Abstract
The MERIT-Hydro networks re-gridded by the Iterative Hydrography Upscaling (IHU) algorithm do not retain exo- or endorheic basin attributes from the original data. Here we developed methods to assign such attributes to those and any other digital river networks. The motivation is that endorheic inland drainage basins are essential for hydrologic modelling of global and regional water balances, land surface water storage, gravity anomalies, sea level rise, etc. First, we create basin attributes that explicitly label endorheic and exorheic catchments by the criteria of direct or hidden connectivity to the ocean without changing their flow direction grid. In the second step we alter the delineation of endorheic basins by the merging algorithm that eliminates small inland watersheds to the adjacent host basins. The resulting datasets have a significantly reduced number of endorheic basins while preserving the total land portion and topology of the inland basins. The data was validated using the Water Balance Model by comparing volume of endorheic inland depressions with modelled water accumulation in their inland lakes.
Similar content being viewed by others
Background & Summary
While connectivity of surface water flows, such as streams and rivers, determine individual river basins or catchments, their drainage to the ocean or to inland seas and depressions classifies them as exorheic or endorheic basins respectively1. Unlike in natural environments where that definition of river catchments can be an oversimplification2, in this work we use a straightforward definition of basin types as those are directly derived from the digital river networks in almost all hydrological computer models.
Digital river networks with flow direction data3,4,5,6,7,8,9 are essential for hydrological10,11,12,13, ecosystem14,15, water resource management10,16,17, hydro-infrastructure18,19,20, and other geoscience models especially in those that simulate surface water routing in streams and rivers21. Flow direction data defines other important entities of the hydrological systems such as river stream order, tributaries, and the extent of the entire river basin. Each drainage basin has an outlet point or river mouth representing the last downstream grid cell of the directional tree graph that defines the river network. Exorheic river basins drain water to the world’s oceans, and, alternatively, there are endorheic (internal) basins that terminate on land and whose river mouths do not connect to the ocean. These endorheic basins terminate at inland lakes or dry depressions. All endorheic basins comprise about 20% of the global land area1. Modelling water storage in these endorheic lakes is very important for understanding its historical dynamics22,23, flooding or recession mitigation24,25. Additionally, interpretation of total water storage (TWS) contributions to the gravity anomalies recorded by the GRACE satellite is important26,27,28.
In many hydrological models, the river network is stored in a two-dimensional matrix where each cell specifies the direction of downstream flow from that grid cell to the next downstream cell. Being a special case of graph theory29,30, namely a directional tree graph, the flow direction matrix records the direction of neighbouring cells of the regular grids by a bit value in a byte number allowing a byte to record all 9 possible directions for flows into and out of a grid cell. Each of the 8 bits in a byte have a Boolean meaning for all 8 cardinal directions starting from east and continuing clockwise31. Eight zeroes in the byte indicate “no flow” or a river mouth outflow condition. Unfortunately, there is no additional room in the 8-bit byte value to hold an extra bit combination to indicate one of the two river mouth flow types, exorheic or endorheic. In some special cases, the network authors sacrifice the limitation of strict flow-in to/from flow-out conversion to record flow-out exorheic mouth direction as byte zero and endorheic mouth direction as byte −1 which are the wrap-around values of 2555. Using this work-around solution may help to differentiate the river mouth types in flow-out direction grids where only one outflow direction is allowed, but it requires off-splitting the exorheic mouth Boolean grid and patching back zero values to the outflow direction grid. The latter likely have been implemented in some hydrological packages, and we have implemented this in the University of New Hampshire Water Balance Model (WBM32).
The recently published high resolution MERIT river network dataset5 does include the special value −1 (255 as a wrap-around byte value), but more recent upscaled derivatives of this product at coarser resolutions by Eilander and others6 did not retain this information making these upscaled networks difficult to use in hydrological models where endorheic basins are explicitly modelled.
In exploring solutions to this oversight, we found that a simple solution to determine endo- and exorheism of the basins was not possible from just the flow direction. Specifically, the absence of nodata cells among the basin mouth neighbours indicate an endorheic basin since a nodata value is assumed to be non-land, and, thus, represents ocean grid cells. This seemingly simple method does not work as many near coastal river mouth cells may have narrow passages, such as estuaries or fiords, connecting them from inland to the ocean which exist on the sub-grid cell level but are not reflected in the coarser resolution upscaled flow direction grid.
This paper describes a river mouth delineation method to determine whether or not a drainage basin is endorheic or exorheic in the Eilander data6 upscaled river network datasets. In addition, we use optional methods for removal of numerous small endorheic watersheds. We tested and validated those methods over the upscaled 5 and 15 arc minute MERIT networks and the resulting value added datasets, with the additional basin type information, are available in the data repository33.
