River routing for GEOSldas #1023

zyj8881357 · 2024-10-31T17:31:13Z

Pull Request Overview

This pull request initiates the implementation of a river routing process within GEOS, starting with an offline model.

Current Implementation: Only the offline version of the river routing model is introduced at this stage.

All model code is located in the directory:
GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp

The model reads runoff data from the Catchment model outputs (variables RUNOFF and BASEFLOW, with type tavg24_2d_lnd_Nx) and generates daily river-related outputs for each catchment area.

For detailed information on configuring and running the offline model, please refer to readme.txt in the above directory.

Next Steps:
This is an initial implementation, and many aspects require further discussion with relevant members to plan the next steps.

Additionally, work is ongoing to develop a pre-processing package for creating input data for the routing model.

…t in GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp

…ting_GEOSroute

zyj8881357 · 2024-12-03T17:01:12Z

New Updates:

1. The river routing model (excluding the reservoir and lake components) is now integrated into GEOSldas.
2. Currently, the model supports EASEv2 M09 and M36 grids. To enable the river routing feature, set RUN_ROUTE to 1 in the exeinp file.
3. Output Details
    River Discharge [m3 s-1]: Daily river discharge data is saved in the "river" directory (created automatically) within the experiment folder.
    The output is in vector format (*.txt files), where each file contains 291,284 values. These values correspond to catchments indexed from 1 to 291,284.
    To convert the vector data into spatial maps, use the 1-minute resolution catchment distribution map available in the CatchIndex variable of the SRTM_PfafData.nc file at:
    /discover/nobackup/projects/gmao/bcs_shared/make_bcs_inputs/land/topo/v1/SRTM-TopoData/SRTM_PfafData.nc
4. Restart Files: Restart files are generated monthly in the restart folder and can be read automatically as it exists. If there is no restart file, zero water storage will be used at the start.
5. The source code of offline routing model has been moved to the "offline" directory under GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/


For questions or further assistance, please contact Yujin Zeng at:
Email: yujin.zeng@nasa.gov

…ting_GEOSroute

mathomp4 · 2025-05-08T18:03:31Z

...GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/reservoir.F90

+  enddo
+
+  ! Deactivate reservoirs if the use_res flag is set to False
+  if(use_res == .False.) active_res = 0


Suggested change

if(use_res == .False.) active_res = 0

if(use_res .eqv. .False.) active_res = 0

GNU was tripping on this as, per Fortran, you have to use .eqv. and .neqv. to compare logicals.

I just fixed this. Thank you!

@zyj8881357 It looks like GNU is finding a couple more. Lines 1007 and 1057 in GEOS_RouteGridComp.F90

I also see similar in lines 66 and 250 in lake_mod.F90

Those have been fixed. Thank you!

zyj8881357 · 2025-05-08T18:17:44Z

New Updates:

The pre-processing package for the input data to routing model has been added. Related comments and documents were supplemented.
At this stage, the CMAKE file is not created for the pre-processing, so the package can be built and run separately (no need building together with the GEOSldas). It is easier for anyone who wants to test and run.

…ndGridComp.F90)

…ent (GEOS_RouteGridComp.F90)

…omp.F90)

gmao-rreichle

@zyj8881357: I started reviewing the routing model PR. I added a couple of minor commits with some cleanup (just a start). More importantly, we'll need to address the comments below, most of which are quite major and will require substantial revisions in how the model is integrated in GEOS.
cc: @weiyuan-jiang @tclune @biljanaorescanin @rdkoster @amolod @lcandre2

gmao-rreichle · 2025-05-21T18:15:12Z

...m_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOS_LandGridComp.F90

@@ -31,7 +31,7 @@ module GEOS_LandGridCompMod
  use GEOS_CatchGridCompMod,   only : CatchSetServices    => SetServices
  use GEOS_CatchCNGridCompMod, only : CatchCNSetServices  => SetServices
  use GEOS_IgniGridCompMod,    only : IgniSetServices     => SetServices
-!  use GEOS_RouteGridCompMod,   only : RouteSetServices    => SetServices
+  use GEOS_RouteGridCompMod,   only : RouteSetServices    => SetServices


