Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Estimating lake water volume fluctuations using Sentinel-2 and ICESat-2 remote sensing data

Alexantonakis, Pavlos LU (2024) In Master Thesis in Geographical Information Science GISM01 20241
Dept of Physical Geography and Ecosystem Science
Abstract
This study focuses on the estimation of lake water volume fluctuations using open-source remote
sensing data and evaluates its accuracy. The research follows a three-step methodology, starting
with water area estimation from Sentinel-2 imagery, followed by water level estimation using
ICESat-2 satellite data, and concluding with the calculation of lake water volume differences,
testing the regression modeling and Triangular Irregular Networks (TINs).
For the water extent estimation, three different methods are being compared – Normalized
Difference Water Index (NDWI), Modified Normalized Difference Water Index (MNDWI), and
Random Forest machine learning algorithm - against manual digitization. All three methods yield
accurate... (More)
This study focuses on the estimation of lake water volume fluctuations using open-source remote
sensing data and evaluates its accuracy. The research follows a three-step methodology, starting
with water area estimation from Sentinel-2 imagery, followed by water level estimation using
ICESat-2 satellite data, and concluding with the calculation of lake water volume differences,
testing the regression modeling and Triangular Irregular Networks (TINs).
For the water extent estimation, three different methods are being compared – Normalized
Difference Water Index (NDWI), Modified Normalized Difference Water Index (MNDWI), and
Random Forest machine learning algorithm - against manual digitization. All three methods yield
accurate results on the quantitative comparison, with performance metrics consistently
exceeding 0.95, indicating the absence of a clear preference among these approaches. During
the qualitative comparison, misclassified areas were revealed, indicating the problems caused by
clouds, shadows and leaves upon the lake which sometime makes the process of accurately map
the water extent difficult even for the human eye.
For the water level estimation, ICESat-2 satellite observations were found to be accurate in lake
water level monitoring, with a standard deviation of approximately 5 cm, surpassing the nominal
accuracy of 6.1 cm. However, limitations are noted in terms of temporal resolution, impacting
their use in combination with Sentinel-2 acquisitions.
Finally, for the estimation of water volume fluctuations, the research explores two distinct
approaches for calculating water volumes within open water surfaces. The first approach
employs regression modeling, deriving a regression equation to estimate water volumes based
on area computations. The second approach involves Triangular Irregular Networks (TINs).
However, the research discerns that the latter method is not applicable in cases where a
significant difference in scale between vertical and horizontal changes exists.
In conclusion, the presented methodology can potentially have benefits for decision-makers and
water organizations since it utilizing open-source data and tools for the estimation of lake water
volumes. Future work may explore more sophisticated data-driven approaches, including
convolutional neural networks, and the incorporation of additional data sources, such as
altimetric data from platforms like Jason-2, to improve accuracy and address limitations related
to clouds and temporal resolution. (Less)
Popular Abstract
Understanding changes in lake water levels is crucial for managing water resources. Our project uses advanced space technology to tackle this task. We aimed to develop a simple, reliable method to monitor changes in lake water volume over time, offering valuable insights for decision-makers.
We utilized satellite imagery from Sentinel-2 satellite to assess the extent of water coverage in lakes. Additionally, we also used data from ICESat-2 satellite to measure water height. Our research explored various techniques for calculating water volume changes, including mathematical models and also geometrical 3D approaches.
Water extend data have been analyzed with various remote sensing methods. All of them provided accurate results which... (More)
Understanding changes in lake water levels is crucial for managing water resources. Our project uses advanced space technology to tackle this task. We aimed to develop a simple, reliable method to monitor changes in lake water volume over time, offering valuable insights for decision-makers.
We utilized satellite imagery from Sentinel-2 satellite to assess the extent of water coverage in lakes. Additionally, we also used data from ICESat-2 satellite to measure water height. Our research explored various techniques for calculating water volume changes, including mathematical models and also geometrical 3D approaches.
Water extend data have been analyzed with various remote sensing methods. All of them provided accurate results which suggests no clear preference among them. However, several challenges have been identified such as the presence of clouds, shadows and leaves over the lake, that were visible in the satellite images.
Water level data proved to be applicable for this task, since their precision was found to exceed our initial expectations. However, their limited frequency posed certain challenges, when it comes to availability of this data source.
For the water level changes estimation, our mathematical approach was found to be effective and provided the expected results. On the other hand, the geometrical method faced certain limitations. Due to the specific lake’s individualities, when it comes to the actual level changes compared to the extent changes, the latter wasn’t able to provide comparable results.
Our research offers a promising approach for water resource management and decision-making. By leveraging freely available data and tools, we can enhance our ability to monitor and understand changes in lake water levels. Future endeavors may involve more advanced methodologies and data from other sources too, to further refine our analysis. (Less)
Please use this url to cite or link to this publication:
author
Alexantonakis, Pavlos LU
supervisor
organization
course
GISM01 20241
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Geography, GIS, Remote Sensing, Machine Learning, Water Volume, Satellite Imagery, Open-source data
publication/series
Master Thesis in Geographical Information Science
report number
175
language
English
id
9151259
date added to LUP
2024-05-02 10:37:25
date last changed
2024-05-02 10:37:25
@misc{9151259,
  abstract     = {{This study focuses on the estimation of lake water volume fluctuations using open-source remote
sensing data and evaluates its accuracy. The research follows a three-step methodology, starting
with water area estimation from Sentinel-2 imagery, followed by water level estimation using
ICESat-2 satellite data, and concluding with the calculation of lake water volume differences,
testing the regression modeling and Triangular Irregular Networks (TINs).
For the water extent estimation, three different methods are being compared – Normalized
Difference Water Index (NDWI), Modified Normalized Difference Water Index (MNDWI), and
Random Forest machine learning algorithm - against manual digitization. All three methods yield
accurate results on the quantitative comparison, with performance metrics consistently
exceeding 0.95, indicating the absence of a clear preference among these approaches. During
the qualitative comparison, misclassified areas were revealed, indicating the problems caused by
clouds, shadows and leaves upon the lake which sometime makes the process of accurately map
the water extent difficult even for the human eye.
For the water level estimation, ICESat-2 satellite observations were found to be accurate in lake
water level monitoring, with a standard deviation of approximately 5 cm, surpassing the nominal
accuracy of 6.1 cm. However, limitations are noted in terms of temporal resolution, impacting
their use in combination with Sentinel-2 acquisitions.
Finally, for the estimation of water volume fluctuations, the research explores two distinct
approaches for calculating water volumes within open water surfaces. The first approach
employs regression modeling, deriving a regression equation to estimate water volumes based
on area computations. The second approach involves Triangular Irregular Networks (TINs).
However, the research discerns that the latter method is not applicable in cases where a
significant difference in scale between vertical and horizontal changes exists.
In conclusion, the presented methodology can potentially have benefits for decision-makers and
water organizations since it utilizing open-source data and tools for the estimation of lake water
volumes. Future work may explore more sophisticated data-driven approaches, including
convolutional neural networks, and the incorporation of additional data sources, such as
altimetric data from platforms like Jason-2, to improve accuracy and address limitations related
to clouds and temporal resolution.}},
  author       = {{Alexantonakis, Pavlos}},
  language     = {{eng}},
  note         = {{Student Paper}},
  series       = {{Master Thesis in Geographical Information Science}},
  title        = {{Estimating lake water volume fluctuations using Sentinel-2 and ICESat-2 remote sensing data}},
  year         = {{2024}},
}