Monitoring the Lake Mead Levels from 2000 to 2025

Abstract

Lake Mead, Nevada

The Mojave Desert has had a large history of drought conditions over the years. Understanding these conditions and how they can affect local water availability has become a large topic of study for California, Nevada, and Arizona. This study investigates how the surface area of Lake Mead has changed from 2000 to 2025 using remote sensing method. In order to analyze these results, I have viewed Lake Mead using Google Earth's Historic Images Timelapse to visually ascertain the lake area differences throughout the years. I used the random trees classification on Landsat images from the years 2000, 2005, 2010, 2015, 2020, 2025, and produced Sankey Diagrams and percent change charts to see how much of the land cover has changed around the lake. This project will focus mostly on the area of the lake, rather than the depth of the water. As the analysis suggests, the lake shows a drastic decrease in area, with a decline of 15,254 hectares (58.9 Square Miles), most notably in the northern and eastern sections of the lake. Continued use and decrease of water levels in the lake could lead to drastic problems for those who rely on the lake for their water needs. If the lake continues to reduce in size, it will result in the loss of not only the water as a resource, but the energy that the lake produces in the Hoover Dam Hydroelectric Plant.

Introduction

Growing up in Southern California, I have had to learn all the ways that we can save water - especially in times of droughts. Southern Californians are no strangers to these types of conditions and water restrictions have been implemented many times throughout the years. I remember when the drought intensities increased and not being able to even water my lawn or plants during certain times of the day. If your water use exceeded the limits, you could have been fined or taxed more. In 2015, Governor Jerry Brown "announced actions that will save water, increase enforcement to prevent wasteful water use, streamline the state’s drought response and invest in new technologies that will make California more drought resilient" (Governor Brown).

When rainfalls or snowpack are reduced due to drought conditions, Californians start to use more water from local lakes and reservoirs to help supplement their needs. This may include water for domestic and agricultural use, and even for industrial and power generation. Using up the water in our lakes, on top of what evaporates or flows out in rivers and streams, could cause us to lose the natural water resources that we need for California. One major source of Southern California's water is the Colorado River, that flows through southern Nevada and the Hoover Dam. "The Colorado River supplies roughly a third of all water for Southern California cities and suburbs. It also supports a large farming industry in Imperial and Riverside Counties" (PPIC). The lake that this project focuses on, Lake Mead, is a storage lake formed by the Colorado River and the Hoover Dam. "Most of the water in this great reservoir comes from snowmelt in the Rocky Mountain range and travels through Lake Powell, the Grand Canyon, and into Lake Mead." (Earth Observatory). It supplies water and serves California, Nevada, Arizona, and even Mexico. It is very important to monitor the lake levels, so we know how much water is going in from the Colorado River, and coming out not only due to the normal seasonal evaporations, but also what we use for our daily needs.

This project investigates how Lake Mead’s surface area has changed over time due to climatic and human factors, and whether recovery trends are observable by 2025. It aims to monitor the area of the lake from 2000 to 2025 and taking intervals every 5 years. This project uses Landsat 5 and Landsat 8 imagery, so it uses the surface area of the lake to make the analyses.

Lake Mead, showing evidence of previous water levels.

Image credit: https://www.visitarizona.com/places/parks-monuments/lake-mead/

Methods

In order to begin my analysis, I used Google Earth's Historical Imagery Timelapse function to visually see the effects time has had on Lake Mead. Next, I used Google Earth Engine to clean up and merge a collection of Landsat 5 and Landsat 8 Images [Figure 1]. From there, I analyzed the area of interest by looking at The Modified Normalized Difference Water Index (MNDWI) trends in the form of a Kendall Mann Test [Figures 2 and 3]. MNDWI "enhances open water features while suppressing noise from built-up land, vegetation, and soil" (Mehta & McCartney). This will give me a visual estimate of the change in the open water features over the date range specified. This was chosen to focus on the area of the lake over the Normalized Difference Water Index test since "MNDWI produced better results than the Normalized Difference Water Index" (Xu, 2006). In addition, I ran the same area in a MNDWI as well as a Normalized Difference Vegetation Index (NDVI) test to see if there were any correlations between a reduction in the water body and increase in vegetation in those same areas.

Then, I used R code to clean and export Landsat images of Lake Mead. I used Landsat 5 Images from 2000, 2005, and 2010 and Landsat 8 images from 2015, 2020 and 2025. I chose Landsat 5 instead of Landsat 7 due to a malfunction in one of the sensors, where "the sensor’s line of sight traced a zig-zag pattern along the satellite ground track" which caused gaps in the image itself (Landsat 7). I brought one of these images into ArcGIS in order to train a classifier to classify the rest of the images. In order to ensure that I did not misclassify areas of the image, I kept the number of classes small. According to crop landcover data released by USDA, most of the area around the lake is in the desert and mostly covered by "barren" or "shrublands" (USDA). So I felt that difference between these two landcovers was not necessary for the analysis of this project. I only care about the water in the form of the surface area of the lake, which means for this project I do not need the difference between barren and shrubland. Thus, I combined those two groups.

