www.nature.

com/scientificdata

OPEN Global Dam Tracker: A database

Data Descriptor of more than 35,000 dams with
location, catchment, and attribute
information
1✉
Alice Tianbo Zhang & Vincent Xinyi Gu2

We present one of the most comprehensive geo-referenced global dam databases to date. The Global
Dam Tracker (GDAT) contains 35,000 dams with cross-validated geo-coordinates, satellite-derived
catchment areas, and detailed attribute information. Combining GDAT with fine-scaled satellite data
spanning three decades, we demonstrate how GDAT improves upon existing databases to enable the
inter-temporal analysis of the costs and benefits of dam construction on a global scale. Our findings
show that over the past three decades, dams have contributed to a dramatic increase in global surface
water coverage, especially in developing countries in Asia and South America. This is an important step
toward a more systematic understanding of the worldwide impact of dams on local communities. By
filling in the data gap, GDAT would help inform a more sustainable and equitable approach to energy
access and economic development.

Background & Summary

As one of the oldest forms of man-made infrastructure, dams have been integral to economic development
throughout human history. They are built to control floods, irrigate crops, supply water, generate electricity, and
ease navigation. Proponents of dams often praise them as a source of low-carbon electricity. Hydropower gen-
erated 16 percent of the world’s total electricity and 60 percent of all renewable electricity in 20191,2. Estimated
to reduce annual emissions by about 2.8 billion tons of carbon dioxide equivalent (CO2e), hydropower is often
seen as the backbone for achieving the goals of affordable and clean energy for all (SDG 7). Developing countries
in Asia, South America, and Africa possess significant untapped hydropower potential3. Globally, at least 3,700
large hydropower plants with a capacity of more than 1 MW and 82,800 smaller plants are operating, under
construction, or being planned4,5.
But harnessing the power of the river comes with concentrated costs. The construction of dams, especially
large dams and those located in transboundary river basins, has wide-ranging socioeconomic, geopolitical, and
environmental implications. The International Commission on Large Dams (ICOLD) defines a large dam as “a
dam with a height of 15 metres or greater from lowest foundation to crest or a dam between 5 metres and 15 metres
impounding more than 3 million cubic metre.” Large dams, such as the Three Gorges Dam in China, displace
millions of people globally and deprive entire communities of their cultural heritage6,7. The displaced rarely
receive adequate compensation and often suffer from an enduring loss of land, job, and wealth long after the
dams are built8. By modifying and fragmenting rivers, dams also exacerbate water scarcity9–11 and worsen geo-
political tensions12,13. As these socioeconomic and geopolitical costs are often borne by marginalized and indig-
enous people14 near the dam catchment area, dams could exacerbate existing social and environmental injustice.
Beyond the socioeconomic and geopolitical implications, dams affect the ecological functioning of river sys-
tems by reducing the downstream transfer of nutrients15 and threatening the natural habitat of freshwater mega-
fauna16, especially migratory fish species17,18. A third of the world’s freshwater fish species are at risk due to dam
construction booms in some of the most biodiverse river basins, such as the Amazon, Congo, and Mekong19,20.
In addition, some dams may intensify rather than mitigate global warming, as these man-made waterbodies
reduce surface albedo and result in a positive radiative forcing21. Climate change mitigation requires a rapid and

1
Washington and Lee University, Department of Economics, Lexington, Virginia, 24450, USA. 2Georgia Institute of
Technology, School of Public Policy, Atlanta, Georgia, 30332, USA. ✉e-mail: [email protected]

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 1

www.nature.com/scientificdata/ www.nature.com/scientificdata

systematic transformation of the energy system toward renewable sources. As such, understanding the costs and
benefits of dam construction is crucial for energy and climate policy.
Quantifying the net benefit of dams requires a comprehensive database with three key characteristics:
1) accurate geo-coordinates of dams, 2) information on completion year, and 3) global in coverage. Existing
databases of dams either lack necessary spatial and temporal detail or are limited in scope. Two recently pub-
lished global dam databases, GOODD22 and GeoDAR23, make significant contributions to geocode and validate
dam and reservoir locations. However, neither database contain publicly available dam attribute information,
limiting their use for inter-temporal studies. To fill this gap, we compile a new global database of dams, the
Global Dam Tracker (GDAT), by consolidating existing global dam datasets with country- and region-specific
data from nonprofit organizations, academics, and governments24. Building upon existing global and regional
dam databases, including AQUASTAT, the Global Reservoir and Dam Database (GRanD), and the World
Resources Institute (WRI) database, we conduct extensive cross-referencing and manual validation to fill in
data gaps and correct erroneous dam attributes. More than 90% of dams in GDAT are geocoded (n = 31,780),
and the coordinates are extensively verified using Google Earth and other geospatial software. Beyond location,
GDAT contains detailed attribute information for each dam, such as completion year, purpose, height, length,
and installed capacity. As such, GDAT is one of the most comprehensive geo-referenced global dam databases
with catchment and attribute information to date, especially for the Global South.
To allow for inter-temporal analysis of the impact of dam construction, we develop an algorithm that uses
GDAT and various state-of-the-art satellite data products to obtain reservoir and catchment areas associated with
dams in GDAT. To demonstrate the use case of GDAT data, we calculate the cumulative change in global surface
water from dam construction and find that dams have substantially altered the location and persistence of sur-
face water around the world. Our results show a cumulative increase of approximately 50,000 km2 of seasonal
and permanent water in dam catchment areas between 1984 and 2018.
The GDAT database could be used for a systematic and global analysis of the impact of dams on local com-
munities. As the urgency of climate change calls for a transition from fossil fuels, hydropower could usher in a
cleaner and more sustainable energy system as mandated by SDG 7. But these benefits should be weighed against
the negative social, geopolitical, and environmental costs. Therefore, a global dam database that improves our
understanding of the net impact of dams is crucial for informing a more sustainable and equitable approach to
economic development.

Methods
In this section, we present an overview of the data collection process and demonstrate an application where GDAT
is used to calculate dam-induced global surface water changes. The first subsection, “Global Dam Tracker (GDAT),”
presents the GDAT database and describes the data sources. The second subsection,“GDAT Use Case,” details our
application of GDAT and outlines an algorithm for calculating surface water changes within dam catchment areas.

Global dam tracker (GDAT). The Global Dam Tracker (GDAT), a database with 35,140 dams in all conti-
nents except Antarctica, is an original data compilation effort24. We build upon existing global and regional dam
databases, including AQUASTAT (www.fao.org/aquastat), the Global Reservoir and Dam Database (GRanD -
globaldamwatch.org/grand)25, and the World Resources Institute (WRI) database (datasets.wri.org/dataset/
globalpowerplantdatabase)26. For each country, we consolidate these global and region databases with available
data from government agencies, academics, and nonprofits. More than 90% of dams in GDAT are geocoded
(n = 31,780), and the coordinates are extensively verified using Google Earth and other geospatial software. This
process creates an extensive cross-validated database that can be used for analysis at various geographical scales.

