علوم محیطی (Dec 2024)
Comparison of Landsat-8 and Sentinel-2 Satellite Images to Estimate the Amount of Chlorophyll-a in Zaribar Lake
Abstract
Introduction: Population growth and pollution caused by the discharge of all types of municipal, industrial, agricultural sewage, and waste disposal valves have caused the spread of pollution and the limitation of water resources. Surface water sources such as seas, lakes, rivers, and reservoirs of dams are more exposed to pollution than underground water sources. This pollution leads to the increase of nutrients and the blooming of algae and their consequences, such as the increase of chlorophyll-a, change in dissolved oxygen, and ultimately the reduction of water quality. Considering the close relationship between water quality and environmental health and quality of life, it is necessary to monitor the quality of surface water. By monitoring the changes in water quality, it is possible to observe, evaluate, and correct the long-term trends of water quality reduction and also predict its quality changes for the future. Due to the fact that the traditional methods of water quality evaluation are time-consuming, risky, and expensive, experts use remote sensing images to control water quality. Material and Methods: In this research, the chlorophyll-a of Zaribar Lake was investigated from Landsat-8 and Sentinel-2 satellite images in 2019 using the Google Earth Engine platform. For this purpose, the water body of the lake was separated from the non-water body using the NDWI index. Then, four spectral indices 2DBA, 3DBA, NDCI, and FLH-Violet were applied on the separated water body from satellite images. Finally, the predicted amount of chlorophyll-a was compared with the actual amount of chlorophyll-a on the ground in order to select the most suitable spectral index and satellite image to estimate the concentration of chlorophyll-a. Results and Discussion: The results obtained from the comparison of spectral indices showed that 2DBA and NDCI indices are more accurate than 3DBA and FLH-Violet indices in both satellite images and were able to predict the chlorophyll-a concentration well. Therefore, 2DBA and NDCI indices were considered the most efficient indices to evaluate the chlorophyll-a concentration. Also, the amount of R2 obtained from 2BDA and NDCI indices in Landsat-8 and Sentinel-2 satellite images were compared to determine which satellite image is able to estimate the concentration of chlorophyll-a with higher accuracy. The results indicated that the amount of R2 in Sentinel-2 images was 2DBA=0.799 and NDCI=0.794 and in Landsat-8 was 2DBA=0.156 and NDCI=0.125. Therefore, Sentinel-2 was able to predict the concentration of chlorophyll-a more accurately than Landsat-8. This is due to the larger size of Landsat-8 cells compared to Sentinel-2, which can make the detection of chlorophyll-a a challenge in small areas. In addition, there was a one-day time interval between ground sampling and the date of Landsat-8 image collection, when the movement of chlorophyll-a concentration had occurred temporally and spatially, on the surface and in the depth of the lake. However, the ground sampling and the taking of Sentinil-2 images were simultaneous and within the same day. Conclusion: Based on the obtained results, it can be concluded that the use of 2BDA and NDCI indices compared to other indices for small areas in Sentinel-2 images provided higher accuracy than Landsat-8 images. The most important reason is the smaller size of cells in Sentinel-2 images. In order to more accurately evaluate the concentration of chlorophyll-a, the lake must be monitored in a time series and different seasons, because a large volume of water flows from the bottom of the lake through rivers and boiling springs every day, on which the concentration of chlorophyll-a depends; Therefore, the concentration of chlorophyll-a in the lake should be evaluated in low water and high water conditions in order to determine its polluting sources, which unfortunately was not addressed in this research due to the lack of sampling.
Keywords
