The
mesoscale
Hydrologic Model
mHM

Cite as:
Cite the code:

Release Notes

Code repository and list of publications

The mHM GitLab repository grants read access to the code. GitLab now allows you to report bugs or issues publicly.

Developers that contribute to the code will be incorporated into the list of authors and will appear in the DOI of the next mHM version.

Please find a list of publications in mhm/doc/mhm_papers.md.

The mHM Team. Photo: Sebastian Wiedling/UFZ
A small "sample" of the team behind the mHM development (unfortunately, not all collaborators were at that day at the UFZ, but their names are below.). Photo: Sebastian Wiedling/UFZ

The list of collaborators behind the development of the mHM source code (Fortran), its data handling routines, pre/post-processors, git repository, the CHS Fortran/Bash/Python libraries, and the key cluster/workstation software maintenance are (listed in chronological order, starting in 2006) :


The mesoscale hydrologic model (mHM) developed by our group is a spatially explicit distributed hydrologic model that uses grid cells as a primary hydrologic unit, and accounts for the following processes: canopy interception, snow accumulation and melting, soil moisture dynamics, infiltration and surface runoff, evapotranspiration, subsurface storage and discharge generation, deep percolation and baseflow and discharge attenuation and flood routing.

Schematic Representation of the Mesoscale Hydrologic Model
Fig. 1: Schematic of resolution levels, data, processes and states in mHM.

The model is driven by hourly or daily meteorological forcings (e.g., precipitation, temperature), and it utilizes observable basin physical characteristics (e.g., soil textural, vegetation, and geological properties) to infer the spatial variability of the required parameters.

The main feature of mHM is the approach to estimate parameters at the target resolution based on high resolution physiographic land surface descriptors (e.g., DEM, slope, aspect, root depth based on land cover class or plant functional types, LAI, soil texture, geological formation type) . The technique was proposed in Samaniego et al WRR 2010 and is called multiscale parameter regionalization (MPR). The MPR technique is crucial to reach flux-matching across scales and to derive seamless parameter fields (Samaniego et al HESS 2017).

To date, the model has been successfully applied and cross-validated in more than 220 basins in Germany (Zink et al. HESS 2017), 300 Pan EU basins (Fig 2) (Samaniego et al. BAMS 2019), as well as India, and USA (Fig. 3, 4) (Rakovec et al 2019), ranging in size from 4 to 550,000 km2 at spatial resolutions (or grid size) varied between 1 km and 100 km. Shown below is the model performance for stream flow simulations over the EU basins.

Performace EU
Fig. 2: mHM performance for daily streamflow simulations over 1266 European basins. High KGE (blue) corresponds to a good fit of simulated discharge with observations. Median KGE is 0.56.
USA CONUS on site
Fig 3. Performance of the CONUS on site parameter estimation over CAMELS basins
CONUS-wide
Fig. 4: Performance of the CONUS-wide (i.e., a single parameter set for all basins) parameter estimation over the CAMELS basins. Lack of model performance in the center of the US is a well know fact. The main reason is the sparse data records used to derive the forcing data sets, as well as prairie potholes that store surface water and are hydrologically disconnected from the basin outlet due to poorly defined drainage systems which are not represented in the model (Mizukami et al. 2017 WRR).

Questions, Bug Report, Suggestions

If you have technical questions concerning mHM, found a bug during your operations or have any suggestions how to improve mHM, please let us know!