & 8 am Swiss1k With Swiss1k Without Satellite Cloud Cover Meteodrone Data Meteodrone Data Shallow Fog No Fog Resolved Shallow Fog Meteodrones in Schaffhausen, Amlikon and Marbach until 5 am Shallow Fog No Fog Resolved Shallow Fog 17

Thunderstorms in St.Gallen 29. – 30.05.17 Difference With Meteodrone Without Meteodrone 29.05.17 29.05.17 29.05.17 Swiss1k was the only model to capture these storm cells and forecasted them 23 hours ahead! 18

Swiss1k Workflow Customer API FTP/E-Mail Production Secondary Data and Primary Data and Control Link Control Link Model Output, GRIB/NETCDF Sea & Lake Surface Temperature Satellite Information Meteodrone Data WRF Ground Station Control Center Weather Station 19

Weather Maritime Digital Terrain Models Satellites Lightnings Radar Drones Stations Data Model Management Layer: … MeteoCache Users, Licenses, Logs Monitoring, Sanesco, Cache 1 Cache 2 Cache N Service Layer: RAM Cleaner … Weather API API 2 API 1 API M Load Balancer Weather API Firewall Open Source * Connectors Internet User Requests * * * * * * … Excel Python MATLAB PHP Google Maps C++ 20

Weather API USP Weather data as On the fly Hyperlocal Variety of formats Simple one-stop Detailed and up- Flexible & fast a single version of calculation for forecasts and connectors in access to high to-date integration & truth most up-to-date delivering different quality weather documentation usage forecasts enhanced programming data worldwide temporal and languages spatial resolution 21

Variety of Possible Integrations Weather API 22

Weather API in MATLAB File Exchange 23

Weather API in MATLAB 24

Weather API in MATLAB 25

Weather API in MATLAB Global, diffuse, direct and clear sky radiation Wind Power 26 MSG Satellite Data Solar Power

Key Takeaways Weath ather er API PI & M MATLAB AB enable us to: ... model gathered drone data … simulate new measurement techniques … implement physical parametrizations … visualize meteorological data … carry out statistical analyses … enrich training of machine & AI learning with weather data … give deeper insights into your weather related business 27

Thank You Meteomatics AG Lerchenfeldstrasse 3 9014 St. Gallen Switzerland Your Contact Meteomatics GmbH Schiffbauerdamm 40 Dr. Martin Fengler Office 4406 10117 Berlin CEO Germany mfengler@meteomatics.com Meteomatics Ltd www.meteomatics.com Sowton Business Center Capital Court Bittern Rd Exeter EX2 7FW United Kingdom 28