merge with main

Author: FlorisBuwaldaDeltares

README.md

@@ -27,12 +27,13 @@ If you are interested in using these products, and do not (yet) want to compile
 - **Delft3D 4 Suite website:** https://www.deltares.nl/en/software-and-data/products/delft3d-4-suite
 - **Delft3D FM Suite:** https://www.deltares.nl/en/software-and-data/products/delft3d-flexible-mesh-suite
 
-and contact our **sales team:** https://www.deltares.nl/en/software-and-data/software-sales-and-support-teams
+and contact our **sales services team:** https://www.deltares.nl/en/software-and-data/software-sales-and-support-teams
 
-## Bug reports
+## Community support
 
-To limit the number of parallel communication channels and issue trackers, the issues tab has been removed on the Delft3D GitHub site.
-In case you encounter bugs, please report them to our **support team:** https://www.deltares.nl/en/software-and-data/software-sales-and-support-teams
+Clients with Service Packages have access to the support team.
+For the wider open source community, we recommend the use of the [GitHub Discussions](https://github.com/Deltares/Delft3D/discussions) tab.
+Please post questions and suggestions there in the Q&A sections.
 
 ## Open Source Community
 

doc/contributing.md

@@ -5,23 +5,44 @@ Below are the specifics about our contributing process.
 If you are entirely new to contributing to open source, [this generic guide](https://opensource.guide/how-to-contribute/) also helps explain why, what, and how to successfully get involved in open source projects.
 
 ## Workflow
- - Create a fork of the Delft3D repository, or request write-access to our Delft3D repository via Delft3D support.
- - Create a JIRA issue ticket at https://issuetracker.deltares.nl describing the bug to be fixed or functionality to be developed.
-   The issue number is relevant for naming the development branch (see below).
-   Third party developers don't have access to the issue tracker; we recommend to communicate with a Deltares contact person.
+### External contributors
+ - Create a fork of the Delft3D repository.
+ - Reach out to Deltares as early as possible if you intend to contribute changes back to the main Delft3D version.
+   This helps us with keeping track of what developments are ongoing.
+   This helps avoid starting work on something that other people are already implementing, and we can give guidance on how best to implement the intended changes.
+   Reach out to us via the **sales services team:** https://www.deltares.nl/en/software-and-data/software-sales-and-support-teams with a brief description of the scope of the code change; they will forward your request internally to the appropriate development team.
+ - Checkout/Clone the repository locally.
+ - Create a branch, ideally using the naming convention below.
+   The frequency of updating your fork/branch from the Deltares main is up to personal taste.
+   Yet, merge from our main as often as possible, and contribute back to us as early as possible.
+ - Implement, test and document the modifications.
+ - Provide a patch-file, or reach out to Deltares to create a pull request:
+   - Although anyone can create a pull request on our repository, our pipelines will only be triggered if the pull request is created by a Deltares employee.
+   - Merging back to our main will typically include the following steps: transfer code changes to a branch in our Delft3D repository, security scan of the changes made, creation of a pull request, review of code, documentation and test cases, and automated code testing.
+     Obviously, with some iterations if one of the steps identifies issues to be resolved before the merge.
+ - To keep legal representation of the Delft3D software indisputable, we ask you to sign a Fiduciary License Agreement (FLA) before the final merge into main.
+   For an explanation why, see [this page](https://fsfe.org/activities/fla/fla.en.html) by the Free Software Foundation Europe.
+   The FLA can be obtained via the Deltares contact person who handles the merging process.
+   Signing the FLA makes sense for code contributions of significant size.
+   For small bug fixes, it's better to send an email with a test case and a description of the recommended code changes than following the formal procedure described above.
+
+### Deltares employees
  - Checkout/Clone the repository locally.
+ - Create a JIRA issue ticket at https://issuetracker.deltares.nl describing the bug to be fixed or functionality to be developed.
+   The issue number is required for naming the development branch (see below).
  - Create a branch using the naming convention below.
-   If no issue number is available, create a research branch starting with `research/`.
    The frequency of updating your branch from main is up to personal taste.
    Yet, merge from main as often as possible, and merge back to main as early as possible.
- - Make and test the modifications.
- - Provide a patch-file, or create a pull request:
-   - Our Continuous Integration pipelines will only be triggered if the pull request is created by a Deltares contact person.
+ - Implement, test and document the modifications.
+   In case of changes by external contributors, this step will include pulling the changes from the external repository into the local branch and at least a security scan of the changes made by the external contributor.
+ - Create a pull request:
+   - Our Continuous Integration pipelines will be triggered automatically by a pull request created by Deltares employees.
      These pipelines consist of (Deltares-internal) TeamCity projects to build the source code (Windows and Linux) and subsequently a set of model simulation testbenches.
      A merge is only possible when all checks succeed.
      The projects will take at least 30 minutes to complete.
-   - You have to assign the Pull request to a core developer for reviewing and testing. When succeeded, the tester/reviewer is allowed to merge into main.
- - Official binary deliveries are only allowed using Deltares TeamCity server
+   - You have to assign the pull request to a core developer for review.
+     If review and all tests pass, the tester/reviewer is allowed to merge into main (signed Fiduciary License Agreement required in case of external contributor).
+ - Official binary deliveries are only allowed using the Deltares TeamCity server.
 
