1r.watershed(1)              GRASS GIS User's Manual             r.watershed(1)
2
3
4

NAME

6       r.watershed  - Calculates hydrological parameters and RUSLE factors.
7

KEYWORDS

9       raster,  hydrology,  watershed, accumulation, drainage, stream network,
10       stream power index, topographic index
11

SYNOPSIS

13       r.watershed
14       r.watershed --help
15       r.watershed [-s4mab]  elevation=name   [depression=name]    [flow=name]
16       [disturbed_land=name]    [blocking=name]    [retention=name]   [thresh‐
17       old=integer]        [max_slope_length=float]        [accumulation=name]
18       [tci=name]        [spi=name]        [drainage=name]        [basin=name]
19       [stream=name]   [half_basin=name]   [length_slope=name]   [slope_steep‐
20       ness=name]    [convergence=integer]    [memory=memory in MB]   [--over‐
21       write]  [--help]  [--verbose]  [--quiet]  [--ui]
22
23   Flags:
24       -s
25           SFD (D8) flow (default is MFD)
26           SFD: single flow direction, MFD: multiple flow direction
27
28       -4
29           Allow only horizontal and vertical flow of water
30
31       -m
32           Enable disk swap memory option: Operation is slow
33           Only needed if memory requirements exceed available RAM; see manual
34           on how to calculate memory requirements
35
36       -a
37           Use positive flow accumulation even for likely underestimates
38           See manual for a detailed description of flow accumulation output
39
40       -b
41           Beautify flat areas
42           Flow direction in flat areas is modified to look prettier
43
44       --overwrite
45           Allow output files to overwrite existing files
46
47       --help
48           Print usage summary
49
50       --verbose
51           Verbose module output
52
53       --quiet
54           Quiet module output
55
56       --ui
57           Force launching GUI dialog
58
59   Parameters:
60       elevation=name [required]
61           Name of input elevation raster map
62
63       depression=name
64           Name of input depressions raster map
65           All non-NULL and non-zero cells are considered as real depressions
66
67       flow=name
68           Name of input raster representing amount of overland flow per cell
69
70       disturbed_land=name
71           Name of input raster map percent of disturbed land
72           For USLE
73
74       blocking=name
75           Name of input raster map blocking overland surface flow
76           For  USLE. All non-NULL and non-zero cells are considered as block‐
77           ing terrain.
78
79       retention=name
80           Name of input raster map with percentages for flow accumulation.
81
82       threshold=integer
83           Minimum size of exterior watershed basin
84
85       max_slope_length=float
86           Maximum length of surface flow in map units
87           For USLE
88
89       accumulation=name
90           Name for output accumulation raster map
91           Number of cells that drain through each cell
92
93       tci=name
94           Name for output topographic index ln(a / tan(b)) map
95
96       spi=name
97           Name for output stream power index a * tan(b)
98           Name for output raster map
99
100       drainage=name
101           Name for output drainage direction raster map
102           Directions numbered from 1 to 8
103
104       basin=name
105           Name for output basins raster map
106
107       stream=name
108           Name for output stream segments raster map
109
110       half_basin=name
111           Name for output half basins raster map
112           Each half-basin is given a unique value
113
114       length_slope=name
115           Name for output slope length raster map
116           Slope length and steepness (LS) factor for USLE
117
118       slope_steepness=name
119           Name for output slope steepness raster map
120           Slope steepness (S) factor for USLE
121
122       convergence=integer
123           Convergence factor for MFD (1-10)
124           1 = most diverging flow, 10 = most converging flow. Recommended: 5
125           Default: 5
126
127       memory=memory in MB
128           Maximum memory to be used (in MB)
129           Cache size for raster rows
130           Default: 300
131

DESCRIPTION

133       r.watershed generates a set of maps indicating: 1)  flow  accumulation,
134       drainage  direction,  the location of streams and watershed basins, and
135       2) the LS and S factors of the Revised  Universal  Soil  Loss  Equation
136       (RUSLE).
137

