MESH Model Guide#

Overview #

MESH (Modélisation Environmentale Communautaire - Surface and Hydrology / Community Environmental Modeling System for Surface Hydrology) is an integrated land surface-hydrological modeling system developed by Environment and Climate Change Canada (ECCC). MESH couples detailed land surface schemes with hydrological routing to provide comprehensive water and energy budget simulations.

Key Capabilities:

Coupled land surface-hydrology modeling
Multiple land surface schemes (CLASS, SVS)
Tile-based land heterogeneity representation
Detailed cold region processes (permafrost, organic soils)
Lakes and wetlands
Routing with WATROUTE or WATFLOOD
Energy and water budgets
Applicable from headwaters to large river basins

Typical Applications:

Cold region hydrology (snow, permafrost, frozen soil)
Wetland-dominated landscapes
Large basin water budgets
Climate change impact studies
Operational flood forecasting (Canada)
Land-atmosphere coupling studies
Agricultural water management

Spatial Scales: Subcatchment (10 km²) to large basins (100,000+ km²)

Temporal Resolution: Sub-hourly to daily

Model Structure #

Land Surface Schemes #

MESH supports multiple physics options:

CLASS (Canadian Land Surface Scheme):

Standard MESH land surface module
3-layer soil model
Detailed snow physics
Canopy processes
Permafrost capable

SVS (Soil, Vegetation, Snow):

Alternative to CLASS
Multi-layer soil and snow
Urban physics
Glacier representation

Spatial Discretization #

Grouped Response Units (GRUs):

Similar to SUMMA, MESH uses GRUs representing subcatchments.

Tiles within GRUs:

Each GRU contains tiles representing land cover types:

GRU 1 (Subcatchment):
├─ Tile 1: Needleleaf Forest (40%)
├─ Tile 2: Cropland (35%)
├─ Tile 3: Wetland (15%)
└─ Tile 4: Urban (10%)

GRU 2:
├─ Tile 1: Grassland (60%)
└─ Tile 2: Cropland (40%)

Process Representation #

Snow:

Multi-layer snowpack
Snow densification and albedo evolution
Blowing snow sublimation
Snow interception by canopy

Soil:

3+ layer moisture and temperature
Frozen soil physics
Infiltration and percolation
Subsurface lateral flow

Evapotranspiration:

Energy balance approach
Stomatal conductance
Soil moisture stress

Runoff Generation:

Infiltration excess
Saturation excess
Subsurface flow
Wetland storage and release

Routing:

WATROUTE or WATFLOOD
Channel routing
Lake routing
Reservoirs (optional)

Configuration in SYMFLUENCE #

Model Selection #

HYDROLOGICAL_MODEL: MESH

Key Configuration Parameters #

Installation and Execution#

Parameter	Default	Description
MESH_INSTALL_PATH	default	Path to MESH executable
MESH_EXE	mesh.exe	MESH executable name
EXPERIMENT_OUTPUT_MESH	default	Output directory

Spatial Configuration#

Parameter	Default	Description
MESH_SPATIAL_MODE	distributed	Spatial mode (lumped, distributed)
MESH_GRU_DIM	default	GRU dimension name
MESH_HRU_DIM	default	HRU (tile) dimension name

Forcing Configuration#

Parameter	Default	Description
MESH_FORCING_PATH	default	Path to forcing files
MESH_FORCING_VARS	default	Forcing variable names
MESH_FORCING_UNITS	default	Forcing units
MESH_FORCING_TO_UNITS	default	Target units for conversion

Data Configuration#

Parameter	Default	Description
MESH_DDB_VARS	default	Drainage database variables
MESH_DDB_UNITS	default	Database units
MESH_DDB_MIN_VALUES	default	Minimum acceptable values
MESH_DS_MAIN_ID	default	Main drainage segment ID

Input File Specifications #

MESH uses NetCDF for most inputs and outputs.

Drainage Database (DDB)#

File: MESH_drainage_database.nc

Contains GRU and tile attributes:

dimensions:
  gru = 50      ! Number of subcatchments/GRUs
  tile = 5      ! Max tiles per GRU

variables:
  int gru_id(gru)
  float gru_area(gru)          ! km²
  float gru_elevation(gru)     ! m
  float gru_slope(gru)         ! degrees
  float gru_latitude(gru)
  float gru_longitude(gru)

  int tile_type(gru, tile)     ! Land cover type
  float tile_frac(gru, tile)   ! Fraction of GRU
  float tile_slope(gru, tile)
  float tile_aspect(gru, tile)