This comment is not about this particular line, but rather about the placement of GEOS_RouteGridComp inside GEOS_LandGridComp. Ultimately, we will need to route runoff from the land and the landice tiles. To do that, I think we need to place GEOS_RouteGridComp one level up, within GEOS_SurfaceGridComp (that is, next to GEOS_LandGridComp). Making this change will likely require extensive re-wiring. I wonder if it's best to tackle this re-wiring first or delay the move until later.

gmao-rreichle · 2025-05-21T18:17:27Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

-  integer, parameter :: N_CatG = 291284
+
+  integer, parameter :: upmax    = 34
+  character(len=500) :: inputdir = "/discover/nobackup/yzeng3/data/river_input/"


We cannot hardwire a user-specific directory here. The inputs that come from this directory need to become part of the bcs datasets. See also comment below about adding parameters into routing restart files.

gmao-rreichle · 2025-05-21T18:20:57Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

+
+  integer, parameter :: upmax    = 34
+  character(len=500) :: inputdir = "/discover/nobackup/yzeng3/data/river_input/"
+  logical, parameter :: use_res  = .True.


Are we sure that we want to hardwire using the reservoir model (if that's what this switch does)? It may be better to include config information in the "RUN_ROUTE" resource variable. E.g., RUN_ROUTE=1 could be the routing model without reservoirs, and RUN_ROUTE=2 could be the routing model with reservoirs

gmao-rreichle · 2025-05-21T18:23:36Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

-  USE catch_constants, ONLY:          &
-       N_CatG => N_Pfaf_Catchs  
-#endif
+  use ROUTING_MODEL,          ONLY: river_routing_lin, river_routing_hyd, ROUTE_DT


river_routing_lin() is commented out below. Do we want to make this a runtime option? We could use RUN_ROUTE for this purpose. E.g., RUN_ROUTE=1 runs the hydraulic model, and RUN_ROUTE=2 runs the linear model. If we never want to run the linear model, it should be removed from the code.

gmao-rreichle · 2025-05-21T18:25:09Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90


 ! -----------------------------------------------------------
 ! RUN -- Run method for the route component
 ! -----------------------------------------------------------
+  subroutine RUN1 (GC,IMPORT, EXPORT, CLOCK, RC )


Why define an empty subroutine RUN1()? Do we need this for anything? If there's a call to RUN1(), it should be removed, then we can remove the empty subroutine.

gmao-rreichle · 2025-05-21T18:33:17Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

+
+    ! Read sub-catchment data 
+    allocate(nsub_global(N_CatG),subarea_global(nmax,N_CatG))
+    open(77,file=trim(inputdir)//"/Pfaf_nsub_"//trim(resname)//".txt",status="old",action="read"); read(77,*)nsub_global; close(77)


Not sure what this and many subsequent read statements do, but my sense is that we need to include the routing model parameters in the routing model restart files. Then the "init" routine would not need tap into external files.
Normally, the process for land model parameters is as follows:

Create parameter datasets that can be archived in the bcs_shared directory (along with all the other bcs files). This is accomplished by make_bcs (and may already be part of the present PR).

Create a routing restart file that includes the routing model parameters. This interfaces with the "remap_restarts" package in the GEOS_Util repo.

When the model is run (in GEOSldas or GEOSgcm), the routing model parameters are read in from the restart file.

gmao-rreichle · 2025-05-21T18:35:00Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

+    if(mapl_am_I_root())print *, "init time is ", YY, "/", MM, "/", DD, " ", HH, ":", MMM, ":", SS    
+    allocate(wriver(ntiles),wstream(ntiles),wres(ntiles))
+    allocate(wriver_global(n_catg),wstream_global(n_catg),wres_global(n_catg))
+    open(77,file="../input/restart/river_storage_rs_"//trim(yr_s)//trim(mon_s)//trim(day_s)//".txt",status="old",action="read",iostat=status)


We should not introduce custom, ASCII restart files for the routing model. Instead, we need to make sure the restarts are handled by MAPL and are netcdf-formatted.

gmao-rreichle · 2025-05-21T18:36:31Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