 ## Branch naming
 For each issue or feature, a separate branch should be created from the main.

src/engines_gpl/dflowfm/packages/dflowfm_kernel/src/dflowfm_data/m_heatfluxes.f90

@@ -68,7 +68,8 @@ module m_heatfluxes
    real(kind=dp), dimension(:), allocatable :: qtotmap
 
    ! Secchi depth variables
-   logical :: spatial_secchi_depth_is_available !< Flag to indicate if spatially varying Secchi depth is available
+   logical :: secchi_depth_is_spatially_varying !< Flag to indicate if spatially varying Secchi depth is available
+   logical :: secchi_depth_is_time_varying !< Flag to indicate if time-varying Secchi depth is available
    real(kind=dp), dimension(:), allocatable, target :: spatial_secchi_depth !< [m] Space-varying Secchi depth {"location": "face", "shape": ["ndx"]}
 
 contains

src/engines_gpl/dflowfm/packages/dflowfm_kernel/src/dflowfm_data/partition.F90

@@ -3543,6 +3543,28 @@ subroutine reduce_int1_max(var)
       return
    end subroutine reduce_int1_max
 
+!> for an array of doubles, take maximum over all subdomains (not over the array itself)
+   subroutine reduce_double_array_max(N, var)
+#ifdef HAVE_MPI
+      use mpi
+#endif
+
+      implicit none
+
+      integer, intent(in) :: N !< array size
+      real(kind=dp), dimension(N), intent(inout) :: var !< array with values to be reduced to global maximum over all subdomains
+
+      real(kind=dp), dimension(N) :: dum
+
+      integer :: ierror
+
+#ifdef HAVE_MPI
+      call MPI_allreduce(var, dum, N, mpi_double_precision, mpi_max, DFM_COMM_DFMWORLD, ierror)
+      var = dum
+#endif
+      return
+   end subroutine reduce_double_array_max
+
    !> for an array over integers, take maximum over all subdomains (not over the array itself)
    subroutine reduce_int_max(N, var)
 #ifdef HAVE_MPI
@@ -3740,8 +3762,8 @@ subroutine reduce_lateral_output()
       real(kind=dp), dimension(:, :), allocatable, save :: lateral_volume_per_layer_buffer
       real(kind=dp), dimension(:, :), allocatable, save :: outgoing_lat_volume_buffer
       real(kind=dp), dimension(:, :, :), allocatable, save :: outgoing_lat_concentration_buffer
-      real(kind=dp), dimension(:), allocatable, save  :: cumulative_value_buffer
-      real(kind=dp), dimension(:), allocatable, save  :: cumulative_weight_buffer
+      real(kind=dp), dimension(:), allocatable, save :: cumulative_value_buffer
+      real(kind=dp), dimension(:), allocatable, save :: cumulative_weight_buffer
       real(kind=dp), parameter :: dsmall = -huge(1.0_dp)
 