Forcing Data (NetCDF)#

File: MESH_forcing.nc

dimensions:
  time = UNLIMITED
  gru = 50

variables:
  double time(time)
  float FSIN(time, gru)        ! Shortwave radiation [W/m²]
  float FLIN(time, gru)        ! Longwave radiation [W/m²]
  float PRE(time, gru)         ! Precipitation [kg/m²/s]
  float TA(time, gru)          ! Air temperature [K]
  float QA(time, gru)          ! Specific humidity [kg/kg]
  float UV(time, gru)          ! Wind speed [m/s]
  float PRES(time, gru)        ! Pressure [Pa]

Initial Conditions #

File: MESH_initial_conditions.ini

Text file with initial states:

! Snow states
SNO  50.0    ! Snow water equivalent [mm]
TSNO -2.0    ! Snow temperature [°C]

! Soil states
TBAR  3.5    ! Soil temperature [°C]
THLQ  0.25   ! Liquid soil moisture [-]
THIC  0.05   ! Frozen soil moisture [-]

Output File Specifications #

Standard Output (NetCDF)#

File: MESH_output_Basin_average.nc

Basin-averaged fluxes:

dimensions:
  time = UNLIMITED

variables:
  double time(time)
  float RUNOFF(time)           ! Total runoff [mm]
  float RECHARGE(time)         ! Groundwater recharge [mm]
  float EVAP(time)             ! Evapotranspiration [mm]
  float SNO(time)              ! Snow water equivalent [mm]
  float TBAR(time)             ! Mean soil temperature [K]
  float THLQ(time)             ! Liquid soil moisture [-]

GRU-level Output (NetCDF)#

File: MESH_output_Grid.nc

Spatially distributed output:

dimensions:
  time = UNLIMITED
  gru = 50

variables:
  float RUNOFF(time, gru)
  float SNO(time, gru)
  float EVAP(time, gru)
  float TBAR(time, gru)
  float THLQ(time, gru)

Streamflow Output (Text)#

File: MESH_streamflow_Gauge_###.txt

YEAR  MONTH  DAY  HOUR  STREAMFLOW_m3s
2015  1      1    0     45.32
2015  1      1    1     45.18
2015  1      1    2     45.05
...

Model-Specific Workflows #

Basic MESH Workflow #

# config.yaml
DOMAIN_NAME: canadian_basin
HYDROLOGICAL_MODEL: MESH

# Domain
DOMAIN_DEFINITION_METHOD: semidistributed
POUR_POINT_COORDS: [-95.0, 54.0]  # Manitoba

# Forcing
FORCING_DATASET: RDRS  # Canadian reanalysis (recommended for Canada)
FORCING_START_YEAR: 2010
FORCING_END_YEAR: 2020

# MESH configuration
MESH_INSTALL_PATH: /path/to/mesh
MESH_SPATIAL_MODE: distributed

Run:

symfluence workflow run --config config.yaml

Cold Region MESH Application #

For permafrost and seasonal frost:

# config.yaml
HYDROLOGICAL_MODEL: MESH

# Cold region domain (Arctic/Subarctic)
DOMAIN_DEFINITION_METHOD: semidistributed
POUR_POINT_COORDS: [-115.0, 67.0]  # Northwest Territories

# Discretization with permafrost classes
DISCRETIZATION:
  soilclass:
    include_permafrost: true
    sources: ['SoilGrids', 'PermafrostMap']
  landclass:
    sources: ['MODIS']

# CLASS land surface scheme (better for cold regions)
MESH_LAND_SURFACE_SCHEME: CLASS

Wetland-Dominated Basin #

For Prairie Pothole or boreal wetlands:

# config.yaml
HYDROLOGICAL_MODEL: MESH

# Wetland representation
DISCRETIZATION:
  landclass:
    sources: ['MODIS', 'WetlandMap']
    include_wetland_class: true

# WATFLOOD routing (handles wetlands well)
MESH_ROUTING_SCHEME: WATFLOOD

Large Basin Application #

For major river basins (e.g., Mackenzie, Nelson):

# config.yaml
DOMAIN_DEFINITION_METHOD: merit_basins
MERIT_BASIN_IDS: [...]  # 100+ subcatchments

HYDROLOGICAL_MODEL: MESH
MESH_SPATIAL_MODE: distributed

# Parallel execution
NUM_PROCESSES: 32

# Routing essential for large domain
MESH_ROUTING_SCHEME: WATROUTE

Calibration Strategies #

MESH has many parameters, but key ones for calibration:

Recommended Parameters #

Runoff generation:

# WATFLOOD parameters
R2 N:  Manning's n for overland flow (0.01-0.3)
RILL:  Rill storage (0.0-10.0 mm)

# CLASS parameters
ZSNL:  Limiting snow depth for albedo (0.01-0.1 m)
ZPLG:  Maximum ponding depth (0.001-0.01 m)

Soil parameters (per soil class):

