Earth Weather

The Earth Weather website displays the following forecast products:

Computer Model Forecast Products

Computer Model	Forecast Products
Computer Model	Surface	Upper-air
European Centre for Medium-range Weather Forecasts (ECMWF) computer model and its ensemble prediction system ("ECMWF model")	Wind (including speed of wind gusts), air temperature, relative humidity, mean sea level pressure and rainfall, as well as a post-processed product of “Area of Potential Thunderstorm” (see paragraph 3)	Wind, air temperature, relative humidity and geopotential height Heights include: 200 hPa (approximately 12,000 metres above sea level) 500 hPa (approximately 5,600 metres above sea level) 700 hPa (approximately 3,000 metres above sea level) 850 hPa (approximately 1,500 metres above sea level) 925 hPa (approximately 1,000 metres above sea level)
ECMWF Artificial Intelligence/Integrated Forecast System ("ECMWF-AIFS model")	Wind, air temperature, relative humidity, mean sea level pressure and rainfall
Fengwu Model	Wind, air temperature, mean sea level pressure and rainfall
Fuxi Model	Wind, air temperature, relative humidity, mean sea level pressure and rainfall
Huawei Cloud Pangu-Weather Model ("Pangu model")	Wind, air temperature and mean sea level pressure

Apart from “Area of Potential Thunderstorm”, the above forecasts cover the area bounded by 0 - 60 °N, 85 - 180 °E.

Sea State Forecasts

Sea state forecasts driven by the wind forecasts of the “ECMWF model”, including wave height/direction, swell height/direction and peak wave period forecasts based on WaveWatch III model, and sea current speed/direction forecast produced by the Observatory’s Operational Marine Forecasting System.

Significant wave height roughly approximates to visually observed wave height. It is not the maximum wave height as the wave height of individual waves can be larger. The swell products displayed in the sea state forecast are the primary swells which are generally not available in areas dominated by wind waves, such as those near the centre of tropical cyclones, extra-tropical depressions or low pressure areas. Wave and swell forecasts cover the area bounded by 0 - 60 °N, 85 - 180 °E, while sea current forecast covers the area bounded by 0 - 50 °N, 100 - 150 °E.
Asia-Pacific City Weather Forecasts

Weather forecasts of selected cities in the Asia-Pacific region that are formulated by using a multi-model consensus forecast system of the Observatory. They include 3-hourly forecasts of the temperature, relative humidity, wind direction, wind speed and weather icons over the next nine days.

Users can select the model to be displayed at the top right hand corner of the webpage:
- Forecast wind directions are presented as animated streamlines.
- Users can scroll through the scrollbar to display computer model products at surface or different heights. Taking the example in the following figure, the scrollbar displays the current layer as 850 hPa, indicating that forecast products at 850 hPa (approximately 1,500 metres above surface) level are being displayed.
- Users can overlay isobars (applicable to surface) or isohypses (applicable to upper-air) by selecting on the menu at the top right corner, to display the distribution of mean sea level pressure or geopotential height. Users can also select colour background for other meteorological elements.
As for sea state forecasts, the direction of wave, swell and sea current is shown in the form of animated streamlines. Users can select a background colour overlay of the corresponding significant wave height, swell height, peak wave period or sea current speed through the menu on the top right hand corner.
When rainfall overlay for “ECMWF model” is selected, the “Area of Potential Thunderstorm” forecasts for the area bounded by 10 - 35 °N, 100 - 130 °E will be simultaneously displayed. The “Area of Potential Thunderstorm” forecasts are post-processed products generated using outputs from the ensemble prediction system of the “ECMWF model”. They indicate areas of relatively high (50-74%; indicated by the icon) and high (75% or above; indicated by the icon) potential of thunderstorm development. Based on verification results, when the thunderstorm potential forecast is greater than 50%, there is around 50% or higher probability that thunderstorms would actually occur.
Users can click on the red dot icon or city name on the map directly to view the time series of the weather forecast for the city. The icon on the time series shows the forecast of wind direction and wind speed. The value in the circle is the forecast wind speed in kilometres per hour; the black arrow head points in the direction to which the forecast wind will blow. The above figure shows the forecast of easterly wind (that is, blowing from east to west) with a wind speed of 15 kilometres per hour.