 #ifdef HAVE_MPI
@@ -3764,21 +3786,21 @@ subroutine reduce_lateral_output()
       if (average_waterlevels_per_lateral%is_used) then
          cumulative_value_buffer = average_waterlevels_per_lateral%cumulative_value
          cumulative_weight_buffer = average_waterlevels_per_lateral%cumulative_weight
-          call MPI_reduce(cumulative_value_buffer, average_waterlevels_per_lateral%cumulative_value, average_waterlevels_per_lateral%num_elements, mpi_double_precision, mpi_sum, 0, DFM_COMM_DFMWORLD, ierror)
-          call MPI_reduce(cumulative_weight_buffer, average_waterlevels_per_lateral%cumulative_weight, average_waterlevels_per_lateral%num_elements, mpi_double_precision, mpi_sum, 0, DFM_COMM_DFMWORLD, ierror)
-          if (my_rank == 0) then
-             do i_element = 1, average_waterlevels_per_lateral%num_elements
-                 average_waterlevels_per_lateral%values(i_element) = average_waterlevels_per_lateral%cumulative_value(i_element) / &
-                                                                     max(average_waterlevels_per_lateral%cumulative_weight(i_element), eps10)
-             end do
-          else
-              ! This is a work-around required to avoid issue in dimr.cpp send() i.e. when reducing negative values
-              average_waterlevels_per_lateral%values = dsmall
-          end if
+         call MPI_reduce(cumulative_value_buffer, average_waterlevels_per_lateral%cumulative_value, average_waterlevels_per_lateral%num_elements, mpi_double_precision, mpi_sum, 0, DFM_COMM_DFMWORLD, ierror)
+         call MPI_reduce(cumulative_weight_buffer, average_waterlevels_per_lateral%cumulative_weight, average_waterlevels_per_lateral%num_elements, mpi_double_precision, mpi_sum, 0, DFM_COMM_DFMWORLD, ierror)
+         if (my_rank == 0) then
+            do i_element = 1, average_waterlevels_per_lateral%num_elements
+               average_waterlevels_per_lateral%values(i_element) = average_waterlevels_per_lateral%cumulative_value(i_element) / &
+                                                                   max(average_waterlevels_per_lateral%cumulative_weight(i_element), eps10)
+            end do
+         else
+            ! This is a work-around required to avoid issue in dimr.cpp send() i.e. when reducing negative values
+            average_waterlevels_per_lateral%values = dsmall
+         end if
       end if
 #endif
       return
-    end subroutine reduce_lateral_output    
+   end subroutine reduce_lateral_output
 
    !> Distribute lateral input to all ranks
    subroutine distribute_lateral_input()
@@ -3813,8 +3835,8 @@ subroutine distribute_lateral_input()
       call MPI_bcast(incoming_lat_concentration, num_lateral_layer_constituent, mpi_double_precision, 0, DFM_COMM_DFMWORLD, ierror)
 #endif
       return
-   end subroutine distribute_lateral_input    
-    
+   end subroutine distribute_lateral_input
+
 !> reduce outputted values at observation stations
 !! NOTE: It seems that, now that we reduce the statistical output before writing, this routine is
 !!       only needed to maintain functionality in unstruc_bmi/get_compound_field
@@ -6389,4 +6411,54 @@ subroutine logical_and_across_partitions(val, allval)
       end if
    end subroutine logical_and_across_partitions
 
+!> Given a list of local flow cell indices, returns a list of local flow cell numbers at their global position in the global union,
+! and -1 for the cells that lie on the other partitions.
+   function reduce_cells(local_cells, ndx) result(global_cells)
+#ifdef HAVE_MPI
+      use mpi
+#endif
+
+      integer, dimension(:), intent(in) :: local_cells !< Local flow cell indices found on this partition
+      integer, intent(in) :: ndx !< number of flow cells (internal + boundary), should match ndx in m_flowgeom
+      integer, dimension(:), allocatable :: global_cells !< Local flow cell indices of the global union
+      integer, dimension(:), allocatable :: global_cellmask, ilocal_s
+      integer :: k, num_cells
+#ifdef HAVE_MPI
+      integer :: mpi_err
+#endif
+
+      allocate (global_cellmask(nglobal_s))
+      global_cellmask = 0
+      ! Mark locally present cells in global cellmask, reduce afterwards
+      global_cellmask(iglobal_s(local_cells)) = 1
+
+#ifdef HAVE_MPI
+      call MPI_Allreduce(MPI_IN_PLACE, global_cellmask, nglobal_s, &
+                         MPI_INTEGER, MPI_MAX, DFM_COMM_DFMWORLD, mpi_err)
+#endif
+
+      num_cells = count(global_cellmask == 1)
+      allocate (global_cells(num_cells))
+
+      ! iglobal_s contains global numbers of local cells, but required are local numbers of global cells, so build an inverse mapping.
+      allocate (ilocal_s(nglobal_s))
+      ilocal_s = -1
+      do k = 1, ndx
+         if (iglobal_s(k) > 0) then
+            ilocal_s(iglobal_s(k)) = k
+         end if
+      end do
+
+      num_cells = 0
+      ! Build global cells from ilocal_s by iterating over global_cellmask.
+      ! Cells that exist globally not on current partition will have -1 in ilocal_s and thus -1 in global_cells.
+      do k = 1, nglobal_s
+         if (global_cellmask(k) == 1) then
+            num_cells = num_cells + 1
+            global_cells(num_cells) = ilocal_s(k)
+         end if
+      end do
+
+   end function reduce_cells
+
 end module m_partitioninfo