The MERIT-Plus river network datasets (in 5 and 15 arc minute resolution) discussed here add value to the original upscaled IHU MERIT data. The main purpose of this work to identify the endorheic and exorheic basin types which are missing in the source datasets. Merging (cleanup) of small endorheic basins introduced a few local changes in flow direction and basin ID data, but made the datasets more suitable for a broader range of hydrological modelling applications that simulate water balance and accumulation in the endorheic lakes and land depressions. Those applications are relevant to studies of climate change impact, the hydrological cycle in arid areas, interpretation of historical and seasonal gravity anomaly trends, water resource management, ecosystem protection, and endorheic lake assessment.
Methods
Source data
To develop and test basin type delineation methods we used the 5 and 15 arc minute networks produced by the Iterative Hydrography Upscaling (IHU) method6. These were upscaled from the original 3 arc second MERIT Hydro raster-based river network by Yamazaki and others5. The IHU upscaled datasets do not retain the river mouth type (endorheic or exorheic) from the source data. The IHU method description does not indicate a reason for this, and we assume the IHU method uses only flow direction data and no additional information to resolve this issue. The methods described below can be contributed to the future versions of the IHU open-source software for its additional utility to delineate drainage basin type (Fig. 1).
In a flow direction dataset all grid cells over land have a valid flow direction value, and therefore a nodata grid cell can be considered as part of the ocean. Identifying exorheic and endorheic basins cannot be carried out through a simple check of adjacency between the basin mouth grid cell and a nodata cell (exorheic) or a basin mouth grid cell with no adjacent nodata cell (endorheic). Unfortunately, this simple approach does not work due to several limitations:
-
1)
there can be a sub-grid cell level narrow passage to the ocean such as a fiord, estuary or human-made channel that is not reflected in a coarser scale grid;
-
2)
unresolved low-land or shallow ocean delta areas covered with vegetation (e.g., mangrove forests in tropics and ice fields in Arctic zones) causing those areas to be seemingly disconnected from the ocean;
-
3)
grid cells over water bodies of large rivers (e.g., Amazon, Lena, Indus) where river channel width exceeds cell size so that it has undesignated flow direction that gives a false single cell drainage basin; and
-
4)
digital elevation model (DEM) data or other errors in the digital river network algorithm that lead to false local flow directions.
Identification of basin types here does not change river flow direction data and basin boundaries. The primary goal of this task is to add endorheic/exorheic basin type labels to all basins that had been lost in the upscaled river network products as described in the “Source Data” section above.
If only the adjacency check is used for the basin type identification in the MERIT-Hydro IHU 6, then most land area would be wrongly classified as endorheic since limitation #3 above will falsely mark all major river basins as endorheic. In order to resolve river mouth connectivity to the ocean and, thus, river basin type attribution, we present here a multi-step procedure to both identify basin type and to merge many smaller endorheic basins into larger drainage basins.
Step 1 - Create initial list of all mouth outlets and assign first classification:
This is done by filtering out all outlets and checking them for the absence of nodata values in the adjacent grid cells (Fig. 2). This results in a list of exorheic mouth points, and the list of potentially endorheic outlets to be further processed in subsequent steps.
This is implemented through a sequence of straight forward actions:
-
1)
Filtering all river mouth grid cells by flow direction value equal to zero. The result is a list of locations (longitude/latitude and/or column/row) of all mouth points.
-
2)
Identification of exorheic mouth points by searching a 3 × 3 kernel around each mouth point and checking for the count of non-nodata cells (i.e., ocean). If the count of non-nodata cells in the kernel is 9, meaning that the river mouth point is surrounded by land cells, then it is listed as a potentially endorheic outlet. A kernel count less than 9 indicates the presence of nodata cells next to the mouth point making it a coastal cell and therefore an exorheic outlet.
-
3)
Lists of both types of outlets resulting from this step are also transferred to binary mask layers for use in the next segmentation step.
Step 2 - Coastal segmentation of mouth outlet clusters:
Both, the original 3 arc second MERIT network and the set of upscaled MERIT river network datasets allow a grid cell to have a flow direction value if it is part land and part ocean. Entirely freshwater grid cells in the estuary of large rivers, such as the Amazon, also have zero values for the flow direction making it potentially endorheic in Step 1. However, these “mouth” grid cells do not accumulate water over time as endorheic lakes and, therefore, cannot be included as endorheic basins. In this step we change the classification of these false endorheic outlets by checking whether these cells are adjacent to a coastal exorheic outlet grid cell or can be connected to it through the chain of other such false outlets (Fig. 3). If a potentially endorheic outlet (i.e., all adjacent cells have no nodata values) can be connected to the coast through a continuous chain of other exorheic outlets, then it is identified as exorheic. The rationale of this step is checking adjacency and/or chain adjacency conditions which does not require the use of any additional dataset, such as a high-resolution ocean coastline vector or ocean high resolution grid mask, making this step self-sufficient and solely based on the flow direction source data itself.
