Increasingly, lakes, small and large, around the world are under threat from different forms of pollution. This threatens a large percentage of the Earth’s freshwater supply, as lakes are often among the most critical resources for freshwater.
However, monitoring the health of lakes, including pollution threats, can be complex, as many lakes are large, some are remote, and some regions have so many lakes that it can take a lot of effort and money to monitor them.
Satellite-based remote sensing is proving to be a useful tool to allow effective monitoring of lakes around the world.
Instruments on current satellites make a variety of monitoring possible. One form of monitoring for lakes is to determine thermal pollution, where water temperature increases to unusually high levels due to human activity.
Using sensors on Landsat 8, specifically Operation Land Image (OLI) and Thermal Infra-Red Scanner (TIRS), lake monitoring of thermal change has significantly improved. The OLI sensor covers the near-infrared and short-wave infrared, while TIRS covers two thermal infrared bands.
Combining these bands, it is possible to determine relative lake temperature, and temperature fluctuations, as reflectance properties change under temperature variation from water surfaces, particularly in shallow waters.
These infrared bands also can sense how heat is absorbed in shallow surfaces.
Additionally, thermal change can occur in rivers and lakes due to thermal or nuclear plants using local water resources. The cooling and subsequent emission of the water can cause potentially high temperate variation, causing potentially significant environmental impact. This makes Landsat an effective platform to monitor energy plant pollution on nearby lakes.
While thermal pollution monitoring is one key area, another is different pollutants entered into lakes, such as total suspended matter (TSM). These measures indicate if more suspended pollutants had entered a lake using measurements over time.
Using green, red, or near‐infrared bands on Landsat 8, it is possible to measure reflectance of a lake’s surface and determine variation of optic reflectance and radiance on the water’s surface. The combination of these measures suggests that suspended matter influences reflectance properties and that variation enables one to measure different levels of pollution or suspended matter pollution in a lake.
Determining what type of pollution, for instance organic or inorganic, is more complex. The differentiation of organic pollution in particular is still considered a challenge for satellite systems to detect.
A recent review of the use of cloud computing and deep neural network models using satellite sensor data, including reflectance and radiance data from multi-spectral satellites (e.g., Landsat, MODIS, and MERIS), could also help in better improving different forms of pollution monitoring.
Measuring spectral ratio variation in monitoring chlorophyll, algae, and oxygen variation using deep learning to detect trends in reflectance variation and their significance over time allows research to classify when pollution levels may reach critical or significant stages.
The utilization of ground-based sensors, to calibrate and check the quality of satellite-based results, along with satellite sensor data, are particularly useful and show the most potential in improving observations.
Classification of what sensor reflectance levels indicate different forms of pollution is now possible by combining satellite and, where available, ground-based data. This could enable a more rapid identification of the threat levels different water sources, including lakes, face for different forms of pollution.
However, challenges remain, including monitoring different forms of nutrients entering lakes, where their signatures are not easily identified by satellite sensors.
Monitoring lakes was considered more challenging using older satellite systems, such as Landsat satellites before Landsat 8, as limited spectral range limited threat identification to lakes.
Now, we can better differentiate thermal and suspended particle pollution, while also determining oxygen and possible some organic pollution, although this is still a challenge.
Combining satellite data with deep learning techniques, however, could improve results because minor variation of reflectance data could be more easily seen as classification for certain pollutant types.
Improved ground-based and satellite-based monitoring, including combining these data, may prove to be the best form of analysis which would then allow easier calibration of satellite sensors to improve future efforts that increasingly depend on satellite-only monitoring.
Deeper water pollution is also still a challenge that future satellite system may be better able to detect.
 For more about how Landsat 8 can be used to monitor water and water temperature, see: Yavari, S. M., & Qaderi, F. (2020). Determination of thermal pollution of water resources caused by Neka power plant through processing satellite imagery. Environment, Development and Sustainability, 22(3), 1953–1975. https://doi.org/10.1007/s10668-018-0272-2.
 For more on measuring thermal pollution from power plants, including nuclear and other thermal plants, see: Issakhov, A., & Zhandaulet, Y. (2020). Numerical Study of Technogenic Thermal Pollution Zones’ Formations in the Water Environment from the Activities of the Power Plant. Environmental Modeling & Assessment, 25(2), 203–218. https://doi.org/10.1007/s10666-019-09668-8.
 For more on how different bands on Landsat 8 are used to measure suspended matter pollution, see: Zhu, W., Huang, L., Sun, N., Chen, J., & Pang, S. (2020). Landsat 8‐observed water quality and its coupled environmental factors for urban scenery lakes: A case study of West Lake. Water Environment Research, 92(2), 255–265. https://doi.org/10.1002/wer.1240.
 For more on how satellite monitoring using deep learning and cloud computing could improve pollution detection, see: Sagan, V., Peterson, K. T., Maimaitijiang, M., Sidike, P., Sloan, J., Greeling, B. A., Maalouf, S., & Adams, C. (2020). Monitoring inland water quality using remote sensing: potential and limitations of spectral indices, bio-optical simulations, machine learning, and cloud computing. Earth-Science Reviews, 205, 103187. https://doi.org/10.1016/j.earscirev.2020.103187.