Description of regional data sources. This section provides an overview of major regional or national data
sources that contributed substantially to GDAT. Detailed information on the data sources for each country
is available in the Region Highlights section. Existing dam databases have limited coverage of developing
and low-income countries. As such, our data collection effort focuses on countries in Africa, Asia, and South
America, where we collected primary data from administrative and other sources. Data for countries in Europe,
North America, and Oceania are mainly obtained from existing global dam databases.
Dams in India were collected from the India Water Resources Information System (WRIS), a database that
is maintained and funded by the government of India (indiawris.gov.in/wris). Reservoir volume data are not
provided by the WRIS and, hence, are not currently available in GDAT for dams in India.
Dams in China were compiled from several nongovernmental sources: AQUASTAT, GRanD, International
Rivers, and the Consultative Group for International Agricultural Research (CGIAR). While the Ministry of
Water Resources of the People’s Republic of China maintains a database on dams in China, it is not released to
the public. Hence, even though the GDAT database has compiled more than 1,000 dams in China, the actual
dam count is expected to be much higher, given that many smaller dams might be missing from the database.
For Continental Southeast Asia, GDAT database builds upon three main data sources: AQUASTAT, GRanD,
and CGIAR Greater Mekong (wle-mekong.cgiar.org). The CGIAR database focuses on the major river basins
of continental Southeast Asia (Mekong, Red, Salween, Irrawaddy) and contains information for more than 800
dams that span China, Thailand, Vietnam, Laos, Cambodia, and Myanmar.
Dams in the Middle East came from AQUASTAT and GRanD, except Turkey and Iran, where we obtained
data from government websites. The Turkish General Directorate of State Hydraulic Works (www.dsi.gov.tr)
publishes information for more than 600 dams, and the Iran Water Resources Management Company (www.
wrm.ir) publishes information for more than 1,000. Neither database, however, is geolocated. Hence, the geo-
graphic coordinates for these two countries are manually verified through GeoNames and Google searches.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 2

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 1 Map of dams in the Global Dam Tracker database. (a) Locations of all dams in the GDAT database, with
each dam represented by a blue point. Large concentrations of dams can be found in the United States, Brazil,
India, South Africa, Europe, and East and Southeast Asia. (b) The catchment area of dams delineated using
drainage flow directions.

Dams in Brazil were compiled from four sources: AQUASTAT, GRanD, the Brazilian National Dam Safety
Information System (SNISB), and Dams in Amazonia (archive.internationalrivers.org). The SNISB for 2016, the
most recent version available, includes almost 23,000 dams. We excluded all dams and reservoirs used exclu-
sively for tailings and hazardous waste storage. The SNISB includes structures with a height greater than or equal
to 15 meters and a total reservoir capacity greater than 3hm3. As a result, the GDAT database is likely missing a
considerable number of small dams in Brazil that are not monitored by the SNISB.
While most of the new data in GDAT came from the developing world, especially Africa, Asia, and South
America, we also cleaned and consolidated data for Europe, Oceania, and North America from existing global
databases and government sources. Data for Oceania mostly came from AQUASTAT and GRanD, except
for Australia, which was collected from the Register of Large Dams maintained by the Australian National
Committee on Large Dams. The GDAT database for Europe contains information from AQUASTAT, GRanD,
the WRI Global Power Plant Database, and reports for more than 3,000 dams in Europe. For North America,
data were compiled from AQUASTAT and GRanD, except for the United States, which included data from
the United States Geological Survey (USGS). USGS lists more than 8,000 dams, dikes, levees, and other
water-engineering structures. We filtered the USGS data to remove non-dam structures.

Description of dam attributes. In addition to greater spatial coverage, the GDAT database contains detailed
attribute information for each dam. These attributes include basic identifying information such as the dam
name and alternate names; geographic attributes such as country, province, river basin, latitude, and longitude;
dam characteristics such as completion and construction years, height, and length; reservoir capacity and area;
purposes; and energy-generating capacity for hydroelectric dams. These attributes served as the basis for data
collection, and any information found in regional and country-specific sources was matched to these attributes
to be included in GDAT in a consistent and standardized format.
Dams are largely concentrated along major river basins in Asia, Africa, Europe, and the Americas (Fig. 1 and
Table 1). Asia has the highest number of dams completed to date, with 9,526 dams, or 27 percent of worldwide dam
construction. North America and South America also have significant dam infrastructure, representing 23 percent
and 21 percent of the global dam count, respectively (Table 1). In terms of installed capacity, Asia and South

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 3

www.nature.com/scientificdata/ www.nature.com/scientificdata

Continent
Asia North America South America Africa Europe Oceania World
Total dam count 9,526 8,333 7,566 6,243 3,201 271 35,140
Count by main purpose
Irrigation dams 5,627 1,130 1,607 253 241 — 8,858
Hydropower dams 1,255 1,553 1,372 345 2,458 121 7,104
Water supply dams 395 1,147 1,118 149 225 136 3,170
Flood control dams 390 1,748 53 18 63 — 2,272
Recreation dams 4 1,323 114 — 8 — 1,449
Other/not specified dams 1,819 1,139 2,894 5,223 198 14 11,287
Key dam attributes
Median completion year 1981 1963 1990 1981 1966 1966 1974
Mean dam height (m) 33 25 18 14 57 37 26
Mean dam length (m) 1,014 1,335 817 391 1,129 No data 1,036
Total installed capacity (MW) 468,681 85,173 193,496 No data 175,258 4,553 927,161
Total reservoir volume (km3) 1,091 13,477 1,179 No data 1,392 17 17,155
Total reservoir area (km2) 43,264 316,058 10,501 No data 146,772 1,169 517,763

Table 1. Summary statistics of dam attributes in GDAT.

Africa Asia Europe

800

600

400
Number of Observations

200

0
North America Oceania South America
800

600

400

200

0
1900 1930 1960 1990 2017 1900 1930 1960 1990 2017 1900 1930 1960 1990 2017

Completion Year

Fig. 2 Distribution of dam completion year by continent.

America account for 50 percent and 20 percent of the global total installed capacity, respectively, while Europe
and North America account for 18 percent and 9 percent, respectively (Table 1).
There are notable differences in the distribution of dam completion years across continents. While most
developed countries in North America, Europe, and Oceania have been witnessing a decline in dam construc-
tion since the 1970s, developing countries in Africa, Asia, and South America have been experiencing a contin-
ued increase in dam construction (Fig. 2). Most notably, the Yangtze basin in China, the Ganges-Brahmaputra
basin in South Asia, and the Amazon basin in South America have many dams currently planned or under con-
struction4. A time series of completion year (Fig. 3) demonstrates a significant acceleration in dam construction
from 1970 to 2000, followed by a slight deceleration over the past decade. Global investments in hydropower
peaked in the second half of the twentieth century in response to a growing desire to diversify energy sources
and reduce dependency on fossil fuels5.
The most common main purposes of dams in GDAT are irrigation and hydroelectricity, which represent 25 percent
and 20 percent of the data (Fig. 4). Additional main purposes include water supply, flood control, recreation,
livestock, and navigation. 32 percent of dams had the main purpose classified as “other” due to limited or miss-
ing information in the primary sources.
In Asia and South America, irrigation is the most common main purpose of dams (Fig. 5). In comparison,
the most common main purposes in North America, Europe, and Oceania are flood control, hydroelectric-
ity, and water supply/storage, respectively. The number of dams shown under each main purpose reflects the
primary and secondary sources used in the data collection. For example, Europe’s high percentage of dams

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 4

www.nature.com/scientificdata/ www.nature.com/scientificdata

1
North America
Europe
Oceania
.8 Africa
Asia
South America

Cumulative Share
.6

.4

.2

0
1900 1930 1960 1990 2017

Completion Year

Fig. 3 Cumulative distribution of dam completion year by continent.

