Frequently Asked Questions

Sunsethue uses a ray-based approach to determine if a sunrise or sunset will be good. The algorithm determines if light can be reflected from the clouds that are visible for a certain location. If a lot of light of the setting sun can be reflected, this results in a high quality score. This is the general intuition, the actual workings are a bit more complex than this.

The sunrise or sunset quality score is a value between 0% and 100% indicating how good the sunrise or sunset is forecasted to be. The score is a summary of multiple aspects of a sunrise or sunset: the colors, duration of the colors, expanse of the colors, etc. In short, it tries to capture how spectacular the sunrise or sunset is forecasted to be in a single number. For more information on how to interpret the score, click the button below.

A lot of information regarding the upcoming sunrises and sunsets are available on Sunsethue. The web app is centered around the sunrise or sunset quality metric, but you can also access information about the cloud cover during sunrise or sunset, the amount of clouds at the horizon and the uncertainty. Of course, you can also access information regarding the time of sunrise and sunset, and about the blue hour and golden hour.

The algorithm uses weather data (mainly cloud cover data) as input. Weather forecasts, and especially cloud cover forecasts, can be wrong sometimes. Small changes in this input may result in a vastly different sunrise or sunset quality. The forecasts generally get more certain as they become closer in time, so check back in regularly for the best forecasts. You can also use the uncertainty metric to assess the level of uncertainty for a given sunrise or sunset.

Since a global weather model is used, the forecasts are available for all locations in the world. One exception: in the periods around the winter and summer solstices, data for parts of Scandinavia, Alaska and Canada are not available. Be careful in mountainous areas, as rapidly changing weather conditions can result in more uncertain and less accurate forecasts.

In the periods around the winter and summer solstices, there is no data for the northern latitudes. The reason for this is that the algorithm makes some approximations regarding the position of the sun, these approximations tend to deteriorate quickly in extreme scenarios (i.e. in northern regions near one of the solstices).

The forecasts are updated four times per day, at around 05:40, 10:20, 16:40, 22:20 UTC. The previous and next expected update are visible in the web-app, so you know when to check the latest forecasts.

There are two main reasons for this. Firstly, the weather data that is used as input gets more and more uncertain for multiple days ahead. Currently, forecasts for more than three days ahead tend to change too often to be useful. Secondly, running the model for more days requires more computational power, and thus results in higher costs.

The Sunsethue website is fully free to use for anyone, no account is needed to access the forecasts. The general quality forecasts for the upcoming days will remain free forever, some more advanced metrics may be moved to a premium version later on. I also chose to not use any ads on Sunsethue to improve the user experience, and because I really dislike ads myself. Currently, the only paid service on Sunsethue are the Premium Alerts.

A mobile app for Sunsethue is currently in development, you can read more about it by clicking the button below.

Sunsethue has a well-documented REST API that you can use in your projects and automations. Using the API you can access the sunrise and sunset quality, time and magic hours for any location. The API is free for personal use for up to 1000 credits per day.

Even though Sunsethue is already more than three years old, it is still in beta. This is mainly because I'm still very actively changing the underlying model to improve the forecasts. I hope to launch” Sunsethue when I’ve settled on a good model architecture and have sufficiently scalable systems in place.