I then took the training shapefile back into R to run a random trees classification on the images and then a 3x3 majority filter to clean up the noise. I chose random trees classification for its ability to "successfully handle high data dimensionality and multicollinearity, being both fast and insensitive to overfitting" (Belgiu). This allowed me use the classified images to compare the ground cover differences. To analyze these differences I ran them through a difference calculator to see the percent growth or loss. Finally, I created Sankey Diagrams to see whether the ground cover turned into wetlands or more barren and shrublands. "The primary value of a Sankey is visualization of the evolution of a population's status over time through transitions" (Otto et al.).

Figure 1: Cloud Masking in Google Earth Engine

Figure 2: Adding MNDWI and NDVI Bands

Figure 3: Creating the Kendall Mann trends

Results

Analyzing all the different images for 25 years, I noticed a large decrease on the surface area of Lake Mead both visually and while using the classifications to find the numerical area of the lake. Over the full 25 years, the lake has reduced in size by about 15,254 hectares (58.9 Square Miles) of water cover. Breaking that number down into five year intervals, I found that between 2000 and 2005 the lake lost 8279.5 hectares of water, between 2005 and 2010 the lake lost 4,595.9 hectares, between 2010 and 2015 the lake lost 2373.5 hectares, between 2015 and 2020 the lake actually gained 3,424.6 hectares, and between 2020 and 2025 the lake lost 3429.72 hectares of water. The largest decrease of the water level was the 2000 to 2005 interval, and the only increase was the 2015 to 2020 interval. Most of the area of water that was lost turned straight into the shrubland or barren land covers rather than turning into wetlands. In fact, the shrubland and barren land cover type showed growth for every interval except when the lake gained water. This shows that as the lake recedes, the area around the lake does not stay wet or moist for long. The only areas that did change to or from wetlands were the Northern and Eastern sections. These are the areas of the lake that are much shallower than the main portions of the lake.

For initial investigation, using the "JRC Global Surface Water Mapping Layers" (JRC), Figure 4, between the years 1984 and 2019, there shows a large amount of "permanent loss" located around the lake on the shorelines. A small amount of "seasonal to permanent" is attributed to the small changes in the water level due to the local rains, snowmelt, and the filling of the lake from the Colorado River.

Figure 4: Categorical classification of change between first, 1984, and last year, 2019.

Image Credit: Screenshot from GEE using JRC Global Surface Water Mapping Layers (JRC)

Using the Kendall Mann Diagrams created in my analysis, Figure 5 and Figure 6, I noticed a decrease in the MNDWI levels around the lake, but an increase in NDVI levels in a lot of the same areas. Interestingly, this test also shows an increase in the NDVI in the northern and eastern parts of the lake as these are the shallower parts of the lake that generally drain into wetlands first, which then allows more vegetation to grow in the wet soils.

Figure 5: MNDWI Kendall Mann Test in GEE

Figure 6: NDVI Kendall Mann Test in GEE

After applying the MNDWI index to Lake Mead images from 2000 to 2025 with 5 year intervals, the visual change of the size of the lake is much more apparent. The differences are most obvious in the Northern and Eastern arms of the lake. The largest visual difference is between 2005 and 2010, since the upper limit of the lake came down from off the Northern limit of the map to below that map limit. We are able to see that upper shores of the lake from 2010 and on.

Classification and Accuracy Plots of Lake Mead

Plots from R shows the results of the classification of Lake Mead at each 5 year interval from 2000 to 2025. Each plot shows the area of study classified into either Shrubland / Barren, Water, or Wetlands. These images help to highlight the decrease or increase of the area of the lake throughout the years.

Lake Mead 2000

Lake Mead 2005

Lake Mead 2010

Lake Mead 2015

Lake Mead 2020

Lake Mead 2025

Percent Change Analysis and Sankey Diagrams from the Classified Plots

Using the classifications from above, the change in the area of the lake is easy to see. All intervals except 2015 to 2020 show a loss in the the amount of pixels, which were converted into area in hectares. The accompanying Sankey Diagrams show that a lot of the loss in water is converted into shrublands and barren Land cover types.

Land Cover Percent Change from 2000 to 2025

Over the entire time interval, the amount of water had around a 30% loss. Since the wetlands are a small portion of the total land cover, even a small change results in a large percent. We see from the Sankey Diagram below that most of the change in water turns into the shrubland / barren Land Cover.

Land Cover Percent Change from 2000 to 2005

Over this interval, I saw around a 19% decrease in water and an increase in both the shrubland / barren as well as the wetlands. This is evident in the changes in the Northern and Eastern segments of the lake, where the elevation changes on the shorelines is much smaller. This allows more of the wetland to form on those gentler slopes.