The website is updated daily at the following times:

Computer Model Forecast Products	Around 4:30 a.m.	Provides 3-hourly or 6-hourly forecasts from the same day up to 8 p.m. on the 14th day
Computer Model Forecast Products	Around 4:30 p.m.	Provides 3-hourly or 6-hourly forecasts from the same day up to 8 a.m. on the 15th day
Sea State Forecasts	Around 8 a.m.	Provides 3-hourly forecasts from the same day up to 8 p.m. on the third day
Sea State Forecasts	Around 8 p.m.	Provides 3-hourly forecasts from the same day up to 8 a.m. on the fourth day
Asia-Pacific City Weather Forecasts	Around noon	3-hourly or 6-hourly forecasts from noon of the same day to midnight on the ninth day
Asia-Pacific City Weather Forecasts	Around midnight	3-hourly or 6-hourly forecasts from midnight of the same day to midnight on the ninth day

In addition to the update times above, unscheduled updates may also be made during the period.

The forecasts presented on this website are directly generated from computer systems without manual adjustments, and may be different from weather information disseminated by the Hong Kong Observatory and other meteorological authorities in the region. Users are advised to refer to the latest weather forecasts on the Hong Kong Observatory website and World Weather Information Service website for official weather forecasts.
In particular, the forecast wind fields are based on direct output from a single computer model, and may not be consistent with the tropical cyclone tracks issued by the Observatory, which considers output from multiple computer weather models and also the latest available observation data. Please refer to the following websites to obtain the latest information on the tropical cyclones and to understand how to interpret tropical cyclone forecast track products:
The accuracy of weather forecasts generally decreases while the error generally increases with forecast period. Users may visit the Extended Outlook webpage which provides the probability forecasts of different weather elements as well as their evolution trends in the next 14 days for Hong Kong.
The “ECMWF model” is one of the major global numerical weather prediction models in the world. Numerical weather prediction models make use of physical formulae and weather observation data to simulate the evolution of the atmosphere. For more information on numerical weather prediction and weather forecasting, please visit:
- Simulating Atmosphere
- Observatory's Blog: Let's talk about Weather Forecasting
On the other hand, “ECMWF-AIFS model”, “Fengwu model”, “Fuxi model” and “Pangu model” are artificial intelligence (AI) weather forecasting models. The Fengwu, Fuxi and Pangu forecasts presented on this website are initialized from operational analysis of ECMWF.
- Forecast products from “ECMWF-AIFS model” are provided by ECMWF under the CC BY 4.0 license. Please refer to the ECMWF website for details on the terms of use. For more information about the system, please refer to the following article:
  - Lang, S., Alexe, M., Chantry, M. et al. AIFS - ECMWF's data-driven forecasting system. arXiv:2406.01465 (2024). https://doi.org/10.48550/arXiv.2406.01465
- The “Fengwu model” is provided by the Shanghai Artificial Intelligence Laboratory. For more information on this weather forecasting model, please refer to the following article:
  - Han, T., Gong, S., Ling, F., et al. FengWu-GHR: Learning the Kilometer-scale Medium-range Global Weather Forecasting. arXiv:2402.00059 (2024). https://doi.org/10.48550/arXiv.2402.00059
- The “Fuxi model” is provided by Artificial Intelligence Innovation and Incubation Institute of Fudan University. For more information on this weather forecasting model, please refer to the following article:
  - Chen, L., Zhong, X., Zhang, F. et al. FuXi: a cascade machine learning forecasting system for 15-day global weather forecast. npj Clim Atmos Sci 6, 190 (2023). https://doi.org/10.1038/s41612-023-00512-1
- Forecasts by “Pangu model” are provided under the CC BY-NC-SA 4.0 license. For more information on this weather forecasting model, please refer to the following article:
  - Bi, K., Xie, L., Zhang, H. et al. Accurate medium-range global weather forecasting with 3D neural networks. Nature 619, 533–538 (2023). https://doi.org/10.1038/s41586-023-06185-3
For background information on interpreting computer model forecast products, please refer to the following link:
- Background Information on Computer Forecast Weather Map
For more information on sea waves and swells, please visit:
- Storm Surge and Sea Waves
- Swells from Distant Typhoons
The background geographical information in this webpage comes from Natural Earth. The aim of using Natural Earth is to aid users in appreciating the geographical location of various weather information. This does not imply that information on the map is endorsed by the Observatory.
The webpage is best viewed by browsers that comply with HTML5 standard and that support JavaScript and WebGL (e.g. Chrome, Firefox, Edge, Safari). If your browser does not support HTML5 or you have disabled WebGL in your browser, the response could be relatively slow and some special features may not be supported. Presentation of the webpage may also vary according to different browsers, computers and operating systems.
If you have any comments on this webpage, please fill in the suggestion form or send us an e-mail.