Skip to content

BDSNP Soil NO Emissions

Overview

The BDSNP module computes soil nitrogen oxide (NO) emissions and writes them to the export state for consumption by MEGAN3 or other schemes. It replaces the previous soil_nox scheme with a more comprehensive parameterization.

Two algorithms are supported:

  • BDSNP (default) — Berkeley-Dalhousie Soil NO Parameterization with biome-specific emission factors, piecewise-linear soil moisture dependence, nitrogen deposition fertilization, and canopy reduction
  • YL95 — Yienger and Levy (1995) empirical model with exponential temperature response and Poisson-like moisture function

Both modes set emissions to zero when soil temperature is below 0°C.

References: - Hudman et al. (2012), Berkeley-Dalhousie Soil NO Parameterization - Yienger, J.J. and H. Levy II (1995), JGR, 100(D6), 11447–11464

Registration Names

  • Native C++: "bdsnp"
  • Fortran bridge: "bdsnp_fortran"

Configuration

physics_schemes:
  - name: bdsnp
    options:
      soil_no_method: bdsnp       # "bdsnp" (default) or "yl95"
      # BDSNP-specific parameters
      fert_emission_factor: 1.0
      wet_dep_scaling: 1.0
      dry_dep_scaling: 1.0
      pulse_decay_constant: 0.5
      # YL95 parameters (also used as fallback)
      biome_coefficient_wet: 0.5
      temp_limit: 30.0
      temp_exp_coeff: 0.103
      wet_coeff_1: 5.5
      wet_coeff_2: -5.55
    input_mapping:
      soil_temperature: TSOIL
      soil_moisture: GWETTOP
      nitrogen_deposition: NDEP
      land_use_type: LANDTYPE
      leaf_area_index: LAI
      biome_emission_factors: BIOME_EF
    output_mapping:
      soil_nox_emissions: SOIL_NO

Parameters

YAML Key Type Default Description
soil_no_method string "bdsnp" Algorithm: "bdsnp" or "yl95"
fert_emission_factor double 1.0 Fertilizer emission factor scaling (BDSNP)
wet_dep_scaling double 1.0 Wet deposition scaling factor (BDSNP)
dry_dep_scaling double 1.0 Dry deposition scaling factor (BDSNP)
pulse_decay_constant double 0.5 Pulsing decay constant (BDSNP)
biome_coefficient_wet double 0.5 Biome emission coefficient (YL95)
temp_limit double 30.0 Max temperature for emission [°C] (YL95)
temp_exp_coeff double 0.103 Exponential temperature coefficient (YL95)
wet_coeff_1 double 5.5 Moisture response coefficient 1 (YL95)
wet_coeff_2 double -5.55 Moisture response coefficient 2 (YL95)

Import Fields

Field Name Units Description
soil_temperature K Soil temperature
soil_moisture fraction Soil moisture [0–1]
nitrogen_deposition kg N/m²/s N deposition rate (BDSNP only)
land_use_type — Land use category (BDSNP only)
leaf_area_index m²/m² LAI for canopy reduction (BDSNP only)
biome_emission_factors — Biome-specific base emission (BDSNP only)

Export Fields

Field Name Units Description
soil_nox_emissions kg NO/m²/s Soil NO emission flux

Algorithms

YL95 Mode

  1. Convert temperature: tc = T_soil − 273.15
  2. If tc ≤ 0: emission = 0 (freezing cutoff)
  3. Temperature factor: t_term = exp(0.103 × min(30, tc))
  4. Moisture factor: w_term = 5.5 × gw × exp(−5.55 × gw²)
  5. Emission: soil_NO = a_biome × UNITCONV × t_term × w_term

BDSNP Mode

  1. Convert temperature: tc = T_soil − 273.15
  2. If tc ≤ 0: emission = 0 (freezing cutoff)
  3. Temperature response: t_response = exp(0.103 × min(30, tc))
  4. Moisture factor (piecewise linear):
  5. SM ≤ 0: 0
  6. SM ≤ 0.3: SM / 0.3
  7. SM > 0.3: 1.0 − 0.5 × (SM − 0.3) / 0.7
  8. N-deposition fertilization: fert = 1 + fert_ef × ndep × (wet_dep + dry_dep)
  9. Canopy reduction: canopy = exp(−0.24 × LAI)
  10. Emission: soil_NO = base_ef × UNITCONV × t_response × sm_factor × fert × canopy

Integration with MEGAN3

BDSNP writes to the soil_nox_emissions export field. MEGAN3 reads this field for the NO emission class. The stacking engine must execute BDSNP before MEGAN3:

physics_schemes:
  - name: bdsnp        # Runs first
  - name: megan3       # Runs second, reads soil_nox_emissions

If soil_nox_emissions is not present when MEGAN3 runs, the NO class contribution is set to zero with a warning.

Implementation Notes

  • Available as both native C++ (Kokkos) and Fortran bridge implementations
  • The pulse factor in BDSNP mode is a stateless placeholder (no rain history tracking)
  • Both modes produce numerically identical results between C++ and Fortran within 1e-6 tolerance