src/engines_gpl/dflowfm/packages/dflowfm_kernel/src/dflowfm_data/unstruc_inifields.f90

@@ -1575,15 +1575,14 @@ subroutine process_parameter_block(qid, inifilename, target_location_type, time_
       use timespace_parameters, only: NCGRID
       use m_missing, only: dmiss
       use fm_location_types, only: UNC_LOC_S, UNC_LOC_U, UNC_LOC_CN, UNC_LOC_GLOBAL, UNC_LOC_S3D
-      use m_flowparameters, only: jatrt, javiusp, jafrcInternalTides2D, jadiusp, jafrculin, jaCdwusp, ibedlevtyp, jawave, waveforcing
-      use m_flowparameters, only: ja_friction_coefficient_time_dependent
-      use m_flow, only: frcu
-      use m_flow, only: jacftrtfac, cftrtfac, viusp, diusp, DissInternalTidesPerArea, frcInternalTides2D, frculin, Cdwusp
+      use m_flowparameters, only: jatrt, javiusp, jafrcInternalTides2D, jadiusp, jafrculin, jaCdwusp, ibedlevtyp, jawave, &
+         waveforcing, ja_friction_coefficient_time_dependent
+      use m_flow, only: frcu, jacftrtfac, cftrtfac, viusp, diusp, DissInternalTidesPerArea, frcInternalTides2D, frculin, Cdwusp
       use m_flowgeom, only: ndx, lnx, grounlay, iadv, jagrounlay, ibot
       use m_lateral_helper_fuctions, only: prepare_lateral_mask
       use fm_external_forcings_data, only: success
       use fm_external_forcings_utils, only: split_qid
-      use m_heatfluxes, only: spatial_secchi_depth
+      use m_heatfluxes, only: spatial_secchi_depth, secchi_depth_is_time_varying
       use m_wind, only: wind_drag_type, CD_TYPE_CONST
       use m_fm_icecover, only: ja_ice_area_fraction_read, ja_ice_thickness_read, fm_ice_activate_by_ext_forces
       use m_meteo, only: ec_addtimespacerelation
@@ -1736,6 +1735,10 @@ subroutine process_parameter_block(qid, inifilename, target_location_type, time_
          call realloc(spatial_secchi_depth, ndx, keepExisting=.true., fill=dmiss, stat=ierr)
          target_location_type = UNC_LOC_S
          target_array => spatial_secchi_depth
+         if (filetype == NCGRID) then
+            time_dependent_array = .true.
+            secchi_depth_is_time_varying = .true.
+         end if
       case ('backgroundverticaleddydiffusivitycoefficient')
          target_location_type = UNC_LOC_S
          call realloc(dicoww, ndx, constant_dicoww)
@@ -2008,7 +2011,7 @@ subroutine finish_initialization(qid)
       use m_grw, only: jaintercept2D
       use m_fm_icecover, only: ja_ice_area_fraction_read, ja_ice_thickness_read
 
-      use m_heatfluxes, only: spatial_secchi_depth_is_available, spatial_secchi_depth
+      use m_heatfluxes, only: secchi_depth_is_spatially_varying, spatial_secchi_depth
       use m_physcoef, only: secchi_depth
       use m_meteo, only: ec_addtimespacerelation
       use m_vegetation, only: stemheight, stemheightstd
@@ -2066,7 +2069,7 @@ subroutine finish_initialization(qid)
       case ('sea_ice_thickness')
          ja_ice_thickness_read = 1
       case ('secchidepth')
-         spatial_secchi_depth_is_available = .true.
+         secchi_depth_is_spatially_varying = .true.
          do n = 1, ndx
             if (spatial_secchi_depth(n) == dmiss) then
                spatial_secchi_depth(n) = secchi_depth(1)