12,000
11,287

8,858
Number of Observations

8,000
7,104

Irrigation
Hydroelectricity
Water Supply
4,000 Flood Control
3,170
Recreation
2,272 Livestock
1,449 Multi-Purpose
Navigation
411
116 197 276 Fisheries
0

Fig. 4 Distribution of main purpose of dams in GDAT – Overall.

used for hydroelectricity is partly due to the high concentration of hydroelectric power plants in the WRI data.
Additional sources will need to be consulted to provide a more complete representation of the dam main pur-
poses by region, particularly for Europe and Oceania.

GDAT Use Case: Calculating Dam-Induced Global Surface Water Changes. To demonstrate how
GDAT would improve our understanding of the global impact of dam construction, we use GDAT to calculate
dam-induced global surface water changes over the past three decades. Following prior work22,27, we apply an
algorithm (Fig. 6a) that uses GDAT and various state-of-the-art satellite data products to analyze temporal surface
water dynamics around dam catchment areas. The algorithm has five main steps.
First, to ensure that dams fall precisely on a river network, we snap all 31,780 geocoded dams in GDAT onto
the closest river lines using a fine-scale global river network dataset at 15 arc-second resolution, HydroRivers.
HydroRivers provides vectorized line network of all global rivers that have a catchment area of at least 10 km2 or
an average river flow of 0.1 cubic meters per second, or both. It encompasses 8.5 million individual river reaches
with an average length of 4.2 km, representing 35.9 million km of rivers globally27. This step ensures the dam
coordinates are exactly aligned with the river network.
Second, we use the USGS Global Multi-Resolution Terrain Elevation 2010 data (GMTED 2010) at 7.5
arc-second resolution to generate the drainage flow direction at each dam site. GMTED 2010, developed by the
USGS and the National Geospatial-Intelligence Agency (NGA), is an enhanced global elevation data of choice
that incorporates the best available data sources28. The GMTED 2010 data represent a significant improvement
in consistency and vertical accuracy over existing global elevation data, such as GTOPO30.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 5

www.nature.com/scientificdata/ www.nature.com/scientificdata

6,000

Number of Observations
4,000

Fisheries
Flood Control
Hydroelectricity
Irrigation
2,000 Livestock
Multi-Purpose
Navigation

Recreation
Water Supply
0
North America Europe Oceania Africa Asia South America

Fig. 5 Distribution of main purpose of dams in GDAT – By continent.

Third, with the PCRaster library in Python, we delineate the catchment area representing the buffer zone
upstream of the dam location containing its reservoir. The PCRaster code assigns a flow direction value to each
pixel in the GMTED elevation data, which it does by analyzing the values of the eight surrounding pixels. The
code then retraces the flow of water starting from each dam by moving upstream until there are no longer any
more upstream pixels or until the algorithm hits another dam29. The area covered by the retraced pixels is the
catchment area for the dam, which is the area we use to calculate surface water changes caused by the dam. In
total, we obtain 28,460 catchment areas (Fig. 1b) corresponding to dams in GDAT. A small subset of GDAT
dams does not have catchment areas due to limitations in the elevation and river network data.
Calculating surface water change requires information on the dam completion year. Since more than 70% of
GDAT dams have known completion years, we use these as training data for an imputation procedure on dams
with missing information (Fig. 6b). Around the year of dam completion, we expect to see a significant change
in the upstream catchment area, such as in Fig. 7. Therefore, as the fourth step, we impute the completion year
for 8,342 dams by detecting trend breaks in the catchment area time series. The satellite-derived surface water
coverage between 1984 and 2019 comes from the Global Surface Water (GSW) V1.1 dataset available through
Google Earth Engine. Developed by the European Commission Joint Research Centre, GSW is a globally con-
sistent, validated dataset that shows the spatial and temporal distribution of global surface water over the past
three decades using three million Landsat satellite images30.
In the fifth and final step, we calculate global surface water changes induced by dam construction. From GSW,
we obtain pixel counts for the “no water,” “seasonal water,” and “permanent water” categories as annual time
series. We then use the completion year information to calculate the median pre- and post-dam pixel counts.
We choose the median as the summary measure to minimize idiosyncratic measurement errors from climatic
or technical factors that may affect satellite image quality. We quantify the temporal surface water dynamics for
11,710 dams completed between 1985 and 2019. We drop dams without pre- or post-completion observation,
i.e., those built before 1984 or after 2019. Below, we provide additional details about the methodology.

Dam catchment areas. We combine GDAT with elevation data to calculate the catchment area that contains the
reservoir of each dam. To trace the drainage flow direction, we use GMTED 2010 with a resolution of 250 meters.
The elevation data allows us to calculate how water would flow between each pixel based on their relative upstream
or downstream location, similar to GOODD22. We first employ PCRaster, an environmental modeling software,
to obtain catchment areas for each dam from elevation data29. We use the local drainage direction function (ldd)
to assign flow directions to each pixel based on the lowest elevation of the eight surrounding pixels. This allows us
to obtain a flow-direction file from elevation data. We then use the subcatchment function, which uses each
dam in GDAT as a starting point and traces all possible trajectories that water could flow toward the dam. If there
is another dam along the path of the water trajectory, the function stops, as all further upstream points would fall
within another dam’s catchment area. We are hence able to capture all upstream points that eventually drain to the
point of each dam, enabling us to obtain the catchment areas for the majority of dams in our database.
Data on flow direction contains noise from the slopes surrounding the river, as water might flow in a different
direction than the river itself22. The pixels where the greatest flow accumulation occur (the path tracing the bot-
tom of the riverbed) are the only locations that accurately reflect the river’s flow. As a result, dams that are even
slightly misplaced relative to the main river path could result in large errors in the catchment area. To address
this, we align dam locations with river network data from HydroRIVERS to ensure that dams fall precisely on
the main river path where flow accumulation occurs. HydroRIVERS captures all pixels with an accumulated
upstream catchment of more than 10 km2, producing a connected network of lines representing accumulated
river flows31. We use the interpolate function from the Shapely library in Python to snap the location of
each dam onto its nearest river network32.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 6

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 6 Conceptual diagram of the algorithm. We develop an algorithm to estimate the changes in global surface
water from dam construction. (a) To calculate the difference in surface water coverage before and after dams are
built, we snap each dam to a river network to correct for its location, calculate the drainage direction, and delineate
its catchment area. (b) Procedures for imputing the completion years of dams when the information is missing.
This allows the analysis of surface-water changes to be expanded beyond dams with known completion years.

The vast majority of dams in the GDAT database are located very close to a HydroRIVERS river network line,
with 30,471 dams (or 95.9 percent) falling within 0.5 degrees of their snapped location. The remaining 1,309
dams were snapped at a distance of more than 0.5 degrees. Some of them were located on islands not covered
by HydroRIVERS, causing them to be snapped onto a continent. Hence, dams with an interpolated distance of
greater than 0.5 degrees may be inaccurate. This is due to limitations in the data resolution, as 0.5 degrees is the
resolution of the runoff and discharge layers used to calculate HydroSHEDS33. In those cases, we discard the
snapped location and instead use the original geo-coordinates. Our final catchment-area calculation procedure,
therefore, consists of three stages: obtaining drainage direction from elevation data via the ldd function from
PCRaster, snapping each dam onto its respective river network via Shapely, and using the drainage direction
and interpolated location to derive a catchment area for each dam. The output is a raster file of dam catchment
areas, which are used to understand how each dam contributed to changes in surface-water coverage.

