Skip to content

Latest commit

 

History

History
135 lines (92 loc) · 5.12 KB

README.md

File metadata and controls

135 lines (92 loc) · 5.12 KB
Raw

(Very) Simple Soil Hydrology Model

DOI

A very simple, point-scale soil hydrology model based on the modified Richards equation from the Community Land Model (CLM), version 5.0, with some equations from CLM 4.x and pedotransfer functions from Balland et al. (2008). The basic model characterizes the vertical movement of water in a soil column and lateral sub-surface drainage at sub-daily time steps, driven by:

  • Initial soil volumetric water content (VWC) state;
  • Mean daily infiltration rate (kg m-2 sec-1);
  • Mean daily potential transpiration (kg m-2 sec-1);
  • Mean soil temperature in each layer;
  • Soil texture and porosity;
  • Fraction of land area that is saturated;

The soil column is considered to be at a point location; i.e., there is no connectivity between lateral sub-surface runoff and any other soil column. The approach here is designed to produce reasonable results for a wide variety of soils. The soil hydrology model has been tested using soil infiltration rates, soil texture, and soil temperature data from the SMAP Level 4 Soil Moisture (L4SM) product and the Catchment land model (Koster et al. 2000).

Significance:

This is a small library that enables the generation of reasonable soil moisture profile time series data at point-scale. With this library, provided you have reasonable surface meteorology driver data, it is not necessary to install, configure, and spin-up a more sophisticated terrestrial biosphere model or land model. Your feedback is welcome! Get in touch with me or submit a Pull Request.

If this library doesn't suit your needs, you might check out these related Python libraries:

Documentation

Read the online documentation here.

Dependencies

  • Python 3.5+
  • scipy 1.3+
  • numpy 1.13.3+
  • cached_property 1.5.1+

Installation

It is recommended that you install within either a virtualenv or conda virtual environment.

$ python setup.py install .

To install using pip:

$ pip install .

To install simsoil in "development mode," which enables you to edit the source code:

$ pip install -e .

Tests can be run by:

$ python simsoil/tests/tests.py

Note that the test suite may take up to a minute to complete.

Getting Started

It's generally necessary to spin-up each soil column to a quasi-equilibrium state, beginning with an arbitrary guess at the initial soil VWC state, e.g., 0.15 m3 m-3. Potential transpiration (PET) can be calculated using the simsoil.transpiration module or taken from MODIS MOD16.

from simsoil.core import InfiltrationModel, SoilModel, DEPTHS, SOC_RATIOS

# Define the soil column
soil = SoilProfile(
  pft = 1, soc = SOC_RATIOS * 1e3, sand = 0.6, clay = 0.1,
  porosity = 0.4, bedrock = 3, slope = 0.01, depths_m = DEPTHS)

model = InfiltrationModel(soil_model = soil)

# e.g., spin-up for 20 years
n_days = 20 * 365

# e.g., Initial soil VWC in each layer = 0.15 m3 m-3
init_vwc = np.ones(DEPTHS.shape) * 0.15

# With adaptive = False, do not use adaptive time stepping; sub-daily
#   time steps will always be hourly (3600 seconds);
# With climatology = True, we can provide a 365-day climatology for
#   temperature, PET, and infiltration that will be recycled every year
results = model.run(
  init_vwc, temp_profile, pet, influx, f_sat, dt = 3600,
  n_days = n_days, climatology = True, adaptive = False)

Suggested driver datasets:

  • Daily infiltration rate, e.g., from SMAP L4SM Geophysical_Data/soil_water_infiltration_flux field.
  • Soil temperature in each layer, e.g., from SMAP L4SM Geophysical_Data/soil_temp_layer* fields (if using same soil layers as the Catchment model).
  • Soil texture (sand, clay content) and porosity from SMAP L4SM Land-Model-Constants_data.
  • Soil texture (sand, clay, and soil organic carbon content) from SoilGrids 250m.

Debugging

In InfiltrationModel.step_daily(), it may be useful to add interactive debugging (after installing ipdb); for example, the following code block might be inserted anywhere after the call to solve_vwc():

# DEBUG: Tracking water flows can be helpful
if self._debug:
    vwc1 = np.add(vwc, x)
    vwc2 = vwc1 + runoff # Runoff is negative
    try:
        vwc3 = rebalance(vwc2, temp_profile, thickness_mm)
    except AssertionError:
        # i.e., rebalance() exceeded max. number of operations
        import ipdb
        ipdb.set_trace()
    if not np.logical_and(0 <= vwc3, vwc3 <= 1).all():
        # i.e., soil VWC exceeds physical limits
        import ipdb
        ipdb.set_trace()

Acknowledgments

Thanks to Colin Brust for providing the MOD16 source code for reference. Thanks also to Randy Koster and Rolf Reichle (NASA GMAO) for their feedback on an early version of this work.