NOTES

139       Without flag -m set, the entire analysis is run in memory maintained by
140       the operating system. This can be limiting, but is very  fast.  Setting
141       this flag causes the program to manage memory on disk which allows very
142       large maps to be processed but is slower.
143
144       Flag -s force the module to use single flow direction (SFD, D8) instead
145       of multiple flow direction (MFD). MFD is enabled by default.
146
147       By  -4  flag the user allow only horizontal and vertical flow of water.
148       Stream and slope lengths are approximately the same as outputs from de‐
149       fault  surface  flow (allows horizontal, vertical, and diagonal flow of
150       water).  This flag will also make the drainage basins look more homoge‐
151       neous.
152
153       When  -a  flag is specified the module will use positive flow accumula‐
154       tion even for likely underestimates. When this flag is not  set,  cells
155       with  a  flow  accumulation value that is likely to be an underestimate
156       are converted to the negative. See below for a detailed description  of
157       flow accumulation output.
158
159       Option  convergence  specifies convergence factor for MFD. Lower values
160       result in higher divergence, flow is more  widely  distributed.  Higher
161       values  result  in higher convergence, flow is less widely distributed,
162       becoming more similar to SFD.
163
164       Option elevation specifies the elevation data on which entire  analysis
165       is based. NULL (nodata) cells are ignored, zero and negative values are
166       valid elevation data.  Gaps in  the  elevation  map  that  are  located
167       within  the  area  of  interest  must  be  filled beforehand, e.g. with
168       r.fillnulls, to avoid distortions.   The  elevation  map  need  not  be
169       sink-filled because the module uses a least-cost algorithm.
170
171       Option  depression  specifies the optional map of actual depressions or
172       sinkholes in the landscape that are large enough to slow and store sur‐
173       face  runoff  from  a storm event.  All cells that are not NULL and not
174       zero indicate depressions. Water will flow into but not out of  depres‐
175       sions.
176
177       Raster  flow  map specifies amount of overland flow per cell.  This map
178       indicates the amount of overland flow units that each  cell  will  con‐
179       tribute to the watershed basin model. Overland flow units represent the
180       amount of overland flow each cell contributes to surface flow. If omit‐
181       ted, a value of one (1) is assumed.
182
183       Raster  retention  map  specifies  correction factors per cell for flow
184       distribution. This map indicates the percentage of overland flow  units
185       leaving each cell. Values should be between zero and 100. If omitted, a
186       value of 100 is assumed.
187
188       Input Raster map or value containing  the  percent  of  disturbed  land
189       (i.e.,  croplands,  and  construction  sites) where the raster or input
190       value of 17 equals 17%.  If no map or value is given,  r.watershed  as‐
191       sumes no disturbed land. This input is used for the RUSLE calculations.
192
193       Option  blocking  specifies  terrain  that  will block overland surface
194       flow. Blocking cells and streams stop the slope length for  the  RUSLE.
195       All cells that are not NULL and not zero indicate blocking terrain.
196
197       Option  threshold  specifies  the minimum size of an exterior watershed
198       basin in cells, if no flow map is input, or overland flow units when  a
199       flow  map  is  given.  Warning: low threshold values will dramactically
200       increase run time and generate difficult to read basin  and  half_basin
201       results.   This  parameter  also  controls  the  level of detail in the
202       stream segments map.
203
204       Value given by max_slope_length option indicates the maximum length  of
205       overland  surface flow in meters. If overland flow travels greater than
206       the maximum length, the program assumes the maximum length (it  assumes
207       that landscape characteristics not discernible in the digital elevation
208       model exist that maximize the slope length).  This input  is  used  for
209       the RUSLE calculations and is a sensitive parameter.
210
211       Output  accumulation  map  contains the absolute value of the amount of
212       overland flow that traverses each cell. This value will be  the  number
213       of  upland  cells  plus  one  if no overland flow map is given.  If the
214       overland flow map is given, the value will be in overland  flow  units.
215       Negative numbers indicate that those cells possibly have surface runoff
216       from outside of the current geographic region.  Thus,  any  cells  with
217       negative  values  cannot  have  their  surface runoff and sedimentation
218       yields calculated accurately.
219
220       Output tci raster map contains topographic index TCI, computed as ln(α
221       /  tan(β))  where α is the cumulative upslope area draining through a
222       point per unit contour length and tan(β) is the local slope angle. The
223       TCI  reflects  the  tendency of water to accumulate at any point in the
224       catchment and the tendency for gravitational forces to move that  water
225       downslope  (Quinn  et  al.  1991).  This value will be negative if α /
226       tan(β) < 1.
227
228       Output spi raster map contains stream power index SPI, computed as α *
229       tan(β)  where  α  is  the  cumulative upslope area draining through a
230       point per unit contour length and tan(β) is the local slope angle. The
231       SPI  reflects the power of water flow at any point in the catchment and
232       the tendency for gravitational forces  to  move  that  water  downslope
233       (Moore  et  al. 1991).  This value will be negative if α < 0, i.e. for
234       cells with possible surface runoff from outside  of  the  current  geo‐
235       graphic region.
236
237       Output  drainage  raster map contains drainage direction.  Provides the
238       "aspect" for each cell measured CCW from East.
239       Figure: Drainage is 8 directions  numbered  counter-clockwise  starting