Calculating surface water changes within dam catchment areas. We calculate the change in surface water cov-
erage within each catchment before and after dam completion. This allows us to understand how the dam has
inundated the surrounding land. We use data from the Global Surface Water (GSW) Explorer developed by the
European Commission Joint Research Centre, which contains surface water coverage data between 1984 and
201930. The dataset divides the world into pixels coded as 1 for no coverage, 2 for seasonal coverage, and 3 for
annual coverage. We mask all pixels that are coded as 0 (no data). Not all years contain robust data due to cloud
cover and satellite-imaging anomalies. Therefore, for each dam, we obtain the median value of all pre-dam
pixels within its catchment area and create a single snapshot of the median pre-dam water coverage. We do
the same for the post-dam pixels (inclusive of the year of completion), obtaining a single post-dam snapshot
of the median water coverage. Due to the masking of no-data pixels, some pixels may have fewer years of data.
The possible median-value pixels include no water (1), varying values of seasonal water (1.5, 2, 2.5), and

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 7

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 7 Change in surface water coverage for three exemplary dams. Dam completion year in parenthesis.
(a) Construction for Ataturk Dam in Turkey was completed in 1990 and the reservoir was filled in 1992.
A comparison of the pre-dam (left) and post-dam (right) images show a clear increase in water pixels after
the completion, a total increase of 983,482 km2. The pre-dam image exhibits a partially filled reservoir,
demonstrating that large dams may take an extended time to fill. (b) Luis Eduardo Magalhaes Dam in Brazil
was completed in 2001. The formerly free-flowing section of the river is visible in the pre-dam image, while the
dam flooded an area of 660,130.5 km2. (c) Mohale Dam in Lesotho, with an imputed completion year of 2004,
illustrates the logic behind our imputation procedure. The actual year of construction is 2002, and the dam was
formally commissioned in 2003–4. After completion, the dam flooded an area of 15,884.5 km2.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 8

www.nature.com/scientificdata/ www.nature.com/scientificdata

permanent water (3). Once we have the median-value snapshots of pre-dam and post-dam water coverage
within the buffer zone of each dam, we calculate the difference in the amount of area covered by both seasonal
and permanent pixels. Given that GSW data spans 1984 and 2019, pre- and post-dam comparisons are only
possible for dams with catchment data and completion years that fall within the GSW time range. This yields us
5,508 dams for which we can calculate water-coverage changes.

Imputation of dam completion years. Calculating surface water changes using dams with available
completion-year data limits us to only 5,508 dams out of more than 30,000 in GDAT. To expand our analysis
to more dams, we developed a machine-driven imputation procedure that calculates the dam’s completion year
based on surface-water data from GSW. Because the imputation procedure requires deriving completion years
from surface-water changes, we can only apply it to dams with catchment data but without recorded completion
years. This totals 8,357 observations. Using GSW data, we obtain the annual pixel counts within the catchment
area of each dam for each year between 1984 and 2019. The result of this step is a data frame of 8,357 dams span-
ning 36 years, with counts of no-water (1), seasonal-water (2), and permanent-water (3) pixels for each year.
Years that contain missing data are filled in with data from the previous year or the most recent year with data.
We use two methods to calculate the imputed completion year. The first method identifies the year with
the largest pixel change as the imputed completion year. We conduct four runs for this method: using 1 pixels
only (no water), 2 pixels only (seasonal water), 3 pixels only (permanent water), and 2 + 3 pixels (any water).
For each run, we obtain the pixel difference between the current and prior years for each year within the GSW
data, which yields 35 observations of year-to-year differences. This yields us a total of four imputation results
using the largest-change method. The second method involves the calculation of the structural break using the
Wald test. A structural break occurs when the trend of a time series abruptly changes34–36. We apply the STATA
estat sbsingle test, which assumes a single structural break with an unknown break date in the data37–40.
This is appropriate for estimating the completion year of a dam when the year is unknown, given that the struc-
tural break in the water-coverage trend would occur when the dam is completed. As with the first method, we
apply the structural-break test with four runs: 1 pixel only (no water), 2 pixels only (seasonal water), 3 pixels
only (permanent water), and 2 + 3 pixels (any water). This yields four more imputation results for each dam. The
above two methods combined yield a total of eight imputation results. The most frequent (mode) year among the
eight imputations is the final calculation for the imputed year. If there are multiple modes, then the earliest mode
is taken. Taking earlier years would account for the fact that it takes time for dams to fill up after completion.

Use case results: dam-induced global surface water changes. Dams have substantially altered the location and
persistence of surface water around the world. Figure 8a shows a cumulative increase of at least 49,715 km2 of
seasonal and permanent water in dam catchment areas between 1984 and 2018. This is likely a lower-bound
estimate due to the absence of satellite data before 1984 and the lack of completion year information for all dams.
The fastest dam-induced surface water expansion has occurred in Asia (Fig. 8b), particularly from the rapid pace
of dam construction in China and India. Growth is slower but still considerable in South America and Africa,
where hydroelectric dams power economic development in Brazil, Venezuela, Zambia, and D.R. Congo. North
America witnessed a large number of dams at the start of the sample period, but few new dams were completed
after 1984 (Figs. 2,3). Similar to North America, dam construction peaked in Europe around the 1960s, and
surface water coverage within dam catchment areas has remained steady since 1984.
Comparing the change in seasonal and permanent water before (pre-dam) and after (post-dam) dam completion
(Fig. 8c,d) gives us insight into the impact of dam construction on surface water dynamics. In Asia, post-dam catch-
ment areas witnessed the largest surface water increase by more than 23,400 km2 (Table 2). This is followed by South
America and Africa, with more than 7,000 km2 of additional surface water due to dam construction. We also observe
an increase in the total pixel count globally and a pronounced increase in Asia as the spatial and temporal resolutions
of satellite data there have improved substantially over the past two decades. This is a data quality limitation for
quantifying the long-term surface water dynamics using satellite data–a caveat documented in the GSW dataset30.
We map changes in global dam catchments by country to understand the spatial distribution of surface water
dynamics (Fig. 9). The continental summaries mask considerable spatial heterogeneity across countries. The
largest contributors to dam-induced surface water change come from developing countries with rapid hydro-
power installation, such as China and Brazil. China alone accounts for more than half of the total surface water
increase in Asia. In Brazil, 3,459 dams contribute to more than 80% of the water expansion in South America.
Developed countries with historically large inventories of dams, such as Canada and the United States, also see
a significant rise in surface water. The expansion in surface water, however, may not be a universal feature of
dams. In a few countries, most notably Zambia, Ireland, and the United Arab Emirates, surface water may have
decreased slightly after dams are built. We cannot rule out the possibility that the decline is due to idiosyncratic
changes in satellite data quality or measurement error, as we have found the surface water loss to be less than
5 km2 in these three countries. Echoing prior findings30, these results suggest that losses in permanent and sea-
sonal water from dams may be more concentrated than gains.