Software implementation of this step is done utilizing an Image2D package that is coded in many popular computer languages. We use the Perl PDL implementation in this work34. Its function “cc8compt()” performs image segmentation by labelling spatially continuous areas (clusters) where values are non-zero, thus providing a mask of river outlet points on a grid of both types. The number eight in the function name indicates that connectivity is checked through each of the four pixel sides and four corners (8-connected cells). Workflow here has two actions:
-
1)
Perform segmentation of the gridded mask layer of all outlet points, which results in a map of labelled clusters of continuous outlet pixels.
-
2)
Each cluster is then checked for intersections with coastline pixels (i.e., pixels bordering nodata grid cells). If there is an intersection with a coastline, then all outlets in this cluster are identified as exorheic.
This step works on the assumption that there are sub-grid cell level passages (such as fiords) to the ocean, and, thus, there cannot reasonably be an endorheic outlet next to the exorheic mouth cell. A close visual inspection of many clusters confirms a presence of sub-pixel passages to the ocean for each grid cell in the cluster (Fig. 3). We used an ultra-high resolution (10 m) land and coastline vector dataset35 to make sure that all chained coastal cluster grid cells are connected to the ocean.
Step 3 (optional) - Use an ocean or land mask:
In this step, we apply a high-resolution vector or grid land mask to check whether the mouth outlet grid cell contains an ocean or located at a minimal distance from the ocean (Fig. 4). The minimal distance is an input parameter with default values set to zero which means no checking by distance. This step is optional as it involves the use of an additional high-resolution land mask dataset which may not be available to the user. In this work we used land vector polygons at 30 m segments from OpenStreetMaps (https://www.openstreetmap.org).
Merging small endorheic drainage basins with neighbouring drainage basins
Identification of endorheic and exorheic drainage basins described in the previous section allows for the classification of many river basins to the exorheic and endorheic types based on the location of river mouth points relative to their location or their cluster location of nodata values grid cells, and the proximity to a high-resolution ocean coastline. No flow direction itself has been altered, however, close visual inspection of the remainder of endorheic basins and their mouth points in the upscaled MERIT river network indicates that there are a large number of small endorheic basins that may need to be removed from the network by merging them to adjacent watersheds. This process of merging basins necessitates a change in flow direction of the network and basin ID mask in those areas where merging is performed.
Whether or not a drainage basin flows to the ocean is a function of its topography, geomorphology, and water balance. Because endorheic basins are partly defined by their climate, any hydrological model run at century time scales will have variation in the water balance and small basins with low “pour points” that are currently endorheic could reconnect to a larger adjoining basin. Also, small size (e.g., up to 100 km2) endorheic basins inside a larger basin requires further assessment, especially if its mouth point is located at its own basin boundary. The location of an outlet point at its own basin boundary of a small endorheic basin can be the result of unresolved connectivity through a narrow canyon-like passage that is not reflected in the source DEM dataset used to produce the river flow data. We therefore implemented a routine to identify those small basins and to reconnect them to their larger adjacent basins. This had the effect of reducing the total number of endorheic drainage basins across the globe while leaving the total endorheic area in the IHU data (after performing steps # 1–3 above) with little change.
We have employed a multi-option approach to provide flexibility in which endorheic basins are merged. This is illustrated in Fig. 5 in which the following sequence of steps is implemented to process each endorheic basin:
-
1)
Filter out all endorheic basins by maximum drainage basin area size parameter. This is the most essential filter since we target for merging only small basins such that their removal will not significantly change the total global endorheic land area. Most of these small basins merge into a host basin which is also endorheic. These cases do not change the total endorheic area.
-
2)
Locate all inside and outside basin boundary grid cells and record the elevation of each. The MERIT dataset comes with the minimum river surface elevation in a grid cell which we use here and refer as “elevation”.
-
3)
For each inside boundary cell, trace the flow path to the basin outlet cell, record it and its flow path length.
-
4)
Identify and mark the pour point grid cell on the inside boundary by the lowest elevation (user option #1, Table 1) or minimum flow path length (user option #2) criteria. Note, these options are mutually exclusive and cannot both be set to True or False, and search for the lowest elevation pour point is skipped if option #2 is set to False.
