LDNDC

> **MANDATORY EXECUTION POLICY** — READ BEFORE PROCEEDING
>
> You MUST run the **actual model binary or package** described in this document.
> If the model fails to import, compile, or execute, you MUST:
> 1. Check `diagnostics/triplets.yaml` for a matching error pattern
> 2. Try the fix suggested in the triplet's `remedy` section
> 3. If still failing, report the error to the user with full details
>
> You MUST NOT substitute a simplified Python formula, regression equation,
> or hand-coded approximation in place of the real model. Doing so produces
> scientifically invalid results and defeats the purpose of the KI.
>
> Before starting, run: `python preflight_check.py` (in this KI directory)

> **CMFD direct reader available:** Use `from ki_tools_common.netcdf_utils import load_cmfd_daily_all` to read CMFD 3-hourly data directly. Returns daily precip (mm), temp (°C with Tmin/Tmax), radiation (W/m²), wind, humidity. Handles subdirectory search (Prec/, Temp/, etc.) and unit conversions automatically.
> to verify that the model binary/package and required data are available.
>
> **DEBUGGING PROTOCOL** — When something goes wrong, follow this order:
> 1. **Check triplets** — `diagnostics/triplets.yaml` may already cover this error
> 2. **Read official docs** — The model's own documentation for expected formats/units
> 3. **Find working examples** — Check `outputs/` or the model's shipped test data
> 4. **Fix the tool** — With knowledge of what "correct" looks like
>
> Do NOT write custom debug scripts. The answers are in the docs and examples.

<!-- KI-MAP:BEGIN (projected by generate_skill_map.py — edit the KI, not this table) -->
## KI map — what to read, and when

| when you need | read | why |
|---|---|---|
| FIRST, always | `preflight_check.py` | run it (`python preflight_check.py`): proves env/binary/data are usable and emits a machine-readable `PREFLIGHT_REPORT=` line. Do not debug a run that never had a healthy environment. |
| to run the pipeline stages | `tools/` (20 tools) | the executable pipeline. Read each tool's argparse (`--help`) before composing a command; SKILL.md's stage table says which tool serves which stage. |
| before running a stage | `docs/s*_*.md` (12 stage docs) | per-stage procedure, verification and traps — the how-to that SKILL.md's overview compresses. |
| on ANY error, before debugging | `diagnostics/triplets.yaml` (27 entries) | symptom → diagnosis → remedy for this model's known failure modes. Check here FIRST; the answer usually exists. Never renumber or rewrite entries. |
| to know what an output IS | `dag.yaml` | the model's identity: every output's medium, units, `validation_rank` (1 = the headline variable) and observability. Scoring and obs-binding read THIS — when asked 'what does this model predict', the dag is the answer, not a guess. |
| when building inputs / parsing outputs | `docs/format_spec.yaml` | exact I/O shapes + `known_issues`, projected from dag + triplets. Regenerate with `ki_tools_common/generate_format_spec.py` after changing either — never hand-edit. |
| to judge a run's skill | `docs/validation_convention.yaml` | how this model's field judges it validated: per-`dag_variable` metrics, directions and CITED pass-bands. A run is graded against these, not against intuition. |
| for claims and thresholds | `docs/gathered_papers.json` (20 papers) + `docs/papers_index.md` | the literature this KI is judged by; each entry's `text_path` is fetched full text in the central paper cache. `role: benchmark` marks the model's own skill paper. |
| for a machine-readable summary | `knowledge_infrastructure.yaml` | the manifest (package, pipeline, validation tier, counts) — projected by `ki_tools_common/generate_ki_manifest.py`; regenerate after structural changes, never hand-edit. |

*Projected 2026-08-17 from the KI's actual contents — 9 components present. Refresh: `python3 ki_tools_common/generate_skill_map.py --ki_dir <this KI>`.*
<!-- KI-MAP:END -->

<!-- KI-TOOL-INDEX:BEGIN (projected by generate_skill_map.py — the discoverability contract: every public tool, exact path; PURPOSE stays human-authored elsewhere) -->
### Executable tool index (projected — complete by construction)