Data Records
The GDAT database is available as a zip folder (GDAT_v1.zip) on Zenodo41. Data records are provided as two ESRI
shapefiles. The first file contains a point layer for dam locations (GDAT_v1_dams.shp). The second file contains
a polygon layer for dam catchment areas (GDAT_v1_catchments.shp). Both shapefiles use the World Geodetic
System 1984 (WGS84) datum. Dam attributes are available in both the point and polygon layers. In addition, a
detailed codebook (GDAT_v1_documentation.xlsx) provides the definition of each variable in the attribute table.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 9

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 8 Summary of changes in water coverage around global dam catchments between 1984 and 2018.
(a) Global cumulative increase in water-covered area (seasonal + permanent) within dam catchments.
(b) Cumulative increase by continent. (c) Cumulative increase by pixel category for pre- and post-dam periods,
showing an increase in permanent- and seasonal-water pixels by around 50,000 km2 and a decrease in no-water
pixels. (d) Dam-induced surface water changes by continent.

Surface Water Coverage (km2)

Continent Number of Dams Pre-dam Post-dam Difference
Asia 3,115 24,279 47,775 23,496
South America 4,119 18,641 26,259 7,617
Africa 1,987 18,681 25,945 7,265
Europe 1,117 29,122 35,249 6,127
North America 1,352 74,684 79,795 5,111
Oceania 20 124 223 100
Total 11,710 165,530 215,245 49,715

Table 2. Summary statistics of global surface water changes by continent.

Technical Validation
Comparison with existing datasets. Dams. We compare GDAT with four existing global dam data-
bases, including GRanD, AQUASTAT, GOODD, and GeoDAR. We focus on GRanD and AQUASTAT when
comparing attributes because GOODD and GeoDAR do not contain publicly available information on dam or
reservoir attributes. As a result, this limits our ability to independently verify data accuracy and check for dupli-
cates. The GDAT database (Figs. 10,11) surpasses GRanD, AQUASTAT, and GeoDAR not only in total dam
count (Fig. 12) but also in attribute information on completion year, location, and generation capacity (Fig. 13).
Specifically, GDAT contains 412 percent and 141 percent more records than the GRanD and AQUASTAT data-
bases, respectively. GDAT also surpasses the total dam count of GeoDAR by 41 percent. For temporal coverage,

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 10

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 9 Surface water change in global dam catchments between 1984 and 2018 by country. The largest increase
in surface water coverage are in developed and rapidly developing countries with significant dam counts, such
as Brazil, Canada, China, India, the United States, Mexico, Turkey, and various Southeast Asian countries. The
data captures changes in the median surface water pixel counts between pre- and post-construction periods
within dam catchment areas. In addition to dam construction, a portion of the surface water changes in some
countries, especially those in Asia, could be attributed to improvements in satellite data quality.

GDAT contains data up to 2018, whereas the GRanD and AQUASTAT databases include dams constructed up
to 2010 and 2013 (Fig. 12). The WRI’s Global Power Plant Database is excluded from this comparison because it
focuses on power plants and, hence, contains mostly hydroelectric dams. For a visual comparison of geospatial
coverage between the databases, please see the Region Highlights section below.
GDAT contains more dams in all six continents of the world (Fig. 14) compared to GRanD and AQUASTAT.
GOODD contains 10 percent more dam observations than GDAT, most of which are in Asia. GeoDAR has
fewer observations than GDAT globally, but it has more coverage in North America, Europe, and Oceania. This
suggests existing global dam databases may be biased toward developed countries and may be underestimating
the existence of dams in lower-income and developing countries. In contrast, GDAT significantly improves data
coverage in South America, Africa, and Asia, where multiple new primary sources were consulted in the data
collection process. Compared to AQUASTAT, countries with significantly higher dam counts in GDAT include
India (1,440 more), China (446 more), Iran (241 more), Brazil (4,863 more), South Africa (4,340 more), and var-
ious countries in the Southeast Asian region. For instance, Myanmar, Thailand, Laos, Cambodia, and Vietnam
collectively have 575 more dams in GDAT than in AQUASTAT. Many of these countries are in the Global South,
where economic development has given rise to rapid dam construction, yet dam-related data is often sparse and
difficult to use. As such, GDAT is one of the most comprehensive geo-referenced global dam databases to date,
with significant improvements in the coverage of the Global South.
As a further step in cross validation, we assess the spatial overlap between GDAT, GOODD, and GeoDAR.
GOODD and GeoDAR are suitable datasets for cross-comparison, because they are independent data collec-
tion efforts that draw from different sources and use different data validation methods. We cannot match and
cross-check individual dams due to the lack of dam name and attribute information in GOODD and GeoDAR.
Instead, we assess the extent to which points in the GDAT, GOODD, and GeoDAR databases fall within the
same geographic areas, which may indicate overlaps in dam coverage. To do so, we draw a 0.1° buffer zone
around dams in each dataset. We then took the buffer zones for GDAT and verified the percentage of dams
in GOODD and GeoDAR that overlapped with the GDAT buffer zones. We repeated this procedure for the
GOODD and GeoDAR buffer zones.
Table 3 shows the percentage of each dataset captured by the 0.1° buffer zones. Around 54% of GDAT dams
are captured by the GOODD and GeoDAR buffer zones. Between 49–52% of GOODD dams are captured by
GDAT and GeoDAR buffer zones, while between 60–70% of GeoDAR dams are captured by GOODD and
GDAT buffer zones. The comparison reveals that a substantial subset of each dataset is unique, highlighting the
need for coordinated data collection efforts to create a globally consistent and comprehensive dam database.

Catchments. We compare the catchment areas of GDAT dams to those of GOODD, which is another global
dam dataset with algorithm-generated catchment. Validating our catchment areas against GOODD22 is the most

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 11

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 10 Map of Global Dam Tracker (GDAT) database – Global dams by location and main purpose.

Fig. 11 Map of Global Dam Tracker (GDAT) database – Global dams by reservoir capacity.

appropriate because both use similar procedures. This includes snapping the dams onto a river network from
HydroRivers and using PCRaster and elevation data to capture the drainage area upstream of each dam.
GDAT and GOODD are comparable in total catchment area coverage. GDAT catchment areas cover a
total of 44.88 million km2, while GOODD catchment areas cover a total of 46.77 million km2. As shown in
Fig. 15, GDAT and GOODD catchment areas overlap substantially. The overlapping area between the two data-
sets is 32.85 million km2, which constitutes 73.20% of GDAT catchments and 70.24% of GOODD catchments.
Nevertheless, many non-overlapping areas in each dataset are due to extremely large catchment areas, especially
in parts of Africa, the Middle East, Central Asia, and Eastern Europe, where few dams exist.
Within the overlapping areas, GDAT and GOODD catchments differ in size and extent. This is due to vari-
ations in regional dam counts between the two datasets. For instance, GOODD catchments cover more area in
Southern China and Southeast Asia, but GDAT contains a higher density of dams in Southwestern China and
Southeast Asia, making GDAT catchments smaller and more concentrated compared to GOODD catchments in
this region. Given the lack of dam attributes in GOODD, we are unable to match catchment areas and compare
them at an individual level. However, the differences between GDAT and GOODD catchment areas demonstrate
the importance of consolidating datasets to ensure consistency.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 12

www.nature.com/scientificdata/ www.nature.com/scientificdata

25,000

20,000

Cumulative Count
15,000
GDAT
AQUASTAT
GRanD
10,000

5,000

0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020

Completion Year

Fig. 12 Cumulative count of dams by completion year – GDAT vs. other databases.

40,000 38,640 38,640

35,140

31,780

30,000
Number of Observations

24,759 24,759
23,274 GDAT
GOODD
GeoDAR
20,000
AQUASTAT
GRanD
14,580

11,324

10,000 8,502

6,454 6,862 6,862

4,413

0 0 0 0 0 0
0
Capacity (MW) Completion Year Location Total

Fig. 13 Comparison between GDAT and other databases – Availability of dam attributes.