240       from  1 in north-east direction (source) Multiplying positive values by
241       45 will give the direction in degrees  that  the  surface  runoff  will
242       travel from that cell.  The value 0 (zero) indicates that the cell is a
243       depression area (defined by the depression input map).  Negative values
244       indicate  that  surface runoff is leaving the boundaries of the current
245       geographic region.  The absolute value of these  negative  cells  indi‐
246       cates  the direction of flow. For MFD, drainage indicates the direction
247       of maximum flow.
248
249       The output basin map contains unique label for  each  watershed  basin.
250       Each  basin  will be given a unique positive even integer.  Areas along
251       edges may not be large enough to create an  exterior  watershed  basin.
252       NULL  values indicate that the cell is not part of a complete watershed
253       basin in the current geographic region.
254
255       The output stream contains stream segments. Values  correspond  to  the
256       watershed basin values.  Can be vectorized after thinning (r.thin) with
257       r.to.vect.
258
259       The output half_basin raster map stores  each  half-basin  is  given  a
260       unique  value.  Watershed basins are divided into left and right sides.
261       The right-hand side cell of the watershed basin (looking upstream)  are
262       given  even values corresponding to the values in basin.  The left-hand
263       side cells of the watershed basin are given odd values  which  are  one
264       less than the value of the watershed basin.
265
266       The  output  length_slope  raster map stores slope length and steepness
267       (LS) factor for the  Revised  Universal  Soil  Loss  Equation  (RUSLE).
268       Equations  taken  from Revised Universal Soil Loss Equation for Western
269       Rangelands (Weltz et al. 1987). Since the LS factor is a  small  number
270       (usually less than one), the GRASS output map is of type DCELL.
271
272       The output slope_steepness raster map stores slope steepness (S) factor
273       for the Universal Soil Loss Equation (RUSLE).  Equations taken from ar‐
274       ticle  entitled  Revised  Slope Steepness Factor for the Universal Soil
275       Loss Equation (McCool et al. 1987).  Since the S factor is a small num‐
276       ber (usually less than one), the GRASS output map is of type DCELL.
277
278   AT least-cost search algorithm
279       r.watershed uses an AT least-cost search algorithm (see REFERENCES sec‐
280       tion) designed to minimize the impact of DEM data errors.  Compared  to
281       r.terraflow,  this algorithm provides more accurate results in areas of
282       low slope as well as DEMs  constructed  with  techniques  that  mistake
283       canopy tops as the ground elevation. Kinner et al. (2005), for example,
284       used SRTM and IFSAR DEMs to compare r.watershed against r.terraflow re‐
285       sults  in  Panama. r.terraflow was unable to replicate stream locations
286       in the larger valleys while r.watershed performed much better. Thus, if
287       forest canopy exists in valleys, SRTM, IFSAR, and similar data products
288       will cause major errors in r.terraflow  stream  output.  Under  similar
289       conditions,  r.watershed will generate better stream and half_basin re‐
290       sults. If watershed divides contain flat to low slope, r.watershed will
291       generate  better  basin results than r.terraflow. (r.terraflow uses the
292       same type of algorithm as ESRI’s ArcGIS watershed software which  fails
293       under  these  conditions.)  Also,  if  watershed divides contain forest
294       canopy mixed with uncanopied areas using SRTM, IFSAR, and similar  data
295       products,  r.watershed  will  generate better basin results than r.ter‐
296       raflow. The algorithm produces results similar to those  obtained  when
297       running r.cost and r.drain on every cell on the raster map.
298
299   Multiple flow direction (MFD)
300       r.watershed  offers  two methods to calculate surface flow: single flow
301       direction (SFD, D8) and multiple flow direction (MFD). With MFD,  water
302       flow is distributed to all neighbouring cells with lower elevation, us‐
303       ing slope towards neighbouring cells as a weighing factor  for  propor‐
304       tional  distribution.  The  AT least-cost path is always included. As a
305       result, depressions and obstacles are traversed with  a  graceful  flow
306       convergence before the overflow. The convergence factor causes flow ac‐
307       cumulation to converge more strongly with higher values. The  supported
308       range  is  1 to 10, recommended is a convergence factor of 5 (Holmgren,
309       1994). If many small sliver basins are created with  MFD,  setting  the
310       convergence  factor  to  a  higher value can reduce the amount of small
311       sliver basins.
312
313   In-memory mode and disk swap mode
314       There are two versions of this program: ram and seg.  ram  is  used  by
315       default, seg can be used by setting the -m flag.
316
317       The ram version requires a maximum of 31 MB of RAM for 1 million cells.
318       Together with the amount of system memory (RAM) available,  this  value
319       can  be  used  to  estimate whether the current region can be processed
320       with the ram version.
321
322       The ram version uses virtual memory managed by the operating system  to
323       store  all  the data structures and is faster than the seg version; seg
324       uses the GRASS segmentation library which manages data in  disk  files.
325       seg  uses only as much system memory (RAM) as specified with the memory
326       option, allowing other processes to operate on the  same  system,  even
327       when the current geographic region is huge.
328
329       Due  to  memory  requirements of both programs, it is quite easy to run
330       out of memory when working with huge map regions. If  the  ram  version
331       runs  out  of  memory and the resolution size of the current geographic
332       region cannot be increased, either more memory needs to be added to the
333       computer, or the swap space size needs to be increased. If seg runs out