-
5)
Check whether the difference in elevation between the pour point and outlet cell (if option #1 is set to True) or the flow path length is equal to or less than the corresponding “Maximum flow path length” input parameter value (if option #2 is set to True). Skip this basin or continue to the next action items based on this check outcome.
-
6)
Trace the flow path from the pour point cell to the outlet and reverse the flow direction of each of the grid cells along that path (Fig. 5).
-
7)
Connect the endorheic basin to the adjacent watershed using the chosen user input option #3 or #4 (lowest elevation or highest catchment area respectively) in the outside boundary cells that are directly adjacent to the pour point. This is done by setting the flow direction on the pour point grid cell toward a cell in one of the adjacent basin’s cells that meet the chosen criteria. Note, these options are mutually exclusive and cannot both be set to True or False. Also, the lowest elevation of the grid cell is not necessarily the same as the highest catchment area cell, because the elevation dataset represents the average grid cell elevation while the river path is near the minimum elevation of that cell and the difference between cell average and minimum elevation can be significant.
-
8)
Set all basin ID grid cells of the merged watershed to the basin ID of the host watershed it is merged to.
The available parameters and options for endorheic basin identification and merging procedure are listed in Tables 1, 2. If only the adjacency check is used for the basin type identification in the MERIT-Hydro IHU 6, then most land area would be wrongly classified as endorheic since limitation #3 above will falsely mark all major river basins as endorheic.
Auxiliary datasets
These methods were applied to upscaled 5 and 15 arc minute MERIT datasets for the endorheic basin identification and elimination of outliers that met certain criteria. These were flow direction and river elevation data sets. Upstream area and the basin ID mask were derived from flow direction data. Basin attributes were also derived from the UNH river database7,32 by matching each MERIT basin’s spatial extent to the known named rivers. Additional attribute files included the names of the host continent, receiving ocean, sea basin, and other characteristics such as basin area and main river length.
Another auxiliary dataset that we used for basin type identification and which does not belong to the MERIT package, is the high resolution land vector polygons from the OpenStreetMaps. The use of this dataset or any other for land/ocean masking is optional, but it helps to resolve connectivity of some river outlets to the ocean, especially, over Arctic and sub-Arctic lands where land elevation gradients are very low allowing very narrow ocean water passages to propagate far inland, but only at the sub-grid cell level only.
Products
We refer to the products of this work as MERIT-Plus reflecting value-added rationale in the delineation of the endorheic basins. As it is described in the previous sections, production of MERIT-Plus datasets involves two fundamental procedures: (1) identification that does not alter the original river flow direction, and (2) merging of small endorheic basins matching certain criteria to their adjacent host basins where this procedure changes the original flow direction and basin ID data. The resulting identity of the endorheic basins is recorded by two added specialized data layers:
-
1)
Gridded layer for endorheic basin IDs only where all other grid cells (ocean and exorheic basins) have nodata values.
-
2)
Signed integer data type for flow direction where endorheic outlets have a conventional value of −1. The numerical value for the flow direction of exorheic outlets is zero.
We used the open source GDAL driver AAIGrid (Arc/Info ASCII Grid) format for these layers which, if needed, can be readily converted to any other GDAL supported format (e.g., GeoTIFF, netCDF) by user preference (http://www.gdal.org/).
Data Records
The MERIT-Plus data public access, use, re-use, and re-distribution is warranted and compliant with the terms and conditions for data sharing by the “International Creative Commons Attribution 4.0” license.
Two upscaled 5 and 15 min MERIT source datasets6 have been processed for MERIT-Plus products. Important parameters, statistics, and basic validation for each of those spatial resolutions are discussed in the sections below as well as in the “README MERIT-Plus Dataset-v2.2.pdf” file of the MSD-LIVE repository33.
MERIT-Plus 05 minute v2.2 Data:
Repository: MSD-LIVE33, Project: Program on Coupled Human and Earth Systems (PCHES) https://data.msdlive.org/records/154gm-kvq48
File format: geoTIFF (.tif), arc ascii (.asc), ESRI shapefile (.shp), Keyhole Markup (.kml), and JavaScript Object Notation (.geojson).
File naming convention: MERIT_plus_05min_v2.2_{Variable}.{Format}
Where {Variable} is one of: IDs, IDsEnR, flwdir, flwdirEnR or upstrArea.
Where {Format} is one of: tif, asc, shp, kml, or geojson.
Date Produced: Dec 2023.
