What are the advantages and disadvantages of using Sentinel-2 data for cloud and atmospheric correction?

The lack of thermal bands in Sentinel-2 can pose problems in approximating and removing atmospheric contributions from thermal infrared radiance data.

Sentinel-2 data offers advantages for cloud correction by enabling effective cloud masking and removal. Its resolution aids in detailed analysis, compensating for cloud-obscured areas by utilizing clear spectral bands. However, drawbacks include occasional cloud cover, hindering data acquisition. Cloud correction methods might omit valuable information or demand complex algorithms, impacting accuracy. Moreover, data processing for the same demands substantial computational resources and time. Despite these challenges, Sentinel-2's rich spectral & temporal coverage and free accessibility through open-source platforms like Sentinel Hub provide valuable insights for cloud correction, benefiting various environmental monitoring applications.

Advantages of Sentinel-2 include high revisit frequency which increases chances of getting cloud-free imagery, high 10-60m spatial resolution enabling detailed cloud detection, and broad spectral range allowing different cloud discrimination indices. However, limitations exist. Regions with persistent clouds will still have high cloud cover in Sentinel-2 data, reducing availability of completely cloud-free images. Complex heterogeneous clouds like thin/broken clouds, mixed pixels, and cirrus are challenging to accurately identify and remove. Large variable cloud shadows can alter surface pixel radiometry/spectroscopy, needing extra processing to mitigate effects.

RESOLV is the direct descendant of Sentinel-2 and Landsat programs. Their free data supported our research, prototyping, and testing to make atmospheric correction convenient, accurate and reliable. Because RESOLV is based on the actual mathematical structure of atmospheric degradation, it sets the highest bar for retrieval of surface reflectance. At resolvearth.com, I share a screenshot from August 3, 2021, of a Landsat 8 image of Okanagan Lake, British Columbia with forest fire smoke and the comparison of NASA’s LaSRC and RESOLV, which works with all VNIR satellites. With RESOLV, correction quality increases with resolution due to higher granularity for assessing the atmospheric effect and higher contrast from reduced pixel mixing.

One of the big advantage of Sentinel-2 Satellite Sensor in general is its 13 Bands, specifically the Red, Green, Blue Bands because of which we can interpret the Imagery as if it is taken for any give Mobile Camera. Having said that, with advantages comes disadvantages, and one of the biggest disadvantage of Sentinel-2 Satellite is that in case of availability of Clouds, we cannot see anything on the ground. Now, that is exactly where we can use the advantage of other Satellite Sensors like Sentinel-1, which can see the ground irrespective of whether we have clouds or not.

Atmospheric correction is a vital process in satellite image processing. It refines images for clear viewing, converting digital signals into surface reflectance. This correction removes the impact of atmospheric aerosol particles, like smoke and dust, which can obscure ground features. Achieving accurate surface reflectance is crucial for robust analytics and insightful machine analysis. RESOLV is the direct descendant of Sentinel-2 and Landsat programs. Their free data supported our research, prototyping, and testing to make atmospheric correction convenient, accurate and reliable. In essence, smallsat imagery, bolstered by RESOLV’s innovation, empowers us not merely as spectators but as active participants in preserving our planet.

Advantages: Dedicated Atmospheric Corrections: Sentinel-2 data is designed for atmospheric studies, and tools such as the Sen2Cor processor are specifically developed for atmospheric correction, ensuring accurate radiometric data. Spectral Resolution: The multiple spectral bands of Sentinel-2 aid in accurate atmospheric correction, particularly in separating the atmospheric effects from surface reflectance. Disadvantages: Complexity: Atmospheric correction methods can be complex and may require accurate ancillary data, making the process computationally intensive. Assumption of Homogeneous Atmosphere: Some atmospheric correction methods assume a homogeneous atmosphere, which might not hold true in regions with varying atmospheric condition.

Advantages of Sentinel-2 for atmospheric correction include a dedicated cirrus band to estimate water vapor, high 12-bit radiometric resolution to detect small radiance variations, and wide swath reducing edge effects. However, disadvantages include no thermal infrared band to derive atmospheric temperature or surface temperature, lack of a coastal aerosol band to estimate aerosols, and no onboard calibration to monitor sensor drift over time. While the visible, near-infrared and shortwave infrared bands allow estimating water vapor and aerosols to some extent, the lack of key bands poses challenges for accurate atmospheric correction. Additional data from other sources may be needed to supplement Sentinel-2.