+          if(mapl_am_I_root())then   
+             !open(88,file="../river/river_storage_"//trim(yr_s)//trim(mon_s)//trim(day_s)//".txt",action="write")
+             !open(89,file="../river/stream_storage_"//trim(yr_s)//trim(mon_s)//trim(day_s)//".txt",action="write")
+             open(90,file="../river/river_flow_"//trim(yr_s)//trim(mon_s)//trim(day_s)//".txt",action="write")             


Output from the routing model should be handled by MAPL HISTORY. We can't publish custom ASCII files with GEOS products.

gmao-rreichle · 2025-05-23T18:23:51Z

@weiyuan-jiang : In a recent commit (85af5a7), I added a USE statement into GEOS_RouteGridComp.F90 that pulls a constant from:
GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/Shared/catch_constants.f90
But this trips up the build because I didn't add the dependency into CMakeLists.txt, which may also need another fix (replace "esmf" with "ESMF::ESMF" in dependencies?). When you get a chance, please edit:
https://github.com/GEOS-ESM/GEOSgcm_GridComp/blob/feature/yujinz/Routing_GEOSroute/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/CMakeLists.txt
accordingly. Thanks!

weiyuan-jiang · 2025-05-23T18:32:16Z

@weiyuan-jiang : In a recent commit (85af5a7), I added a USE statement into GEOS_RouteGridComp.F90 that pulls a constant from: GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/Shared/catch_constants.f90 But this trips up the build because I didn't add the dependency into CMakeLists.txt, which may also need another fix (replace "esmf" with "ESMF::ESMF" in dependencies?). When you get a chance, please edit: https://github.com/GEOS-ESM/GEOSgcm_GridComp/blob/feature/yujinz/Routing_GEOSroute/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/CMakeLists.txt accordingly. Thanks!

I think we need to add dependencies "GEOS_LandShared ESMF::ESMF" . catch_constant is from GEOS_LandShared and ESMF::ESMF is a new standard way. Just added

weiyuan-jiang · 2025-07-15T16:48:24Z

I am quite confused about the variable names. "N_catg" was used as global tile number before. Can it be changed to N_pfaf_g here? The "ntiles" here seem n_pfaf_local to me. ( correct me if I am wrong)

weiyuan-jiang · 2025-07-15T16:53:34Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

+             if(it>0)then
+                RUNOFF_ACT(i)=RUNOFF_ACT(i)+route%subarea(j,i)*runoff_global(it)/1000.   
+             endif
+             if(it==0)exit


exit or cycle ?

weiyuan-jiang · 2025-07-15T16:54:37Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

+       call MPI_allgatherv  (                          &
+            QOUT_CAT,  route%scounts_cat(mype+1)      ,MPI_REAL, &
+            QOUTFLOW_GLOBAL, route%scounts_cat, route%rdispls_cat,MPI_REAL, &
+            MPI_COMM_WORLD, mpierr)      


All the MPI_COMM_WORLD should be replaced by the comm obtained from VM

weiyuan-jiang · 2025-07-15T17:02:13Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

-
+    VERIFY_(STATUS)     
+    route%nt_global = nt_global
+    ! Determine the resolution


use gridname to determine the resolution

weiyuan-jiang · 2025-07-15T17:10:22Z

...cs_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp/GEOS_RouteGridComp.F90

-    ! define minCatch, maxCatch
+
+    allocate(ims(1:ndes))
+    ! define catchment space for this processor
    call MAPL_DecomposeDim ( n_catg,ims,ndes ) ! ims(mype+1) gives the size of my partition


It seems to to me the distribution of pfaf has nothing to do with the distribution of tiles. Is there any proximity that can be taken advantage of ?