334       of memory, additional disk space needs to be freed up for  the  program
335       to run.  The r.terraflow module was specifically designed with huge re‐
336       gions in mind and may be useful here as an alternative,  although  disk
337       space requirements of r.terraflow are several times higher than of seg.
338
339   Large regions with many cells
340       The  upper  limit  of  the  ram  version  is 2 billion (231 - 1) cells,
341       whereas the upper limit for the seg version is 9 billion-billion (263 -
342       1 = 9.223372e+18) cells.
343       In  some situations, the region size (number of cells) may be too large
344       for the amount of time or memory  available.  Running  r.watershed  may
345       then  require  use  of  a  coarser resolution. To make the results more
346       closely resemble the finer terrain data, create a map layer  containing
347       the lowest elevation values at the coarser resolution. This is done by:
348       1) Setting the current geographic region equal  to  the  elevation  map
349       layer  with g.region, and 2) Use the r.neighbors or r.resamp.stats com‐
350       mand to find the lowest value for an area equal in size to the  desired
351       resolution.  For example, if the resolution of the elevation data is 30
352       meters and the resolution of the geographic region for r.watershed will
353       be 90 meters: use the minimum function for a 3 by 3 neighborhood. After
354       changing to the resolution at which r.watershed will be  run,  r.water‐
355       shed  should  be  run using the values from the neighborhood output map
356       layer that represents the minimum elevation within the  region  of  the
357       coarser cell.
358
359   Basin threshold
360       The  minimum  size of drainage basins, defined by the threshold parame‐
361       ter, is only relevant for those watersheds with a single stream  having
362       at least the threshold of cells flowing into it.  (These watersheds are
363       called exterior basins.)  Interior drainage basins contain stream  seg‐
364       ments  below  multiple tributaries.  Interior drainage basins can be of
365       any size because the length of an interior stream segment is determined
366       by the distance between the tributaries flowing into it.
367
368   MASK and no data
369       The  r.watershed  program does not require the user to have the current
370       geographic region filled with elevation values.  Areas  without  eleva‐
371       tion  data  (masked  or NULL cells) are ignored. It is NOT necessary to
372       create a raster map (or raster reclassification) named  MASK  for  NULL
373       cells.  Areas without elevation data will be treated as if they are off
374       the edge of the region. Such areas will reduce the memory necessary  to
375       run  the  program.  Masking out unimportant areas can significantly re‐
376       duce processing time if the watersheds of interest occupy a small  per‐
377       centage of the overall area.
378
379       Gaps (NULL cells) in the elevation map that are located within the area
380       of interest will heavily influence the analysis: water will  flow  into
381       but  not  out of these gaps. These gaps must be filled beforehand, e.g.
382       with r.fillnulls.
383
384       Zero (0) and negative values will be treated  as  elevation  data  (not
385       no_data).
386
387   Further processing of output layers
388       Problem  areas, i.e. those parts of a basin with a likely underestimate
389       of flow accumulation, can be easily identified with e.g.
390         r.mapcalc "problems = if(flow_acc < 0, basin, null())"
391       If the region of interest contains such problem areas, and this is  not
392       desired,  the computational region must be expanded until the catchment
393       area for the region of interest is completely included.
394
395       To isolate an individual river network using the output of this module,
396       a number of approaches may be considered.
397
398       1      Use a resample of the basins catchment raster map as a MASK.
399              The  equivalent  vector  map method is similar using v.select or
400              v.overlay.
401
402       2      Use the r.cost module with a point in the river  as  a  starting
403              point.
404
405       3      Use  the v.net.iso module with a node in the river as a starting
406              point.
407
408       All individual river networks in the  stream  segments  output  can  be
409       identified  through  their ultimate outlet points. These points are all
410       cells in the stream segments output with negative  drainage  direction.
411       These  points  can  be  used  as  start  points  for  r.water.outlet or
412       v.net.iso.
413
414       To create river mile segmentation from a vectorized  streams  map,  try
415       the v.net.iso or v.lrs.segment modules.
416
417       The stream segments output can be easily vectorized after thinning with
418       r.thin. Each stream segment in the vector map will have  the  value  of
419       the  associated  basin.  To  isolate subbasins and streams for a larger
420       basin, a MASK for the larger basin can be created with  r.water.outlet.
421       The  stream segments output serves as a guide where to place the outlet
422       point used as input to r.water.outlet.  The basin threshold  must  have
423       been  sufficiently  small  to  isolate  a  stream network and subbasins
424       within the larger basin.
425
426       Given that the drainage  is  8  directions  numbered  counter-clockwise
427       starting  from  1 in north-east direction, multiplying the output by 45
428       (by 45. to get a double precision floating point raster map  in  r.map‐
429       calc)  gives  the  directions  in degrees. For most applications, zeros
430       which indicate depressions specified by depression and negative  values
431       which  indicate  runoff  leaving  the region should be replaced by NULL
432       (null() in r.mapcalc).  The following command performs  these  replace‐
433       ments:
434       r.mapcalc "drainage_degrees = if(drainage > 0, 45. * drainage, null())"
435       Alternatively, the user can use the -a flag or later the abs() function
436       in r.mapcalc if the runoff is leaving the region.
437