20,000
18,853

18,000

16,000
Number of Observations

14,000

GDAT
12,000
GOODD
GeoDAR
10,000 9,526
8,813 8,674 AQUASTAT
8,333
GRanD
8,000 7,566

6,354 6,413 6,392 6,243

6,000
4,330

4,000 3,201
3,556

2,677 2,839
2,252
2,048 1,963 1,947
2,000 1,282 1,321
1,507
996
726 646
299 255 259 271 430

0
North America Asia Europe South America Africa Oceania

Fig. 14 Comparison between GDAT and other databases – Dam count by continent.

Region highlights. Africa. Dams in Africa (Fig. 16) are concentrated in Sub-Saharan Western Africa
(Niger river basin), Southern Africa, the Nile river basin, and Morocco. South Africa alone has published infor-
mation for over 4,000 dams. Among the largest dams in Africa is the Aswan High Dam in Egypt, whose reservoir
(Lake Nasser) can hold roughly 132 km2 of water, and the Akosombo Dam in Ghana, with a reservoir capacity of
48 km2 and covers more than 3 percent of Ghana’s land area. In addition to the dams on the continent itself, small
dams are also present on the small island nations Cape Verde and Mauritius. Most dams in Africa are hydroelec-
tricity, irrigation, or water supply dams.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 13

www.nature.com/scientificdata/ www.nature.com/scientificdata

Dataset
Buffer Zones (0.1°) GDAT GOODD GeoDAR
GDAT 100% 48.9% 63.9%
GOODD 53.6% 100% 70.0%
GeoDAR 54.0% 52.1% 100%

Table 3. Cross-comparison of datasets–Percentage of overlap of dam location within 0.1° buffer zones.

Fig. 15 Comparison between GDAT and GOODD catchment areas. GDAT covers a total catchment areas
of 44.88 million km2, while GOODD covers a total of 46.77 million km2. Catchment areas from GDAT and
GOODD have 32.85 million km2 of overlap.

The large number of dams shown in South Africa is mainly due to the availability of published data from the
South African government (www.dws.gov.za/hydrology). For much of the continent, the predominant data-
bases used were AQUASTAT and GRanD. Other than South Africa, the rest of the continent has a relatively
small number and low density of dams compared to the rest of the world based on available data. Nevertheless,
many large hydropower projects are currently being planned, including nearly 150 projects being tracked by
International Rivers. Many large dams are funded by China, which, according to International Rivers, commit-
ted more than $3 billion to dam construction in Africa between 2001 and 200742. Policy drivers such as China’s
Belt and Road Initiative have been channeling more funds from China to large infrastructure projects in Africa,
such as the Kafue Gorge Lower dam in Zambia, the Gilgel Gibe III Dam in Ethiopia, and a $5.8 billion hydro-
power project in Nigeria that involves four dams43.

Asia. Most GDAT dams in Asia are concentrated in India and Eastern China (Fig. 17). India alone has over 4,000
dams, while China has more than 1,000. Japan, Turkey, Iran, and Southeast Asia each have more than 500 dams.
The largest concentrations of dams in India can be found in the central-western part of the country. While
most dams in India are small-scale irrigation dams, significant numbers of hydroelectric and water-supply dams
are scattered throughout the central-western parts of the country. The northern and northwestern parts of the
country (Rajasthan, Delhi, Uttar Pradesh, Bihar), where the elevation is flatter and lower than the central and
southern regions, have a significantly lower density of dams. There are very few dams in the Indo-Gangetic Plain
in the northeastern part of the country, as well as in the Thar Desert in Rajasthan.
Most dams in China are located in Sichuan and Yunnan, two provinces with high elevations and several
fast-flowing rivers. Southern China and the Yangtze basin have more dams than Northern China, and dams are
mostly concentrated in mountainous areas–there are no dams in the North China Plain. The country has many
flood control dams relative to surrounding countries and regions.
There is an especially high density of dams on the watersheds of the upper Yangtze, upper Mekong, and
upper Salween rivers, which collectively form the Three Parallel Rivers biosphere reserve. Many of these dams
are planned, under construction, or canceled. Being cross-border rivers, dams on the Yangtze, Mekong, and
Salween rivers are extensively tracked by international organizations because of the geopolitical implications of

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 14

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 16 GDAT Africa – Number and main purpose of dams by location. Many major river basins throughout
Africa are heavily dammed, such as the Nile, Western African rivers (Niger, Volta, Senegal), and Southern
African rivers (Zambezi, Limpopo). South Africa shows a particularly high concentration of dams, in part due
to better data accessibility.

Fig. 17 GDAT Asia – Number and main purpose of dams by location. For Asia, high concentrations of dams
exist in major river basins such as the Mekong, Yangtze, Yellow, and Tigris/Euphrates, as well as in many regions
such as the Indian subcontinent, Indochinese peninsula, Java, Japan, the Korean peninsula, and the Anatolian
peninsula. Many of these regions are home to rapidly developing economies with a high population density.

dam construction, which partly explains the larger number of dams shown in these areas. Fear among Southeast
Asian countries of China’s upstream dam-building activities have prevailed, despite regional organizations’ best
efforts, due to an unequal power dynamic44. Further proposals to construct large hydropower dams on China’s
southwestern rivers have caused concern among downstream neighbors45.
There is a high concentration of dams in peninsular Malaysia and on the island of Java, Indonesia. Turkey
and Iran have the highest numbers and concentrations of dams in the Middle East, especially because of the
aggressive dam-construction policies implemented by both countries. The highest density of dams in Iran is
found in the northwestern part of the country around Lake Urmia, an endorheic lake that has shrunken signif-
icantly due to dam construction46.
Because of the upstream location of Turkey and Iran on the Tigris and Euphrates river basins, respectively,
dam construction in both countries has raised geopolitical concerns by neighboring downstream countries such
as Iraq and Syria. Turkey, which has constructed dams on the Tigris and Euphrates basins as early as the 1950s,
has proposed more large-dam projects that are expected to significantly reduce the flow of major rivers into Syria
and Iraq. The country has also engaged in water politics against its downstream neighbors47. As in Africa, the
damming of rivers has caused wetlands and previously fertile agricultural lands to dry up, displacing people and
triggering geopolitical tensions48.

Oceania. There are more than 500 dams in Australia and roughly 70 in New Zealand (Fig. 18). Dams in
Australia are mostly concentrated in the coastal regions of the continent and on Tasmania, while dams in New
Zealand are relatively evenly distributed between the North and South Island. A few Pacific island countries

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 15

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 18 GDAT Oceania – Number and main purpose of dams by location. On the Australian mainland, dams
are mostly concentrated in the Eastern region, along the Great Dividing Range. Significant numbers of dams are
also present in Tasmania, as well as on the islands of New Zealand.

Fig. 19 GDAT Europe – Number and main purpose of dams by location. For Europe, the high frequency of
hydroelectric dams is due to the large number of observations supplied by the World Resources Institute Global
Power Plant database. Significant concentrations of dams are present in the Iberian peninsula, Southern France,
the Alps, Scotland, and Southern Norway.

(Fiji, Samoa) and Papua New Guinea also contain a small number of dams. Dams in Oceania are used mostly
for hydroelectricity and water supply and are no bigger than a few thousand cubic meters in reservoir capacity.