Land Cover Percent Change from 2005 to 2010

Here, the water shows about a 10-12% decrease, and another large decrease in the wetlands.

Land Cover Percent Change from 2010 to 2015

Here, again, we see more drying of the wetlands from the previous intervals. The water continued to decrease, losing another 9-10%.

Land Cover Percent Change from 2015 to 2020

This interval shows the only positive gain of water in the lake during the entire study period. The lake showed about a 9% change in area.

Land Cover Percent Change from 2020 to 2025

This interval is almost the inverse of the previous interval. After the gain in water from the previous interval, this one loses all and a bit more water than was gained. We saw a loss of about 10% in the water.

Discussion

The results of my analysis shows a large decrease in the area of Lake Mead between the years 2000 and 2025. As the climate warms and the need for water in human activities increases during times of drought, the water in the lake is being used faster than it can recover. The five year interval that saw the largest change in surface area was 2000 to 2005 with a 19% decrease. This decrease is more evident in the changes in the Northern and Eastern segments of the lake because the elevation changes on the shorelines is much more gradual. This allows more of the wetland to form on those gentler slopes. This also shows up in the NDVI test that showed a positive trend, suggesting a growth in vegetation for that area. According to Drought.Gov, around 60% of Nevada was experiencing extreme drought conditions, which shows a correlation with the loss of water. During the 2005 to 2010 interval, the lake experienced another 10% decrease in area. This interval also shows that the wetlands created in the previous interval are starting to dry further, creating more shrublands / barren land covers. Interestingly, some of the wetlands have changed into water, but this could have been a misclassification from the algorithm in one of these two plots. These 10 years have illustrated the greatest change in the lake itself.

Between 2010 and 2015, the lake lost another 9-10%. This period of time started off with light drought conditions, but by 2015, the drought had increased again, contributing to the loss of water (Drought.gov). There was also a 50% decrease in the wetlands which were further drying into shrublands / barren land cover. The only period of time where the water gained in size was 2015 to 2020. This is supported by the data that during that period of time, the Colorado River experienced "an above average snow season in 2019" and "recent rains in the Lower Basin, which reduced Colorado River water uses and increased tributary flows into Lake Mead" (Collum, 2024). During this period, precipitation was higher than in previous intervals. According to the Bureau of Reclamation, the lake was "refilled enough by the end of 2016 to avoid cuts in water deliveries in 2017" (NASA). Unfortunately, in the last interval, 2020 to 2025, Nevada was subjected to more extreme drought conditions. This time, however, around 40% of the state was under "exceptional drought" conditions (Drought.gov). These conditions finally ended in 2023 where Nevada experienced another water year. Despite this, Lake Mead experienced another 10% decrease in area. Currently, Southern Nevada and Lake Mead is experiencing strong drought conditions. According to the Southern Nevada Water Authority, "since 2000, snowfall and runoff into the basin have been well below normal." They describe the current conditions "the worst drought in recorded history" (Southern Nevada Water Authority).

Image Credit: Drought.gov, https://www.drought.gov/states/nevada

Extended research could also be done on how much water is being drawn out from the lake for each state's needs. This would give a better idea of how much of the lake's area is being reduced due to just the drought conditions, and how much is being pulled out for human activities. More studies could be done that specifies more on the land cover types and soil types around the lake itself. This may lend more information to exactly what type of soil the receding water leaves behind. This may also give more exact data on water level changes. Reducing the size of the study area by cropping out the large areas of shrubland / barren could also lead to better Sankey Diagrams and percentage change plots by bringing each of these land types closer in area. Since we are seeing large change percentages in the wetlands and small change percentages in the shrublands / barren, it could lead to having the percentages be much more meaningful and verifiable. Further investigations can alse be done to correlate the area of the lake with the actual depth of the lake. Current information from the US Bureau of Reclamation that as of April 2025, the lake has a water depth of 1062.23 feet (Bureau of Reclamation), about 30 feet lower than the end of 2020 and 15 feet lower than 2015 levels. This could help further support and explain the amount of water loss in the lake.

Conclusion

Lake Mead is a valuable resource for many in California, Nevada, Arizona, and Mexico. Keeping a note on how much the lake fills naturally and how much of the water human activities are using should be very important. Analyzing the surface are of the lake shows a drastic decrease in the lake area, which is linked to the drop of available water in the lake. The most drastic drop in surface are was between the years 2000 and 2005. The only years that the lake saw an increase in area and water level was between 2015 and 2020. This increase was likely due to above-average precipitation in the Colorado River Basin in 2019. However, even with the increase of the lake levels, the following 5 years have overtaken that positive gain to bring the lake levels even lower. If human activities do not change the way we use our water, and where we get our water sources, we may end up losing the lake entirely over time. This will cause a water shortage in the agricultural efforts in the southwest states, as well as the loss of energy resources through the Hoover Dam. Ongoing monitoring of Lake Mead’s water levels and usage will be crucial for effective water management in the Southwest.