EXAMPLES

439       These examples use the Spearfish sample dataset.
440
441   Convert r.watershed streams map output to a vector map
442       If you want a detailed stream network, set the threshold  option  small
443       to create lots of catchment basins, as only one stream is presented per
444       catchment. The r.to.vect -v flag preserves the catchment ID as the vec‐
445       tor category number.
446         r.watershed elev=elevation.dem stream=rwater.stream
447         r.to.vect -v in=rwater.stream out=rwater_stream
448
449       Set a different color table for the accumulation map:
450         MAP=rwater.accum
451         r.watershed elev=elevation.dem accum=$MAP
452         eval `r.univar -g "$MAP"`
453         stddev_x_2=`echo $stddev | awk ’{print $1 * 2}’`
454         stddev_div_2=`echo $stddev | awk ’{print $1 / 2}’`
455         r.colors $MAP col=rules << EOF
456           0% red
457           -$stddev_x_2 red
458           -$stddev yellow
459           -$stddev_div_2 cyan
460           -$mean_of_abs blue
461           0 white
462           $mean_of_abs blue
463           $stddev_div_2 cyan
464           $stddev yellow
465           $stddev_x_2 red
466           100% red
467         EOF
468
469       Create  a  more detailed stream map using the accumulation map and con‐
470       vert it to a vector output map. The accumulation cut-off, and therefore
471       fractal  dimension, is arbitrary; in this example we use the map’s mean
472       number of upstream catchment cells (calculated in the above example  by
473       r.univar) as the cut-off value. This only works with SFD, not with MFD.
474         r.watershed elev=elevation.dem accum=rwater.accum
475         r.mapcalc ’MASK = if(!isnull(elevation.dem))’
476         r.mapcalc "rwater.course = \
477          if( abs(rwater.accum) > $mean_of_abs, \
478              abs(rwater.accum), \
479              null() )"
480         r.colors -g rwater.course col=bcyr
481         g.remove -f type=raster name=MASK
482         # Thinning is required before converting raster lines to vector
483         r.thin in=rwater.course out=rwater.course.Thin
484         r.colors -gn rwater.course.Thin color=grey
485         r.to.vect in=rwater.course.Thin out=rwater_course type=line
486         v.db.dropcolumn map=rwater_course column=label
487
488   Create watershed basins map and convert to a vector polygon map
489         r.watershed elev=elevation.dem basin=rwater.basin thresh=15000
490         r.to.vect -s in=rwater.basin out=rwater_basins type=area
491         v.db.dropcolumn map=rwater_basins column=label
492         v.db.renamecolumn map=rwater_basins column=value,catchment
493
494       Display output in a nice way
495         r.relief map=elevation.dem
496         d.shade shade=elevation.dem.shade color=rwater.basin bright=40
497         d.vect rwater_course color=orange
498