Every public tool in this KI, by exact path. What each is FOR lives in the
human-written Tool Inventory above; `--help` on any of these prints its arguments.

| tool (exact path) | invocation |
|---|---|
| `tools/run_ldndc_bengbu_ghg.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/run_ldndc_bengbu_ghg.py --help` |
| `tools/run_ldndc_rice_paddy_ch4.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/run_ldndc_rice_paddy_ch4.py --help` |
| `tools/s10_vic_coupling/vic_soil_to_ldndc_site.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s10_vic_coupling/vic_soil_to_ldndc_site.py --help` |
| `tools/s10_vic_coupling/vic_to_ldndc_climate.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s10_vic_coupling/vic_to_ldndc_climate.py --help` |
| `tools/s10_vic_coupling/vic_to_ldndc_soilwater.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s10_vic_coupling/vic_to_ldndc_soilwater.py --help` |
| `tools/s1_project_setup/create_project_structure.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s1_project_setup/create_project_structure.py --help` |
| `tools/s1_project_setup/generate_project_xml.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s1_project_setup/generate_project_xml.py --help` |
| `tools/s2_site_config/generate_site_xml.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s2_site_config/generate_site_xml.py --help` |
| `tools/s2_site_config/hwsd_to_ldndc_soil.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s2_site_config/hwsd_to_ldndc_soil.py --help` |
| `tools/s3_setup_modules/generate_setup_xml.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s3_setup_modules/generate_setup_xml.py --help` |
| `tools/s4_climate_prep/convert_forcing_to_ldndc_climate.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s4_climate_prep/convert_forcing_to_ldndc_climate.py --help` |
| `tools/s4_climate_prep/validate_climate_file.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s4_climate_prep/validate_climate_file.py --help` |
| `tools/s5_airchemistry_prep/generate_airchemistry_file.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s5_airchemistry_prep/generate_airchemistry_file.py --help` |
| `tools/s6_management_config/generate_management_xml.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s6_management_config/generate_management_xml.py --help` |
| `tools/s7_species_params/validate_species_params.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s7_species_params/validate_species_params.py --help` |
| `tools/s8_execution/run_ldndc.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s8_execution/run_ldndc.py --help` |
| `tools/s9_output_parsing/aggregate_annual_budget.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s9_output_parsing/aggregate_annual_budget.py --help` |
| `tools/s9_output_parsing/parse_physiology_output.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s9_output_parsing/parse_physiology_output.py --help` |
| `tools/s9_output_parsing/parse_soilchemistry_output.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s9_output_parsing/parse_soilchemistry_output.py --help` |
| `tools/s9_output_parsing/parse_watercycle_output.py` | `KISSPATH_PYTHON_ENV/bin/python {KI}/tools/s9_output_parsing/parse_watercycle_output.py --help` |

*20 public tools; `_`-prefixed helpers and packaging files excluded.*
<!-- KI-TOOL-INDEX:END -->

---

## Data Preparation

### Forcing data

**Data Sources**: Use `from ki_tools_common.load_forcing import load_daily_forcing` for CMFD/MSWX/NASA POWER.
Then convert to LDNDC climate.txt format using this KI's tool: `tools/s4_climate_prep/convert_forcing_to_ldndc_climate.py`

**Data Validation Reference**: See `data_ki/CMFD/SKILL.md` for CMFD unit documentation and known traps.

---

# LandscapeDNDC (LDNDC) — Knowledge Infrastructure

> **Model**: LandscapeDNDC v1.37 (KIT/IMK-IFU)
> **Domain**: Terrestrial biogeochemistry — C/N cycling, GHG emissions (N2O, CO2, CH4), nutrient leaching, crop yield
> **Pipeline stages**: 10 | **Tools**: 19 | **Skill documents**: 10 | **Diagnostic triplets**: 22
> **Binary**: `model/ldndc/ldndc-1.37.linux64/bin/ldndc` (pre-built, Linux 64-bit)
> **Status**: Validated on Bengbu wheat-maize (2000-2005) — N2O=4.9 kgN/ha/yr matches field measurements

## Overview

LandscapeDNDC is a modular, CLI-native C++ biogeochemistry framework for simulating biosphere-atmosphere-hydrosphere exchange processes. It replaces the GUI-bound DNDC model with a scalable, XML-configured system suitable for site and regional-scale simulations.