Spatial Metadata:
Extent: X: −180 to + 180
Extent Y: −60 to + 85 Resolution: 0.083333 decimal degrees (5 arc minutes)
Coordinate reference system: longitude/latitude reference system: longitude/latitude (WGS84 datum)
Projection in PROJ.4 notation: “ + proj = longlat + datum = WGS84”
Rows and columns: 1740, 4320
Units:
IDs: None
IDsEnR: None
flwdir: None
flwdirEnR: None
upstrArea: km2
Nodata value: Oceans, open-water, and Antarctica in the geoTIFF and ascii files have the no-data value of −9999. Exception, MERIT_plus_05min_v2.2_flwdir.tif has a nodata value of 247.
File format: tab delimited text (.csv)
File naming convention: MERIT_plus_05min_v2.2_{Variable}.csv Where {Variable} is one of: IDs or IDsEnR
Units:
IDs: None
IDsEnR: None
MERIT-Plus 15 minute v2.2 Data:
Repository: MSD-LIVE33, Project: Program on Coupled Human and Earth Systems (PCHES) https://data.msdlive.org/records/154gm-kvq48
File format: geoTIFF (.tif), arc ascii (.asc), ESRI shapefile (.shp), Keyhole Markup Language (.kml), and JavaScript Object Notation (.geojson).
File naming convention: MERIT_plus_15min_v2.2_{Variable}.{Format}
Where {Variable} is one of: IDs, IDsEnR, flwdir, flwdirEnR or upstrArea.
Where {Format} is one of: tif, asc, shp, kml, or geojson.
Date Produced: Dec 2023.
Spatial Metadata:
Extent: X: −180 to + 180
Extent Y: −60 to + 85 Resolution: 0.25 decimal degrees (15 minutes)
Coordinate reference system: longitude/latitude reference system: longitude/latitude (WGS84 datum)
Projection in PROJ.4 notation: “ + proj = longlat + datum = WGS84”
rows and columns: 580, 1440
Units:
IDs: None
IDsEnR: None
flwdir: None
flwdirEnR: None
upstrArea: km2
Nodata value: Oceans, open-water, and Antarctica in the geoTIFF and ascii files have the no-data value of −9999. Exception, MERIT_plus_15min_v2.2_flwdir.tif, has a nodata value of 247.
File format: tab delimited text (.csv)
File naming convention: MERIT_plus_15min_v2.2_{Variable}.csv Where {Variable} is one of: IDs or IDsEnR
Units:
IDs: None
IDsEnR: None
Technical Validation
MERIT-Plus river network datasets in 5 arc minute resolution
We used a multi-purpose 43 river network processing utility “networkTools” developed at UNH (https://github.com/wsag/WBM/tree/main/utilities) to build MERIT-Plus products by performing both identification and merging procedures with processing parameter values listed in Table 2. Network basin counts and other statistics are presented in Table 3.
The original IHU upscaled MERIT data set has 130,704 unique drainage basins all of which have outflow direction at their outlets equal to zero, and so cannot be identified as endorheic or exorheic. Most of these are small basins (91.2% of them are smaller than 10 grid cells in size or approximately 70 km2) that are located near ocean coastline and islands. Those numerous small coastal basins were found to be connected to the ocean through sub-pixel passages (e.g., fiords) resolved with a 10 m resolution land coastline dataset (see “Methods” section above) and composing chained coastal river outlet clusters (Fig. 4). Identification of endorheic basins yields a count of 7,738 basins that comprise 19% of the global land area (Table 3). Merging of the small endorheic basins that match filtering criteria (Table 2) reduces their count significantly to 1,708 while leaving the global endorheic land area almost unchanged. Visual checks of the merged basins indicate that most of them are in arid or semi-arid regions where formation of dunes or other landforms that block outlet passages are common.
In order to evaluate the procedures outlined above we used the UNH Water Balance Model (WBM32) to explore endorheic lake water accumulation. The question was whether any given endorheic basin would be likely to fill and overflow or will evapotranspiration rates exceed water accumulation rates, with the basin remaining hydrologically disconnected at the surface. The model was driven using 40 years of the MERRA2 historical climate drivers36 with all human hydrological components turned off37. WBM calculated endorheic lake water storage change using the difference between water inflow and evaporation from the lake surface where the latter is a function of storage and lake geometry and bathymetry. If the lake storage and size exceeds the depression capacity, then it is flagged as a false endorheic basin under historical climate conditions. Checking all endorheic basins by this criterion we found only 12 such outliers before merging, and after the merging none of those were found (Table 3).
