The holy grail for landscape photographers is to be able to accurately predict sunrise and sunset colors. I’m sure we’ve all had the experience of thinking that the sunset was going to be great only to have the colors fizzle out. Or perhaps you thought there wasn’t much potential and started driving home only to be stuck on the interstate when the colors exploded.
In this article, I will discuss how I predict the colors for sunrise & sunset as well as link to the various tools that I use to inform my predictions.
Requirements for Sky Color
First, lets talk about what we need to get those colorful skies. In some ways, it’s simple and we need only two things:
- High clouds to catch the sun’s rays as it rises/sets.
- A clear horizon so that the sunlight is not blocked.
The ideal situation is that the sun is at or below the horizon and the light rays have a direct and unobstructed line of sight to the underside of the clouds in your composition. As the light travels through the atmosphere, the blue light rays scatter more than when the sun is directly overhead. This leaves only the gorgeous warm yellows, oranges, pinks and reds to illuminate the underside of the clouds.
However prediction gets complicated for multiple reasons. First, we have to predict many hours in advance. For sunrise, I need to decide the night before whether I’m going to get up early and for sunset I need at least an hour of lead time for local shoots. Second, “clear” and “high” are not binary attributes. There are obviously varying degrees of both characteristics and they interact to either let the colors shine through or to block the sun.
For example, in this next picture, although there is a marine layer (low clouds on the horizon), it was not dense enough nor high enough to prevent a spectacular sunset.
Sometimes the day can be completely overcast and gray, but if there is even a small opening in the clouds, the sky can explode at sunset. In the next scene, the opening was the small bright wedge that you can see by the end of the pier. As sun dipped below the horizon, it was able to ignite the sky through the wedge. This was one of the most overpowering sunsets I’ve experienced and I actually found it to be too colorful for my taste.
In this third example, we have multiple cloud formations. On the horizon, there is an extremely thick low-lying cloud layer that blocks the sun. But we also have high wispy clouds that can still caught the sunlight colors. In the center of the frame, there are a few low clouds blowing in from offshore that were not high enough to catch any direct light.
In this last example, we have a cloud formation that is almost directly overhead of the shooting position. Most of the cloud was not illuminated but the trailing edge was just high enough to catch the light as the sun dipped below the marine layer and horizon. There is another cloud in the distance that was too low for sunset color.
You will have to use your judgement about the cloud height and clearness of the horizon. This will be difficult at first, but as you gain experience, it will become easier and you will make more accurate predictions.
Short Time Frame Predictions (0-4 hours beforehand)
I use different methods for predicting whether the sunrise/sunset is going to be good depending on the lead time. For short time-frame predictions, I try to get a direct read of the cloud cover based on current conditions. I then consider wind speed and direction before making a decision.
Naked Eye View of the Horizon
My preference is to get a direct look at the horizon with my own eyes. For sunrise you can often see the color starting to emerge well before the colors peak, perhaps as much as hour before the official sunrise time. The peak color usually occurs perhaps 15-20 minutes before sunrise so that leaves you some time to get to your shooting location or decide if you want to go back to sleep.
For sunset I will look at the horizon and judge the size and density of any clouds that might block the light. I will also look for any breaks in the cloud cover where the sun could shine through. Some of the most colorful sunsets I’ve experienced were when the sky was almost completely overcast but there was a small opening of clear sky for the sun to come through.
Although my office window faces west, I can’t actually see the horizon due to buildings blocking the view so I will sometimes go to the top of a nearby parking garage (10 stories high).
Another option is to ping a photography friend who is in a better location. Since I’m often shooting seascapes, I’ll message friends who work near the coast to get their read on the sky.
Real-Time Satellite View of Cloud Cover
In addition to a naked eye view, I will check real-time satellite images of cloud cover. I use the GOES satellites (Geostationary Operational Environmental Satellites) operated by NOAA. To see the satellite images first go to
Then select your region. This will take you to a new page with a number of imagery options based on the light wavelength. I usually select the first option: True Color daytime, multispectral IR at night.
This reveals a real-time image of the current cloud cover conditions. Looking at my area, San Diego, shows that there aren’t any clouds for hundreds of miles. Not a good candidate for sky color although it’s great conditions for shooting the night sky.
Below is a night image using infrared imaging. On the image, the blue colors are liquid water clouds (e.g. fog or stratus) while white or gray represents higher ice clouds. Note that the city lights do not come from the satellite and are instead added from separate static source.
This was a few minutes before sunrise and you can see a layer of clouds on the south side of San Diego. I couldn’t shoot the sunrise this morning but thought there was a good chance of getting decent color in the sky as the forecast from the night before was for high clouds (see section below on long range forecasts).
The website also provides an animation of the last hour’s worth of satellite images. This will show the direction and speed of cloud movement which can help your prediction. I typically look at this if the cloud pattern is sparse and there’s more than hour before sunrise or sunset.
If I can’t get a direct visual of the horizon, I will look at webcams that are pointed in the correct direction. I live in San Diego and there are many webcams here because apparently surfers also check these too. Here are a few I like to look at
It can be tricky judging cloud density on the webcam feed and you will probably need some experience comparing the webcam’s view to the actual situation when you go out to photograph. Webcams also use automatic exposure and white balance which can dramatically change the color and brightness of the image.
