Automated detection of Yarka island (Baikal lake) based on NDWI, MNDWI and WRI spectral indices

DOI: 10.35595/2414-9179-2025-2-31-212-220

View or download the article (Rus)

About the Authors

Fedor M. Andreyev

Irkutsk State University,
1, Karla Marksa str., Irkutsk, 664003, Russia,
E-mail: fmandreev@yandex.ru

Ekaterina N. Sutyrina

Irkutsk State University,
1, Karla Marksa str., Irkutsk, 664003, Russia,
E-mail: ensut78@gmail.com

Abstract

The paper presents the results of an automated algorithm designed to detect Yarki Island on satellite imagery and estimate its area utilizing Python-based tools. Employing the developed algorithm for calculating spectral indices—WRI (Water Ratio Index), NDWI (Normalized Difference Water Index), and MNDWI (Modified Normalized Difference Water Index)—we processed 86 satellite scenes from Landsat-5 TM, Landsat-8 OLI, and Landsat-9 OLI spanning the period between 2008 and 2024. These spectral indices served as a means to isolate land areas within the image dataset. To enhance the precision of constructing the land mask derived from WRI data, we adjusted the binarization threshold according to synchronously recorded water levels of Lake Baikal. This calibration process enabled the threshold’s adaptation to variations in lake water levels. We evaluated the Root Mean Square Error (RMSE) to assess the accuracy of these methods. Additionally, reference masks of Yarki Island were created, allowing us to compare the accuracy of automated area determination using distinct spectral indices: WRI, NDWI, and MNDWI. Our findings revealed that the WRI index demonstrated superior performance with RMSE = 0.010 km2 after incorporating adjustments to the binarization threshold corresponding to Lake Baikal’s water level. In comparison, the NDWI (RMSE = 0.067 km2) and MNDWI (RMSE = 0.087 km2) exhibited lower accuracies. By implementing this dynamic threshold adjustment approach, the method improved land detection accuracy based on WRI values by approximately 20 % relative to a static threshold.

Keywords

remote sensing, Landsat, water indices, morphometric characteristics, adaptive binarization

References

  1. Boldanova E.V. Modeling the Dependence of the Area and Volume of Posolsky Sor on the Level of Lake Baikal. Izvestiya Irkutskogo Gosudarstvennogo Universiteta. Seriya “Nauki o Zemle” (The Bulletin of Irkutsk State University. Series “Earth Sciences”), 2023. V. 44. P. 33–43 (in Russian). DOI: 10.26516/2073-3402.2023.44.33.
  2. Feyisa G.L., Meilby H., Fensholt R., Proud S.R. Automated Water Extraction Index: A New Technique for Surface Water Mapping Using Landsat Imagery. Remote Sensing of Environment, 2014. No. 140. P. 23–35. DOI: 10.1016/j.rse.2013.08.029.
  3. Gao B.-C. NDWI—A Normalized Difference Water Index for Remote Sensing of Vegetation Liquid Water from Space. Remote Sensing of Environment, 1996. V. 58. P. 257–266.
  4. Ji L., Zhang L., Wylie B. Analysis of Dynamic Thresholds for the Normalized Difference Water Index. Photogrammetric Engineering and Remote Sensing, 2009. V. 75. No. 11. P. 1307–1317.
  5. Kataev M.Yu., Bekerov A.A. Methodology for Detecting Water Bodies Using Multispectral Satellite Measurements. Proceedings of TUSUR University, 2017. V. 20. No. 4. P. 105–108 (in Russian).
  6. Kurganovich K.A., Noskova E.V. Using Water Indices to Assess Changes in the Water Surface Areas of Steppe Soda Lakes in the Southeast of Transbaikalia Based on Remote Sensing Data. Transbaikal State University Journal, 2015. No. 6 (121). P. 16–24 (in Russian).
  7. Laonamsai J., Julphunthong P., Saprathet T., Kimmany B., Ganchanasuragit T., Chomcheawchan P., Tomun N. Utilizing NDWI, MNDWI, SAVI, WRI, and AWEI for Estimating Erosion and Deposition in Ping River in Thailand. Hydrology, 2023. V. 10. P. 70. DOI: 10.3390/hydrology10030070.
  8. Manilyuk T.A., Maslova A.V. Study of Lake Water Surfaces Using Water Indices Based on Remote Sensing Data. Transbaikal State University Journal, 2017. V. 23. No. 3. P. 4–11 (in Russian). DOI: 10.21209/2227-9245-2017-23-3-4-11.
  9. Morozova V.A. Calculation of Indices for Identifying and Analyzing Characteristics of Water Bodies Using Remote Sensing Data. Current Problems of Territorial Development, 2019. No. 2. P. 1–12 (in Russian).
  10. Mustafa M., Khalid H., Modher H., Modher A. Using Water Indices (NDWI, MNDWI, NDMI, WRI and AWEI) to Detect Physical and Chemical Parameters by Apply Remote Sensing and GIS Techniques. International Journal of Research—GRANTHAALAYAH, 2017. V. 5. P. 117–128. DOI: 10.29121/granthaalayah.v5.i10.2017.2289.
  11. Potemkina T.G., Potemkin V.L. Sediment Runoff of Lake Baikal Tributaries: Changes and Trends. Izvestiya Irkutskogo Gosudarstvennogo Universiteta. Seriya “Nauki o Zemle” (The Bulletin of Irkutsk State University. Series “Earth Sciences”), 2023. V. 43. P. 79–90 (in Russian). DOI: 10.26516/2073-3402.2023.43.79.
  12. Potemkina T., Sutyrina E., Potemkin V. Changing of the Riverine Sediment Load Supply into Lake Baikal: The Natural and Anthropogenic Causes (Russia). Quaternary International, 2019. V. 524. P. 57–66.
  13. Pshenichnikov A.E. Using Spectral Indicators in Identifying Thermokarst Lakes Based on Remote Sensing Data. Geoinformatika, 2021. No. 2. P. 67–74 (in Russian). DOI: 10.47148/1609-364X-2021-2-67-74.
  14. Sun F., Sun W., Chen J., Gong P. Comparison and Improvement of Methods for Identifying Waterbodies in Remotely Sensed Imagery. International Journal of Remote Sensing, 2012. V. 33. No. 21. P. 6854–6875.

For citation: Andreyev F.M., Sutyrina E.N. Automated detection of Yarka island (Baikal lake) based on NDWI, MNDWI and WRI spectral indices. InterCarto. InterGIS. Moscow: MSU, Faculty of Geography, 2025. V. 31. Part 2. P. 212–220. DOI: 10.35595/2414-9179-2025-2-31-212-220 (in Russian)