We also checked the total/global endorheic land area and its location throughout the global land surface to the original MERIT Hydro5 and other known sources1,5,38 (Fig. 6). The match of MERIT-Plus to those is very good except the MERIT Hydro has a few extra locations in SW China, SE coast of South America, and few smaller areas in a wet temperate and tropical climate zone such as Indonesia. The authors of the latter dataset5 explain those as being karst drainage basins that are connected to the adjacent exorheic basins through the underground passages noting that those do not meet the common definition of “endorheic” basins by the connectivity to the ocean and, thus, are intrinsic to this particular dataset. Since we use basin connectivity to the ocean in the MERIT-Plus data, those (karst) endorheic area mismatches should be considered as an invalid basin type identification. The endorheic basin land fraction (18.81%, Table 3) match well with data from other sources1,8,39.
MERIT-Plus river network datasets in 15 arc minute resolution
Production of the MERIT-Plus river network data in 15 arc minute resolution was created using the same approach and software with altered input parameters (Table 4) to adjust for the coarser grid cell size. For example, the “Maximum area” parameter (P1) was increased to 3000 km2 to account for extra space of inscribing of actual basin boundaries to the coarser grid.
Summary and statistics of the resulting MERIT-Plus endorheic basins at 15 minute resolution (Table 5) have approximately three times fewer basins as compared to the 5 minute network. However, the difference between endorheic basin numbers after identification and especially after merging are not that different since most of the small erroneous endorheic basins have been eliminated by the IHU upscaling process6.
Validation of the basin type by WBM endorheic lake simulation using the same logic and approach as for the 5 minute network found two false basins (Table 5) which were merged to adjacent exorheic basins.
The global distribution of 15 minute endorheic basins (Fig. 6) and their land fraction (Table 5) is similar to those of the 5 minute resolution network with the expected reduced granularity due to the coarser resolution.
Uncertainty analysis
There are two aspects of the MERIT-Plus data production for the identification of endorheic and exorheic basins. The first one is relevant to assigning a Boolean value or flag for the basin endo- exorheic attribute without changing the flow direction data of the original source dataset6, and the second one is for merging small endorheic basins to host catchments. For both, we have conducted sensitivity analysis of our methods by the permutation of processing parameters that affect the output product. The uncertainty then is assessed from the variability of total endorheic area and match to known alternative endorheic basin maps described earlier in this section.
The summary of this analysis is given in the Table 6. The most important uncertainty analysis result is that ignoring or turning off the sub-grid cell ocean mask leads to significant mismatch of the resulting endorheic land areas to the known areas (indicated in the Table 6 “Location Mismatch” column) suggesting that the “observed” (this ocean mask) connectivity of basin outlets to the ocean is essential to produce the MERIT-Plus exo- and endorheic basin identification. Sensitivity, and, thus, the results uncertainty to the variability of the other dataset production parameters is fairly low assuring validation and quality of the data.
Code availability
Code used in this paper is available here: https://github.com/wsag/WBM/tree/main/utilities.
This GitHub wsag/WBM repository is licensed under the “GNU General Public License v3.0” and is one of the open-source public software access and use licenses.
File names:
1. networkTools- Executable. Usage: » networkTools -v JOB_PARAMETERS.init
2. networkTools_manual.init- processing parameters *.init file template. Use corresponding options for the endorheic delineation of a given river network. Input/Output options and parameters are described in it.
References
Vorosmarty, C. J., Fekete, B. M., Meybeck, M. & Lammers, R. B. Global System of Rivers: Its role in organizing continental land mass and defining land-to-ocean linkages. Global Biogeochem. Cycles 14, 599–621 (2000).
Meybeck, M., Dürr, H. H. & Vörösmarty, C. J. Global coastal segmentation and its river catchment contributors: A new look at land-ocean linkage. Global Biogeochem. Cycles 20, https://doi.org/10.1029/2005GB002540 (2006).
Naden, P. S. in Macroscale modelling of the hydrosphere Vol. IAHS Publication #214 (ed Wilkenson, B.) 67–79 (IAHS Press, 1993).
Lin, P., Pan, M., Wood, E. F., Yamazaki, D. & Allen, G. H. A new vector-based global river network dataset accounting for variable drainage density. Scientific Data 8, 28, https://doi.org/10.1038/s41597-021-00819-9 (2021).
Yamazaki, D. et al. MERIT Hydro: A High-Resolution Global Hydrography Map Based on Latest Topography Dataset. Water Resources Research 55, 5053–5073, https://doi.org/10.1029/2019WR024873 (2019).
Eilander, D. et al. A hydrography upscaling method for scale-invariant parametrization of distributed hydrological models. Hydrol. Earth Syst. Sci. 25, 5287–5313, https://doi.org/10.5194/hess-25-5287-2021 (2021).
Fekete, B. M., Vörösmarty, C. J. & Grabs, W. High-resolution fields of global runoff combining observed river discharge and simulated water balances. Global Biogeochem. Cycles 16, 15-11–15-10, https://doi.org/10.1029/1999GB001254 (2002).