References

Amita , Mehta, and Sean McCartney. “Mapping and Monitoring Lakes and Reservoirs with Satellite Observations.” NASA Applied Sciences, 9 Feb. 2021, appliedsciences.nasa.gov/sites/default/files/2021-03/Lake_Webinar_Week1_v6.pdf.
Belgiu, Mariana, and Lucian Drăguţ. “Random Forest in remote sensing: A review of applications and Future Directions.” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 114, Apr. 2016, pp. 24–31, https://doi.org/10.1016/j.isprsjprs.2016.01.011.
Bureau of Reclamation, Lower Colorado Region Web Team. “Lower Colorado River Operations.” Lower Colorado River Operations | Lower Colorado Region | Bureau of Reclamation, May 2025, www.usbr.gov/lc/region/g4000/hourly/mead-elv.html.
“California’s Water: The Colorado River.” PPIC.ORG/WATER, Public Policy Institute of California, 1 Nov. 2018, www.ppic.org/wp-content/uploads/californias-water-the-colorado-river-november-2018.pdf.
“Google Earth Engine Platform Documentation.” Google Earth Engine | Google for Developers, Google, developers.google.com/earth-engine. Accessed 8 May 2025.
“Governor Brown Directs First Ever Statewide Mandatory Water Reductions.” California.Gov, Office of Governor Edmund G Brown Jr, 1 Apr. 2015, archive.gov.ca.gov/archive/gov39/2015/04/01/news18913/index.html.
“JRC Global Surface Water Mapping Layers, v1.2 | Earth Engine Data Catalog | Google for Developers.” Google, Google, developers.google.com/earth-engine/datasets/catalog/JRC_GSW1_2_GlobalSurfaceWater. Accessed 8 May 2025.
“Landsat 7.” USGS, www.usgs.gov/landsat-missions/landsat-7. Accessed 10 May 2025.
McFeeters, S. “The use of Normalized Difference Water Index (NDWI) in the Delineation of Open Water Features.” International Journal of Remote Sensing, 1996.
Mehta, Amita, and Sean McCartney. “Mapping and Monitoring Lakes and Reservoirs With ...” NASA Applied Sciences, 9 Feb. 2021, appliedsciences.nasa.gov/sites/default/files/2021-03/Lake_Webinar_Week1_v6.pdf.
“Modified Normalized Difference Water Index (MNDWI).” Miscellaneous Indices Background, NV5 Geospatial Software, www.nv5geospatialsoftware.com/docs/BackgroundOtherIndices.html. Accessed 8 May 2025.
“Nevada.” Drought.Gov, www.drought.gov/states/nevada. Accessed 10 May 2025.
News, Know Your Water. “Lake Mead Ends 2019 above 1090’ – but 2020 Still Brings Tier Zero Declaration.” Central Arizona Project, 10 Sept. 2024, knowyourwaternews.com/lake-mead-ends-2019-above-1090-but-2020-still-brings-tier-zero-declaration/.
Otto, Ethan, et al. “Overview of sankey flow diagrams: Focusing on symptom trajectories in older adults with advanced cancer.” Journal of Geriatric Oncology, vol. 13, no. 5, June 2022, pp. 742–746, https://doi.org/10.1016/j.jgo.2021.12.017.
“Storage Capacity of Lake Mead.” National Parks Service, U.S. Department of the Interior, www.nps.gov/lake/learn/nature/storage-capacity-of-lake-mead.htm. Accessed 8 May 2025.
“United States Department of Agriculture.” USDA, www.nass.usda.gov/Research_and_Science/Cropland/sarsfaqs2.php#what.5. Accessed 8 May 2025.
“Visualizing the Highs and Lows of Lake Mead.” NASA, NASA, earthobservatory.nasa.gov/images/88099/visualizing-the-highs-and-lows-of-lake-mead#:~:text=Lake%20Mead%20is%20now%20roughly,lowest%20level%20since%20the%201930s. Accessed 8 May 2025.
“Water Shortages.” Water Shortages, Southern Nevada Water Authority, www.snwa.com/water-resources/drought-and-shortage/index.html#:~:text=The%20Colorado%20River%20Basin%20is,Lake%20Mead%20and%20Lake%20Powell. Accessed 10 May 2025.
Xu, H. “Modification of Normalised Difference Water Index (NDWI) to Enhance Open Water Features in Remotely Sensed Imagery.” International Journal of Remote Sensing, vol. 27, no. 14, 2006, pp. 3025–3033.

Page updated

Google Sites

Report abuse