Europe. In Europe, the highest concentration of dams can be found in southern France, Scotland, southern
Norway, the Iberian Peninsula, and the Alpine regions of Central Europe (Fig. 19). On average, dams in Western
Europe tend to be smaller in reservoir capacity than those in Eastern Europe and Russia, even though they
are more numerous. While the rate of dam construction was high in the past two centuries, it has significantly
slowed in the present day.

North America. The United States and Canada have some of the highest concentration of dams in the world (Fig. 20).
After removing dikes and other non-dam structures, we obtain 7,039 dams from the USGS (nationalmap.gov).
The GDAT database lists more than 600 dams in Canada, most of which are concentrated in Ontario, Quebec,
and British Columbia. Canada boasts some of the largest hydroelectric dams in the world and is among the
world’s largest hydropower generators.
In the United States, California and Texas stand out as having the largest dam counts, while most states have
more than 100. Although there are many large dams in the United States, the vast majority are small. The USGS
data also contain information for dikes and small-river hydroelectric plants that do not necessarily involve dams.
To ensure consistency, we removed dikes and non-dam structures from the USGS database.
Dam construction in the United States has taken place since the early nineteenth century to improve nav-
igation on major rivers and provide electricity. During that time, the American public treated water as a com-
modity that needed to be improved through channeling and establishing waterworks. The utilitarian mindset
toward water resources persisted throughout the Progressive Era under then-president Theodore Roosevelt, who

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 16

www.nature.com/scientificdata/ www.nature.com/scientificdata

Fig. 20 GDAT North America – Number and main purpose of dams by location. Data for the United States
mainly come from USGS, with small dikes and non-dam structures removed from the database. In Canada,
most dams are concentrated closer to the southern border, with Quebec having a high concentration of
hydroelectric dams. Central Mexico also contains a high concentration of dams.

Fig. 21 GDAT South America – Number and main purpose of dams by location. The Amazon basin is home
to several large dams, while the highest concentration of dams can be found in the eastern and southeastern
regions of the country. The Andes also contain many dams, particularly in Peru and central Chile.

believed that natural resources needed to be used efficiently to provide the greatest good for the greatest number
of people49. In subsequent decades, the United States constructed many dams to power its industrial economy.
Many dams in the United States are now old and have reached or exceeded their design life, leading to a need
for dam removal50.

South America. In South America (Fig. 21), Brazil has over 5,200 dams (if dikes were included, Brazil would
have the largest recorded number of dams and dikes). Most dams in Brazil are located along the east coast around
the population centers of São Paulo, Rio de Janeiro, and Belo Horizonte. Most dams in Brazil are used for hydro-
electricity and water supply. Brazil depends on hydroelectricity for over 60 percent of its electric power supply,
and a recent boom in the small hydropower sector has put the country on track toward an energy surplus51.
However, Brazil’s hydropower construction boom in the Amazon basin has resulted in the destruction and
inundation of large rainforest areas. For example, the Belo Monte Dam, the world’s fourth-largest hydroelectric
project, flooded 260 square miles of forested lowlands and displaced more than 20,000 indigenous people52. In
response to intense local and international resistance from environmentalists and indigenous groups, the Brazilian
government recently announced a major shift away from its policy of building mega-dams in the Amazon basin53.
Although the construction of large mega-dams may ramp down, Brazil is still experiencing a rapid expansion
of small hydroelectric plants. Massive investments by the private sector after 1995 were stimulated by economic
incentives and new regulations within the energy market, resulting in hundreds of new small dams. A recent
study by researchers at the University of Washington estimates that 1,007 small hydroelectric plants are currently
operating in Brazil5. An additional 35 are under construction, and 156 are approved and awaiting final licensing.
Their research reports that 33 small hydropower plants have been constructed per year on average from 2001 to
2016, a growth rate 14 times as fast as that witnessed in the 1990s.

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 17

www.nature.com/scientificdata/ www.nature.com/scientificdata

Other Andean countries (Bolivia, Chile) also have significant numbers of dams, more than 600 of which are
located in the Peruvian Andes and used mostly for irrigation and hydroelectricity. In other parts of the Andes,
most dams are used for hydroelectricity, water supply, irrigation, and recreation. The Guri Dam in Venezuela, a
hydroelectric dam with a reservoir volume of 135 km3, ranks among the largest dams in South America and in
the world. According to AQUASTAT, Uruguay is home to the second largest number of dams in South America,
following Brazil. However, after spot-searching individual listings, the authors could only find location data for 4
of the 878 dams and reservoirs included. Additional research will be needed to confirm the validity and physical
location of these structures.
Code availability
Code54 for replicating results in this article is publicly available on Zenodo (https://doi.org/10.5281/zenodo.6784716).
We use Python (versions 3.6 and 3.7), Stata MP (version 15.1), Google Earth Engine (https://earthengine.google.com)
to obtain dam catchment data and conduct analysis.

Received: 1 July 2022; Accepted: 8 February 2023;