Hydraulic conductivity
Field capacity
Wilting point
Porosity

Routing parameters:

Channel Manning’s n (0.01-0.1)
Channel geometry (if not from DEM)

Snow parameters:

Snow albedo decay rate
Fresh snow albedo

Parameter Bounds #

Typical ranges for key parameters:

# Overland flow
R2N:      [0.01, 0.3]       # Manning's n overland
RILL:     [0.0, 10.0]       # Rill storage [mm]

# Soil hydraulic
SAND_Ksat: [1e-6, 1e-4]     # Sandy soil Ksat [m/s]
CLAY_Ksat: [1e-7, 1e-5]     # Clay soil Ksat [m/s]

# Channel routing
R2CH:     [0.01, 0.1]       # Manning's n channel

# Snow
ZSNL:     [0.01, 0.1]       # Snow depth for albedo [m]
ZPLS:     [0.1, 10.0]       # Max snow interception [mm]

Known Limitations #

Complexity:
- Steep learning curve
- Many input files and parameters
- Detailed land surface data required
Computational Cost:
- Energy balance calculations expensive
- Tile-based approach multiplies computations
- Large domains require significant resources
Data Requirements:
- Detailed land cover and soil maps
- All forcing variables needed (8 variables)
- Lakes and wetland databases
Limited Adoption Outside Canada:
- Primarily used in Canada
- Documentation focused on Canadian applications
- Less community support compared to international models
Calibration Challenges:
- High-dimensional parameter space
- Parameters interact complexly
- Requires expert knowledge

Troubleshooting #

Common Issues #

Error: “MESH executable not found”

# Verify installation
MESH_INSTALL_PATH: /absolute/path/to/mesh/bin
MESH_EXE: mesh.exe

Error: “Drainage database format error”

Check NetCDF structure:

ncdump -h MESH_drainage_database.nc

# Verify dimensions: gru, tile
# Verify required variables

Error: “Forcing variable not found”

# Check forcing NetCDF
ncdump -h MESH_forcing.nc

# Ensure all 7 variables: FSIN, FLIN, PRE, TA, QA, UV, PRES

Simulation crashes

Check initial conditions (reasonable ranges)
Verify tile fractions sum to 1.0 per GRU
Check forcing for missing data or extremes
Reduce timestep if unstable

Poor performance in cold regions

# Use CLASS scheme
MESH_LAND_SURFACE_SCHEME: CLASS

# Include permafrost classes
DISCRETIZATION:
  soilclass:
    include_permafrost: true

Wetlands not working

# Ensure wetland tile type defined
DISCRETIZATION:
  landclass:
    include_wetland_class: true

# Use WATFLOOD routing
MESH_ROUTING_SCHEME: WATFLOOD

Performance Optimization #

Speed up execution:

Reduce number of tiles (aggregate similar classes)
Coarser timestep (hourly vs 30-min)
Simplify routing (use WATROUTE instead of WATFLOOD)
Parallel execution (MPI-enabled MESH)

Improve accuracy:

Detailed tile representation (more classes)
Finer timestep (30-min or 1-hr)
Include lakes explicitly
Use high-resolution DEM for slopes

Additional Resources #

MESH Documentation:

Official site: https://wiki.usask.ca/display/MESH
User Manual: https://wiki.usask.ca/display/MESH/MESH+Documentation
GitHub: MESH-Model/MESH-Dev

CLASS Documentation:

CLASS manual: https://cccma.gitlab.io/classic/

Publications:

Pietroniro et al. (2007): “Development of the MESH modelling system for hydrological ensemble forecasting” https://doi.org/10.1016/j.jhydrol.2006.07.015
Haghnegahdar et al. (2017): “Multicriteria sensitivity analysis as a diagnostic tool for understanding MESH model behavior” https://doi.org/10.1002/hyp.11358

SYMFLUENCE-specific:

Configuration: MESH parameter reference
Calibration and Optimization: Calibration best practices
SUMMA Model Guide: Comparison with SUMMA
Troubleshooting: General troubleshooting

Canadian Datasets:

RDRS (Regional Deterministic Reanalysis System): Forcing data for Canada https://collaboration.cmc.ec.gc.ca/cmc/cmoi/product_guide/
CanVec: Canadian vector geospatial data
Canadian Digital Elevation Model (CDEM)

Training:

University of Saskatchewan MESH training
ECCC workshops (for Canadian users)

Applications:

Used in Canada’s operational flood forecasting
Provincial water resource management (Saskatchewan, Alberta)
Climate change impact studies across Canada