REFERENCES

500           •   Ehlschlaeger  C.  (1989).  Using the AT Search Algorithm to De‐
501               velop Hydrologic Models from Digital Elevation  Data,  Proceed‐
502               ings  of  International  Geographic  Information Systems (IGIS)
503               Symposium ’89, pp 275-281 (Baltimore, MD, 18-19 March 1989).
504               URL: http://chuck.ehlschlaeger.info/older/IGIS/paper.html
505
506           •   Holmgren P. (1994).  Multiple  flow  direction  algorithms  for
507               runoff  modelling  in grid based elevation models: An empirical
508               evaluation.  Hydrological Processes Vol 8(4), 327-334.
509               DOI: 10.1002/hyp.3360080405
510
511           •   Kinner D., Mitasova H., Harmon R., Toma L., Stallard R. (2005).
512               GIS-based  Stream Network Analysis for The Chagres River Basin,
513               Republic of Panama. The Rio Chagres: A  Multidisciplinary  Pro‐
514               file of a Tropical Watershed, R. Harmon (Ed.), Springer/Kluwer,
515               p.83-95.
516               URL: http://www4.ncsu.edu/~hmitaso/measwork/panama/panama.html
517
518           •   McCool et al. (1987). Revised Slope Steepness  Factor  for  the
519               Universal  Soil  Loss  Equation,  Transactions  of the ASAE Vol
520               30(5).
521
522           •   Metz M., Mitasova H., Harmon R. (2011). Efficient extraction of
523               drainage  networks  from  massive, radar-based elevation models
524               with least cost path search, Hydrol. Earth Syst. Sci.  Vol  15,
525               667-678.
526               DOI: 10.5194/hess-15-667-2011
527
528           •   Moore  I.D.,  Grayson R.B., Ladson A.R. (1991). Digital terrain
529               modelling: a review of hydrogical, geomorphological,  and  bio‐
530               logical applications, Hydrological Processes, Vol 5(1), 3-30
531               DOI: 10.1002/hyp.3360050103
532
533           •   Quinn  P.,  K. Beven K., Chevallier P., Planchon O. (1991). The
534               prediction of hillslope flow paths for distributed hydrological
535               modelling  using  Digital  Elevation  Models, Hydrological Pro‐
536               cesses Vol 5(1), p.59-79.
537               DOI: 10.1002/hyp.3360050106
538
539           •   Weltz M. A., Renard K.G., Simanton J. R. (1987).  Revised  Uni‐
540               versal Soil Loss Equation for Western Rangelands, U.S.A./Mexico
541               Symposium of Strategies for Classification  and  Management  of
542               Native  Vegetation  for  Food Production In Arid Zones (Tucson,
543               AZ, 12-16 Oct. 1987).
544

SEE ALSO

546        g.region, r.cost, r.drain, r.fillnulls, r.flow,  r.mask,  r.neighbors,
547       r.param.scale,  r.resamp.interp, r.terraflow, r.topidx, r.water.outlet,
548       r.stream.extract
549

AUTHORS

551       Original version: Charles Ehlschlaeger, U.S.  Army  Construction  Engi‐
552       neering Research Laboratory
553       Faster sorting algorithm and MFD support: Markus Metz <markus.metz.gis‐
554       work at gmail.com>
555       Retention for flow distribution by Andreas Gericke (IGB Berlin)
556

SOURCE CODE

558       Available at: r.watershed source code (history)
559
560       Accessed: Saturday Jan 21 20:39:30 2023
561
562       Main index | Raster index | Topics index | Keywords index  |  Graphical
563       index | Full index
564
565       © 2003-2023 GRASS Development Team, GRASS GIS 8.2.1 Reference Manual
566
567
568
569GRASS 8.2.1                                                     r.watershed(1)
Impressum