Published: xx xx xxxx

References
1. IEA. Renewables 2020. Tech. Rep., IEA (2020).
2. IEA. Electricity Information: Overview. Tech. Rep., IEA (2021).
3. Gernaat, D. E. H. J., Bogaart, P. W., van Vuuren, D. P., Biemans, H. & Niessink, R. High-resolution assessment of global technical and
economic hydropower potential. Nature Energy 2, 821–828 (2017).
4. Zarfl, C., Lumsdon, A. E., Berlekamp, J., Tydecks, L. & Tockner, K. A global boom in hydropower dam construction. Aquat. Sci. 77,
161–170 (2015).
5. Couto, T. B. A. & Olden, J. D. Global proliferation of small hydropower plants - science and policy. Front. Ecol. Environ. 16, 91–100
(2018).
6. World Commission on Dams. Dams and Development: A New Framework for Decision-making: Report of the World Commission on
Dams (Earthscan, 2000).
7. Zhang, A. T. Within but without: Involuntary displacement and economic development. Preprint at https://doi.org/10.2139/
ssrn.3358089 (2018).
8. Wang, P., Lassoie, J. P., Dong, S. & Morreale, S. J. A framework for social impact analysis of large dams: a case study of cascading
dams on the Upper-Mekong River, China. J. Environ. Manage. 117, 131–140 (2013).
9. Haddeland, I. et al. Global water resources affected by human interventions and climate change. Proc. Natl. Acad. Sci. USA 111,
3251–3256 (2014).
10. Schindler, D. W. & Donahue, W. F. An impending water crisis in Canada’s western prairie provinces. Proc. Natl. Acad. Sci. USA 103,
7210–7216 (2006).
11. Noori, R. et al. Anthropogenic depletion of Iran’s aquifers. Proc. Natl. Acad. Sci. USA 118 (2021).
12. Müller, M. F., Yoon, J., Gorelick, S. M., Avisse, N. & Tilmant, A. Impact of the Syrian refugee crisis on land use and transboundary
freshwater resources. Proc. Natl. Acad. Sci. USA 113, 14932–14937 (2016).
13. Del Bene, D., Scheidel, A. & Temper, L. More dams, more violence? A global analysis on resistances and repression around
conflictive dams through co-produced knowledge. Sustainability Sci. 13, 617–633 (2018).
14. Fearnside, P. M. Environmental and Social Impacts of Hydroelectric Dams in Brazilian Amazonia: Implications for the Aluminum
Industry. World Dev. 77, 48–65 (2016).
15. Maavara, T. et al. Global phosphorus retention by river damming. Proc. Natl. Acad. Sci. USA 112, 15603–15608 (2015).
16. Zarfl, C. et al. Future large hydropower dams impact global freshwater megafauna. Sci. Rep. 9, 18531 (2019).
17. Ziv, G., Baran, E., Nam, S., Rodrguez-Iturbe, I. & Levin, S. A. Trading-off fish biodiversity, food security, and hydropower in the
Mekong River Basin. Proc. Natl. Acad. Sci. USA 109, 5609–5614 (2012).
18. Barbarossa, V. et al. Impacts of current and future large dams on the geographic range connectivity of freshwater fish worldwide.
Proc. Natl. Acad. Sci. USA 117, 3648–3655 (2020).
19. Winemiller, K. O. et al. Balancing hydropower and biodiversity in the Amazon, Congo, and Mekong. Science 351, 128–129 (2016).
20. Dugan, P. J. et al. Fish migration, dams, and loss of ecosystem services in the Mekong basin. Ambio 39, 344–348 (2010).
21. Wohlfahrt, G., Tomelleri, E. & Hammerle, A. The albedo-climate penalty of hydropower reservoirs. Nat Energy 6, 372–377 (2021).
22. Mulligan, M., van Soesbergen, A. & Sáenz, L. GOODD, a global dataset of more than 38,000 georeferenced dams. Sci Data 7, 31
(2020).
23. Wang, J. et al. GeoDAR: georeferenced global dams and reservoirs dataset for bridging attributes and geolocations. Earth System
Science Data 14, 1869–1899 (2022).
24. Zhang, A. T., Urpelainen, J. & Schlenker, W. Power of the River: Introducing the Global Dam Tracker (GDAT). Tech. Rep., Center
on Global Energy Policy (2018).
25. Lehner, B. et al. High-resolution mapping of the world’s reservoirs and dams for sustainable river-flow management. Front. Ecol.
Environ. 9, 494–502 (2011).
26. Byers, L., et al. A Global Database of Power Plants. Tech. Rep., World Resources Institute (2018).
27. Lehner, B. & Grill, G. Global river hydrography and network routing: baseline data and new approaches to study the world’s large
river systems. Hydrol. Process. 27, 2171–2186 (2013).
28. Danielson, J. J. & Gesch, D. B. Global Multi-Resolution Terrain Elevation Data 2010 (GMTED2010). Tech. Rep., US Geological
Survey (2011).
29. Karssenberg, D., Schmitz, O., Salamon, P., de Jong, K. & Bierkens, M. F. A software framework for construction of process-based
stochastic spatio-temporal models and data assimilation. Environmental Modelling & Software 25, 489–502 (2010).
30. Pekel, J.-F., Cottam, A., Gorelick, N. & Belward, A. S. High-resolution mapping of global surface water and its long-term changes.
Nature 540, 418–422 (2016).
31. Linke, S. et al. Global hydro-environmental sub-basin and river reach characteristics at high spatial resolution. Scientific data 6, 1–15
(2019).
32. Gillies, S. The shapely user manual. https://pypi.org/project/Shapely (2013).
33. Lehner, B. HydroRivers Technical Documentation Version 1.0 (2019).
34. Hansen, B. E. The new econometrics of structural change: dating breaks in us labour productivity. Journal of Economic perspectives
15, 117–128 (2001).
35. Chu, C.-S. J., Stinchcombe, M. & White, H. Monitoring Structural Change. Econometrica 64, 1045–1065 (1996).
36. Piehl, A. M., Cooper, S. J., Braga, A. A. & Kennedy, D. M. Testing for structural breaks in the evaluation of programs. Rev. Econ. Stat.
85, 550–558 (2003).

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 18

www.nature.com/scientificdata/ www.nature.com/scientificdata

37. Davies, R. B. Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika 74, 33–43 (1987).
38. Andrews, D. W. Tests for parameter instability and structural change with unknown change point. Econometrica: Journal of the
Econometric Society 821–856 (1993).
39. Kim, H.-J. & Siegmund, D. The likelihood ratio test for a change-point in simple linear regression. Biometrika 76, 409–423 (1989).
40. Quandt, R. E. Tests of the hypothesis that a linear regression system obeys two separate regimes. Journal of the American statistical
Association 55, 324–330 (1960).
41. Zhang, A. T. & Gu, V. X. Global Dam Tracker (GDAT): A global database of more than 35,000 dams with location, catchment, and
attribute information. Zenodo. https://doi.org/10.5281/zenodo.7616851 (2023).
42. Hathaway, T. What is Driving Dams in Africa? Tech. Rep., International Rivers (2010).
43. Mongalvy, S., Malingha, D. & Sguazzin, A. Nigeria to Start Building $5.8 Billion Power Plant in 2018. Bloomberg News (2018).
44. Wong, C. China and the Mekong: The Floodgates of Power. The Diplomat Accessed: 2022-5-27 (2016).
45. Walker, B. China gives green-light to new era of mega-dams. China Dialogue. Accessed: 2022-5-27 (2013).
46. Rahimi, A. & Breuste, J. Why is Lake Urmia Drying up? Prognostic Modeling With Land-Use Data and Artificial Neural Network.
Front. Environ. Sci. 9 (2021).
47. Jongerden, J. Dams and politics in turkey: Utilizing water, developing conflict. Middle East Pol. 17, 137–143 (2010).
48. Levkowitz, J. Iraq wilting: How creeping drought could cause the next crisis. Middle East Institute. Accessed: 2022-5-27 (2018).
49. Billington, D. P., Jackson, D. C. & Melosi, M. V. The History of Large Federal Dams: Planning, Design, and Construction in the Era of
Big Dams (2005).
50. Ho, M. et al. The future role of dams in the United States of America. Water Resour. Res. 53, 982–998 (2017).
51. EIA. Country Analysis Brief - Brazil. EIA. Accessed: 2022-5-27 (2022).
52. Amazon Watch. Fact sheet: The Belo Monte Dam. Amazon Watch. Accessed: 2023-2-6 (2011).
53. Watts, J. Brazil raises hopes of a retreat from new mega-dam construction. The Guardian (2018).
54. Zhang, A. T. & Gu, V. X. Replication package for “Global Dam Tracker: A database of more than 35,000 dams with location,
catchment, and attribute information.” Zenodo. https://doi.org/10.5281/zenodo.6784716 (2023).

Acknowledgements
GDAT data collection was supported in part by the faculty grants program through the Center on Global Energy
Policy (CGEP) at Columbia University. We thank Wolfram Schlenker for guidance and Johannes Urpelainen,
Ryan Kennedy, and S.P. Harish for helpful discussions. Joanie Coker and Sid Shah provided excellent research
assistance. Support for the publication of this article was provided by the Class of 1956 Provost’s Faculty
Development Endowment at Washington and Lee University.

Author contributions
A.T.Z.: Conceptualized and designed the study, acquired funding, supervised and led the research team,
developed analytical tools, performed data analysis and verification, and wrote the paper. V.X.G.: Collected
and cleaned data, developed analytical tools, performed data analysis and verification, and wrote the paper.
All authors read and approved the manuscript.

Competing interests
The authors declare no competing interests.

Additional information
Correspondence and requests for materials should be addressed to A.T.Z.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2023

Scientific Data | (2023) 10:111 | https://doi.org/10.1038/s41597-023-02008-2 19