Geostationary satellites offer a spectacular images of the Earth, and in particular about the movement of clouds. They are widely used to help in weather forecasting and they are also a great source of information for solar resource estimation. Hello, say I want to know the solar resource available for a place where no nearby irradiance measurements are available. Well, let's zoom out because geostationary satellites observations offer ways to retrieve solar irradiance. One example, here are Meteosat satellite images for one summer day from sat24.com. One image every 15 minutes from midnight to midnight. Let's focus on the Paris region, marked with the starts. It becomes evident from the image that the day was free of clouds around Paris. Actually, there were no clouds within hundreds of kilometers around. Let's see now the global horizontal irradiance; the GHI for this day. Three curves are shown. The ground-based measurements with a pyranometer, the estimations from the satellite images, and a theoretical curve that corresponds to the situation with no clouds. All these three curves agree for this day because it's a sunny day. Another day. This one starts also without clouds over our site, but I cloud mass progressively approaches from the West until it covers the sky from around 10:00 AM until practically the end of the day. Here we see that the satellite-derived irradiance in red is able to capture the irradiance evolution shown by the ground measurements; the blue curve and that the clouds have a reducing effect for this day compared to the theoretical cloud less curve as expected. But how does this work? Meteosat is at 36,000 kilometers of altitude and it continuously scans the same part of the globe. In particular, the Meteosat Second Generation, or MSG, takes measurements with the spatial resolution of 1-3 kilometers at the nadir at the vertical of the satellite, and every 50 minutes there is one full disk image. For each satellite image pixel, irradiance is measured by the onboard instrument; the SEVIRI, at different wavelengths in the solar end infrared bands. The irradiance in the solar bands that are reflected back to the satellite come from different parts of the atmosphere and from molecules, aerosols, and clouds scattering, as well as from the ground reflection. There are several methods that exist to retrieve solar irradiance from this geostationary satellite measurements. The article from Wang et al 2019 collects all of them. What are their strong points of satellite-derived irradiance? Well, global coverage. They cover almost all the globe as seen by this representation with the main geostationary satellites. Then high-resolution data. A new image is scanned every 10-30 minutes and with few kilometers of resolution. Long-term data; more than 20 years of data is now available, which allows the studying irradiance variability, a stability over time, which make a good source, for example, for quality cross-checking of ground-based data and easy to access. Many freely available solar irradiance products exist based on satellite observations. Here we see two notable examples of products, PVGIS from the Joint Research Center and the CAMS Radiation Service from Copernicus. What are the limitations? Well, phenomena like aerosols are not well detected with geostationary satellites and might lead to biases in the estimated solar irradiance. Data at sunrise and sunset is less reliable, that is when the sun is low. The parallax effect, that is when misleading atmospheric conditions, clear, cloudy, is obtained by the satellite sensor. Let's see for these parallax effect some examples. The figure shows a situation when a Cloud only exists in the sun to surface path, but not in the surface to satellite path. In which case, the satellite overestimates the irradiance. The opposite case is shown here. When a Cloud only exists in the surface to satellite path, but not in the sun to surface path; that is when the satellite underestimates solar irradiance. But the main challenge for satellite-derived solar irradiance could be the spatiotemporal representativeness that comes from the resolution of few kilometers and they instantaneously taken images at the steps of about 50 minutes, which make it impossible to capture the real variability observed at small scales. The spatial representativeness is particularly challenged for sites that are in complex terrains like valleys, mountain regions where cloudiness has a high local-scale variability. The temporal representativeness is challenged, in particular, by clouds that are smaller than the pixel size. This cloud induce large GHI, variability at the ground, but due to the satellite resolution, the effect is a smooth. Let's see an example for this. In this day, clouds were seen by the satellite as thin, uniform, as smooth layers that were changing smoothly throughout the day. This translates in their smoothness in the red curve that is observed in the figure. However, luckily observed GHI in blue was much more variable with large peaks because of this sub-pixel reality that was full of Cumulus clouds as seen by the Fisheye Camera Photos. This is more clouds, we're moving near the sound direction and thus producing larger than ration effects like this one around 10:00 AM. Enlarge enhancement peaks like this one before noon. Another challenge is often to distinguish between clouds and the snow both appear bright in the visible image. For example, see this image. It is not absolute to say, what is actually a cloud and what is a snow? For this example, both are widely present. All this being said, let's temporarily zoom out and see that satellite data is a great source of information to have our first answer about the solar potential of a given place. See an example of the compression between satellite-derived solar irradiance from a particular product versus the ground-based measurements from a high-quality pyranometer at a monthly scale for the case of Palazzo in France. A high correlation of the points is clearly seen both for winter month with low irradiance and summer month with high irradiance. At a yearly basis, this is how satellite-derived and ground-based irradiance compare at Palazzo. Satellite estimations in red capture the year-to-year variability fairly well with, in this case, a slight of estimations on average, that's about 3.4 percent. Geostationary satellites are constantly getting upgraded with the new versions of these satellites and new products becoming available with better temporal and spatial resolutions. So the use of this data is supposed to be even more relevant for solar energy applications. Thank you.