Lehner, B. & Grill, G. Global river hydrography and network routing: baseline data and new approaches to study the world’s large river systems. Hydrological Processes 27, 2171–2186, https://doi.org/10.1002/hyp.9740 (2013).
Wu, H. et al. A new global river network database for macroscale hydrologic modeling. Water Resources Research 48, https://doi.org/10.1029/2012WR012313 (2012).
Grogan, D. S., Wisser, D., Prusevich, A., Lammers, R. B. & Frolking, S. The use and re-use of unsustainable groundwater for irrigation: a global budget. Environmental Research Letters 12, doi:Artn 034017 https://doi.org/10.1088/1748-9326/Aa5fb2 (2017).
Sutanudjaja, E. H. et al. PCR-GLOBWB 2: a 5 arcmin global hydrological and water resources model. Geoscientific Model Development 11, 2429–2453, https://doi.org/10.5194/gmd-11-2429-2018 (2018).
Hanasaki, N. et al. An integrated model for the assessment of global water resources – Part 1: Model description and input meteorological forcing. Hydrol. Earth Syst. Sci. 12, 1007–1025, https://doi.org/10.5194/hess-12-1007-2008 (2008).
Liang, X., Lettenmaier, D. P., Wood, E. F. & Burges, S. J. A simple hydrologically based model of land surface water and energy fluxes for general circulation models. Journal of Geophysical Research: Atmospheres 99, 14415–14428, https://doi.org/10.1029/94JD00483 (1994).
Guswa, A. J. et al. Ecosystem services: Challenges and opportunities for hydrologic modeling to support decision making. Water Resources Research 50, 4535–4544, https://doi.org/10.1002/2014WR015497 (2014).
Brewer, S. K. et al. Synthesizing models useful for ecohydrology and ecohydraulic approaches: An emphasis on integrating models to address complex research questions. Ecohydrology 11, e1966, https://doi.org/10.1002/eco.1966 (2018).
Liu, J. et al. Achieving sustainable irrigation water withdrawals: global impacts on food security and land use. Environmental Research Letters 12, 104009, https://doi.org/10.1088/1748-9326/aa88db (2017).
Döll, P. & Siebert, S. Global modeling of irrigation water requirements. Water Resources Research 38, 8-1–8-10, https://doi.org/10.1029/2001wr000355 (2002).
Rougé, C. et al. Coordination and control – limits in standard representations of multi-reservoir operations in hydrological modeling. Hydrol. Earth Syst. Sci. 25, 1365–1388, https://doi.org/10.5194/hess-25-1365-2021 (2021).
Zuidema, S. et al. Interplay of changing irrigation technologies and water reuse: example from the upper Snake River basin, Idaho, USA. Hydrol. Earth Syst. Sci. 24, 5231–5249, https://doi.org/10.5194/hess-24-5231-2020 (2020).
Hanasaki, N., Kanae, S. & Oki, T. A reservoir operation scheme for global river routing models. J. Hydrol. 327, 22–41, https://doi.org/10.1016/j.jhydrol.2005.11.011 (2006).
Gleason, C. J. Hydraulic geometry of natural rivers: A review and future directions. Progress in Physical Geography 39, 337–360, https://doi.org/10.1177/0309133314567584 (2015).
Yao, F. et al. Lake storage variation on the endorheic Tibetan Plateau and its attribution to climate change since the new millennium. Environmental Research Letters 13, 064011 (2018).
Wang, L., Wang, J., Li, M., Zhu, L. & Li, X. Lake area and volume variation in the endorheic basin of the Tibetan Plateau from 1989 to 2019. Earth Syst. Sci. Data Discuss. 2021, 1–36, https://doi.org/10.5194/essd-2021-331 (2021).
Yapiyev, V., Sagintayev, Z., Inglezakis, V. J., Samarkhanov, K. & Verhoef, A. Essentials of Endorheic Basins and Lakes: A Review in the Context of Current and Future Water Resource Management and Mitigation Activities in Central Asia. Water 9, 798 (2017).
Zhang, Z., Zheng, Y., Han, F., Xiong, R. & Feng, L. Recovery of an endorheic lake after a decade of conservation efforts: Mediating the water conflict between agriculture and ecosystems. Agricultural Water Management 256, 107107, https://doi.org/10.1016/j.agwat.2021.107107 (2021).
Abhishek & Kinouchi, T. Synergetic application of GRACE gravity data, global hydrological model, and in-situ observations to quantify water storage dynamics over Peninsular India during 2002-2017. J. Hydrol., 126069, https://doi.org/10.1016/j.jhydrol.2021.126069 (2021).