Yujin Zeng and others added 6 commits October 23, 2024 21:22

branch feature/yujinz/Routing_GEOSroute started, offline model was pu…

7faae9f

…t in GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSroute_GridComp

change some paths for Discover

54950b8

script build is replaced by run

982bdd9

readme.txt added

f2a19da

update K_str

bf8b9a8

Merge remote-tracking branch 'origin/develop' into feature/yujinz/Rou…

71e44c5

…ting_GEOSroute

zyj8881357 added 0 diff The changes in this pull request have verified to be zero-diff with the target branch. Draft labels Oct 31, 2024

zyj8881357 requested a review from gmao-rreichle October 31, 2024 17:31

zyj8881357 self-assigned this Oct 31, 2024

zyj8881357 requested a review from a team as a code owner October 31, 2024 17:31

zyj8881357 marked this pull request as draft October 31, 2024 17:51

removing some Chinese characters in the comments

c6e341b

zyj8881357 changed the title ~~Feature/yujinz/routing geo sroute~~ River routing for GEOSldas, offline mode Oct 31, 2024

Yujin Zeng added 4 commits November 5, 2024 11:45

more cleaning up

d2a0c95

Merge remote-tracking branch 'origin/develop' into feature/yujinz/Rou…

0d8ebcb

…ting_GEOSroute

move offline model to dir offline

5e9377a

routing model was implemented to GEOSldas

7ceed5a

zyj8881357 changed the title ~~River routing for GEOSldas, offline mode~~ River routing for GEOSldas Dec 3, 2024

Yujin Zeng and others added 10 commits December 6, 2024 12:40

back to TAKI

17b7d4b

back to Discover

431674c

a small bug fix for the offline model

5d98b30

update the offline model to the coupled model

abcf705

Merge branch 'develop' into feature/yujinz/Routing_GEOSroute

416a94f

reservoir module added

96cbcd6

temporarily disable flood control for dams

ebe5287

preprocessing for routing model added

a81bcea

code clean up, and comment&documents added

0c5d779

Merge remote-tracking branch 'origin/develop' into feature/yujinz/Rou…

af2ec8b

…ting_GEOSroute

mathomp4 reviewed May 8, 2025

View reviewed changes

Yujin Zeng added 2 commits May 8, 2025 14:23

'==' changed to '.eqv.'

9a08d7b

'.eqv.' replaces '=='

32fe955

zyj8881357 requested a review from lcandre2 May 13, 2025 15:25

gmao-rreichle and others added 4 commits May 21, 2025 09:13

Merge branch 'develop' into feature/yujinz/Routing_GEOSroute

1fb57bb

changed variable name NUM_CATCH -> NUM_CATCH_ENS for clarity (GEOS_La…

ef58210

…ndGridComp.F90)

avoid duplicate hard-coding of #Pfaf catchments; fixed indent, alignm…

85af5a7

…ent (GEOS_RouteGridComp.F90)

added comments to clarify connected "RUNOFF" variable (GEOS_LandGridC…

4e3cfec

…omp.F90)

gmao-rreichle reviewed May 21, 2025

View reviewed changes

Update CMakeLists.txt

e367061

zyj8881357 force-pushed the feature/yujinz/Routing_GEOSroute branch from f4d3c73 to e367061 Compare June 9, 2025 14:48

Yujin Zeng and others added 3 commits June 18, 2025 12:06

copying river_input through make_bcs

f1c450d

river routing input data pointing to bcs

c60c59b

-diag-disable=10448 added in the building of routing_model preproc

636cbf1

zyj8881357 mentioned this pull request Jun 27, 2025

River routing for GEOSldas GEOS-ESM/GEOSldas_GridComp#131

Draft

weiyuan-jiang reviewed Jul 15, 2025

View reviewed changes

	if(use_res == .False.) active_res = 0
	if(use_res .eqv. .False.) active_res = 0

River routing for GEOSldas #1023

Are you sure you want to change the base?

River routing for GEOSldas #1023

Uh oh!

Conversation

zyj8881357 commented Oct 31, 2024

Uh oh!

zyj8881357 commented Dec 3, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zyj8881357 commented May 8, 2025

Uh oh!

gmao-rreichle left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gmao-rreichle commented May 23, 2025

Uh oh!

weiyuan-jiang commented May 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

weiyuan-jiang commented Jul 15, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

weiyuan-jiang Jul 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

zyj8881357 commented Dec 3, 2024 •

edited

Loading

weiyuan-jiang commented May 23, 2025 •

edited

Loading

weiyuan-jiang Jul 15, 2025 •

edited

Loading