You may also have to do some searching on Google to find a good set of webcams for your area. Keep in mind that these often go in and out of service. In addition, double check the time-stamp as sometimes they may broadcast an older feed.
Long Time frame Predictions (4+ hours beforehand)
For longer range forecasts, I use a few different approaches and tools to predicting sky colors.
Weather Models (ECMWF, GFS, NAM, etc)
There are multiple weather models that can forecast the location, density, and height of cloud cover. I use these when I need to predict more than 4 hours ahead of time. The most popular models that you may encounter are:
- ECWMF (European Centre for Medium-Range Weather Forecasts)
- GFS (Global Forecast System)
- NAM (North American Mesoscale Model)
These models have different algorithms, spatial resolutions, and update frequencies. In my experience the ECWMF model (European) has been the most accurate for me in San Diego which is somewhat surprising because GFS and NAM are made by US organizations. But you should test this for your specific area.
I find the most convenient place to view the forecast is on windy.com (also available as a phone app). You can see both the cloud cover percentage and altitude of the cloud base.
Windy can also display GFS information but I prefer to view that model’s forecast here:
First select a region and then choose either “Lower Dynamics” or “Upper Dynamics” and pick cloud fraction.
Depending on what site you visit, you may see the forecast time listed in your local time zone or as “zulu” time. Zulu time refers to Universal Coordinated Time (UTC) which was formerly called Greenwich Mean Time. In San Diego, we are at UTC-7 during daylight saving time and UTC-8 during standard time.
Rule of Thumb
Sometimes I use a simple rule to predict sunrise/sunset color:
IF there has been a good sunrise (sunset) in the past few days AND there is a forecast for partial cloud cover THEN there is a good chance of color at sunrise (sunset)
This can be helpful during the fall/winter months in my area as we will often get a few good sunrises/sunsets in a row.
Prediction Websites and Apps
Finally, there are websites and apps that take in all of the forecast data and try to automatically predict sunrise/sunset color. The two most popular ones are SunsetWx and Skyfire (IOS app only).
In my experience, SunsetWx is not very good. It’s based on the GFS & NAM forecast models which are not as accurate as ECWMF in my area. But if SunsetWx predicts a very colorful sunrise/sunset I will often check my other sources. I don’t use Skyfire as it requires a paid subscription but friends who have used it say that it is only so-so at making preditions.
Here is an example prediction map from SunsetWx. A more colorful sunset is denoted by warmer colors (i.e. yellow, orange, and red).
In this section, I’ll include some examples of predictions I’ve made. This is a work in progress and I’ll add more over time.
Sunset 1/2/20 in San Diego
About 2.5 hours before sunset, I was deciding on whether to go to the coast in La Jolla and shoot the sunset. Here are screenshots of the information I looked at before deciding:
- satellite image of clouds
- animation of cloud movement
- SunsetWx prediction
- webcam facing west to the coast
The satellite images and animation suggest good conditions for sunset (partly cloudy with breaks for the sun to light the sky). SunsetWx is negative and suggest poor conditions. The webcam indicates that the clouds are thicker than I would expect from the satellite images but because it’s winter, I know that the sun can often shine through the clouds to provide color.
Ultimately I decided to go and here are a few shots from my camera with a flat profile and minimal processing:
Learn seasonal & local patterns for your area
I think it’s important to learn seasonal and local weather patterns. For example, in San Diego
- We get heavy marine layer that blocks the sun in spring and summer months. During this time we might have one good sunset a month. So the prior probability is very low and unless there are other strong signals I assume there will be lackluster colors.
- Sunset season starts around beginning of November and goes until March/April. We might average about one good sunrise/sunset a week. But there will be some weeks where we get as many as three or four great days in a row.
- Due to topography, various spots tend to develop certain cloud patterns which you can leverage. For example, Balboa Park is located on the top of a hill and often gets clouds that form as the wind comes in from offshore. These can often light up (although you have to be in the right spot to take advantage of it).
- Santa Ana winds which blow hot dry air from inland areas out to the coast often occur in the fall (although they can be year round). Usually this means very clear air and no clouds which is bad for sunrise/sunset color but great for night sky photography.
- The coast often develops heavy fog or cloud cover that rolls in right around sunset. This blocks the sky color because you are literally standing in the cloud, but the sunset can still be great if you move a few miles inland.
I was initially very bad at predicting sunrise and sunset color but this is a skill that can be learned. I started by watching the cloud formations and weather information and making a prediction of how I thought the sky would turn out regardless if I could actually go out and shoot that day. I then made note of how the sunrise/sunset turned out – this helped calibrate how I interpreted the various information sources.
Eventually with enough feedback, you’ll become better at predicting sky colors. However, prediction is inherently hard and there’s a limit as to how accurate you can be. I don’t expect to ever be more the 30-50% accurate at predicting a colorful sunset and significantly less than that for sunrise. However, my motto is when in doubt, go out and shoot. In the worst case, you get to spend a pleasant hour or two in the outdoors.