Ciracì, E., Velicogna, I. & Swenson, S. Continuity of the Mass Loss of the World’s Glaciers and Ice Caps From the GRACE and GRACE Follow-On Missions. Geophys. Res. Lett. 47, e2019GL086926, https://doi.org/10.1029/2019gl086926 (2020).
Loomis, B. D., Felikson, D., Sabaka, T. J. & Medley, B. High-Spatial-Resolution Mass Rates From GRACE and GRACE-FO: Global and Ice Sheet Analyses. Journal of Geophysical Research-Solid Earth 126, doi:ARTN e2021JB023024 https://doi.org/10.1029/2021JB023024 (2021).
Bender, E. A. & Williamson, S. G. Lists, Decisions and Graphs. 173 (S. Gill Williamson, 2010).
Deo, N. Graph theory with applications to engineering and computer science. (Courier Dover Publications, 2017).
Smith, T. R. & Park, K. K. Algebraic approach to spatial reasoning. International Journal of Geographical Information Systems 6, 177–192, https://doi.org/10.1080/02693799208901904 (1992).
Grogan, D. S. et al. Water balance model (WBM) v.1.0.0: a scalable gridded global hydrologic model with water-tracking functionality. Geosci. Model Dev. 15, 7287–7323, https://doi.org/10.5194/gmd-15-7287-2022 (2022).
Prusevich, A., Lammers, R. & Glidden, S. MERIT-Plus Dataset: Delineation of endorheic basins in 5 and 15 min upscaled river networks. MSD-LIVE Data Repository https://doi.org/10.57931/2248064 (2023).
Glazebrook, K., Williams, R., Jeness, T. & Burke, T. A. PDL::Image2D - Miscellaneous 2D image processing functions, https://metacpan.org/pod/PDL::Image2D (2022).
Gong, P. et al. Stable classification with limited sample: transferring a 30-m resolution sample set collected in 2015 to mapping 10-m resolution global land cover in 2017. Sci Bull (Beijing) 64, 370–373, https://doi.org/10.1016/j.scib.2019.03.002 (2019).
Molod, A., Takacs, L., Suarez, M. & Bacmeister, J. Development of the GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2. Geoscientific Model Development 8, 1339–1356, https://doi.org/10.5194/gmd-8-1339-2015 (2015).
Kåresdotter, E., Destouni, G., Ghajarnia, N., Lammers, R. B. & Kalantari, Z. Distinguishing Direct Human-Driven Effects on the Global Terrestrial Water Cycle. Earths Future 10, https://doi.org/10.1029/2022EF002848 (2022).
Earth Resources Observation And Science (EROS) Center. in USGS EROS Archive - Digital Elevation - HYDRO1K (https://doi.org/10.5066/F77P8WN0, 2018).
Verdin, K. L. Hydrologic Derivatives for Modeling and Analysis—A new global high-resolution database. U.S. Geological Survey Data Series 153, 16, https://doi.org/10.3133/ds1053 (2017).
Wolfe, R. MODIS Land Digital Elevation Model and Land/Water Mask in the Sinusoidal Grid Version 6.0. (NASA GSFC Report, https://landweb.modaps.eosdis.nasa.gov/QA_WWW/forPage/user_guide/DEM.pdf, 2013).
Acknowledgements
We would like to thank Danielle Grogan (UNH) for reviewing an earlier draft of this paper. This material was based upon work supported by the U.S. Department of Energy, Office of Science, Biological and Environmental Research Program, Earth and Environmental Systems Modelling, MultiSector Dynamics under Cooperative Agreement DE-SC0022141 and DE-SC0022141; the National Aeronautical and Space Administration, Earth Science Division’s High Mountain Asia program (grant no. 80NSSC20K1595), and the Earth Science Division’s Sea Level Change program (grant no. 80NSSC20K1296); and the Swedish funding agency Formas (grant no. 2017-00, 608).
Author information
Authors and Affiliations
Contributions
Conceptualization: A.A.P., R.B.L. Methodology: A.A.P., R.B.L. Investigation: A.A.P. Visualization: A.A.P. Funding acquisition: R.B.L. Project administration: R.B.L. Supervision: R.B.L. Writing: A.A.P., R.B.L., S.J.G. Data management: S.G., A.A.P. Data documentation: S.J.G., R.B.L., A.A.P. Data submission: S.J.G.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
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 Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted 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 licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Prusevich, A.A., Lammers, R.B. & Glidden, S.J. Delineation of endorheic drainage basins in the MERIT-Plus dataset for 5 and 15 minute upscaled river networks. Sci Data 11, 61 (2024). https://doi.org/10.1038/s41597-023-02875-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41597-023-02875-9