One widely used tool for atmospheric correction of Sentinel-2 data is the Sen2Cor processor, which is provided by the European Space Agency (ESA). Sen2Cor applies algorithms to perform the atmospheric and topographic correction of Sentinel-2 Level-1C products, generating Level-2A products with more accurate surface reflectance values. Another tool is the Atmospheric and Radiometric Correction of Satellite Imagery (ARCSI) software, which provides a set of tools for the atmospheric correction of optical satellite imagery, including Sentinel-2 data.

El procesado de imágenes de nivel L2A quiere presentar atención al tipo de fenómeno que se desea analizar y observar. La corrección atmosférica ayuda a mejorar la visualización de la imagen, pero puede interferir en la visualización de fenómenos específicos que ocurren en la propia atmósfera alterando la visualización de la imagen. Por ejemplo, cuando aplicamos correcciones atmosféricas sobre zonas con presencia de tormentas de arena o emisiones de humo proveniente de los incendios.

Advantages: Open Source Tools: Several open-source tools and algorithms are available for cloud and atmospheric correction of Sentinel-2 data, fostering collaboration and accessibility. Research Community Support: The Sentinel-2 user community actively contributes to the development and improvement of correction methods. Disadvantages: User Expertise Required: Implementing correction methods might require specialized knowledge, and users need to be familiar with the intricacies of the chosen correction algorithm.

Methods and tools: The Sen2Cor software by ESA automates the atmospheric correction process for Sentinel-2 data. SNAP (Sentinel Application Platform) offers user-friendly tools for cloud masking and atmospheric correction. Use of the FLAASH module in ENVI software for accurate atmospheric correction.

RESOLV software is based on the actual mathematical structure of atmospheric degradation, it sets the highest bar for retrieval of surface reflectance. At resolvearth.com, you can witness a typical RESOLV correction: 30-m pixels of corrected from hazy to crystal clear. Correction quality increases with resolution due to higher granularity for assessing the atmospheric effect and higher contrast from reduced pixel mixing.

Here are points from my perspectives- Advantages: -Freely Available Sentinel-2 data -High Resolution, 10m for some band -Multiple Bands -Processing Tools like Sen2Cor from ESA Disadvantages: -Cloud Cover Issues for Dense cloud -Aerosol Challenges -Intensive Computational Demand Overall, Sentinel-2 data offers a valuable and accessible option for cloud and atmospheric correction, particularly for scenes with moderate cloud cover and minimal aerosols. However, careful consideration of limitations and potentially using complementary data sources is crucial for achieving accurate results.

Temporal Resolution: Sentinel-2's revisit time can influence the frequency of cloud-free observations. Depending on the application, the temporal resolution may need to be balanced with cloud correction requirements. Ancillary Data: Availability and quality of ancillary data, such as meteorological information, can impact the accuracy of cloud and atmospheric correction methods. In summary, while Sentinel-2 offers valuable advantages for cloud and atmospheric correction due to its high spatial resolution and dedicated tools, challenges remain, especially in regions with persistent cloud cover and the need for sophisticated correction methods.

When working with Sentinel-2 data, avoid a one-size-fits-all approach to cloud and atmospheric correction. Carefully assess your specific region and application to inform appropriate correction methods. For example, coastal areas may require more specialized aerosol models while forested regions need checks for cirrus confusion. Urban mapping can utilize temporal compositing more readily than crop monitoring. Evaluate prevailing atmospheric and cloud conditions to select optimal processing algorithms. Validate results using ground truth data from field sites or air quality networks tailored to your zones. With thoughtful tuning informed by ground data, Sentinel-2 can provide robust regionally-optimized cloud and atmospheric correction.

We all share a debt of gratitude to the Landsat and Sentinel-2 programs. They’ve taught us how to use their imagery, and especially the immense value from its application. Shining examples of societal value, these programs continue to support the monitoring and management of existential threats from climate change, environmental degradation and food insecurity. My new software, RESOLV, is the direct descendant of Sentinel-2 and Landsat programs. Their free data supported our research, prototyping, and testing to make atmospheric correction convenient, accurate and reliable. RESOLV is the only software package available to correct smallsat data to surface reflectance.