## 1. Model Identity

| Property | Value |
|----------|-------|
| Full name | LandscapeDNDC |
| Version | v1.37 |
| Language | C++ |
| Primary domain | Terrestrial biogeochemistry |
| Spatial mode | Site and regional-scale simulations |
| Binary | `model/ldndc/ldndc-1.37.linux64/bin/ldndc` |

## 2. What This Model Does

LandscapeDNDC simulates C/N cycling, greenhouse gas emissions, nutrient leaching, crop growth, soil carbon dynamics, and water-cycle processes for terrestrial systems. It is configured through XML input files and run through the real LDNDC binary; do not replace it with simplified formulas.

## 3. Input Requirements

**Exact shapes live in `docs/format_spec.yaml`** (projected from dag + triplets; regenerate it after changing either source, never hand-edit). This section explains the operational intent and common traps; the spec file is the contract.

### 3.1 Meteorological Forcing

| Variable | Unit model expects | Source dataset | Source unit | Conversion |
|----------|-------------------|----------------|-------------|------------|
| Precipitation | mm/day | CMFD 3-hourly | kg/m2/s | multiply each 3-hour step by 10800, then sum 8 steps |
| Precipitation | mm/day | MSWX 3-hourly | mm/3hr | sum 8 steps |
| Temperature | degC | CMFD 3-hourly | K | subtract 273.15 |
| Temperature | degC | MSWX 3-hourly | degC | none |
| Shortwave radiation | W/m2 | VIC/forcing pipeline | W/m2 | none |
| Wind speed | m/s | forcing pipeline | m/s | none |
| Relative humidity | % | forcing pipeline | % | enforce 0-100 range |

Use `from ki_tools_common.load_forcing import load_daily_forcing` for CMFD/MSWX/NASA POWER, or `from ki_tools_common.netcdf_utils import load_cmfd_daily_all` when direct CMFD 3-hourly NetCDF reading is needed.

### 3.2 Static Inputs

| Input | Source | Tool that prepares it |
|-------|--------|----------------------|
| Soil properties | HWSD or VIC soil parameters | `tools/s2_site_config/hwsd_to_ldndc_soil.py`, `tools/s10_vic_coupling/vic_soil_to_ldndc_site.py` |
| Project structure | User/project settings | `tools/s1_project_setup/create_project_structure.py` |
| Site XML | Soil profile, canopy, initial conditions | `tools/s2_site_config/generate_site_xml.py` |
| Species parameters | LDNDC species parameter files | `tools/s7_species_params/validate_species_params.py` |

### 3.3 Configuration Files

| File | Format | Notes |
|------|--------|-------|
| `project.xml` | XML | Master schedule, source paths, and output paths |
| `site.xml` | XML | Soil profile, canopy, and initial conditions |
| `setup.xml` | XML | Module selection and output configuration |
| `climate.txt` | tab-separated text | Meteorological forcing |
| `airchem.txt` | text | CO2 concentration and N deposition |
| `mana.xml` | XML | Sowing, fertilizer, harvest, flood, and drain events |
| `parameters_species.xml` | XML | Crop and vegetation parameters |

## Validated Results — Bengbu Wheat-Maize (Default Parameters)

| Metric | LDNDC (default) | Published Range | Source |
|--------|:---:|:---:|--------|
| N2O emission | 4.9 kgN/ha/yr | 1.9-6.7 kgN/ha/yr | Sichuan field 3yr |
| N2O EF | 1.4% | 1-2% | IPCC Tier 1 |
| NO3 leaching (yr1) | 41.4 kgN/ha | 38-60 kgN/ha | NCP field measurements |
| SOC change | +463 kgC/ha/yr | +113-350 kgC/ha/yr | National avg to straw return |
| Total GHG | 12,234 kgCO2eq/ha/yr | — | — |

**Key finding**: Default LDNDC parameters with HWSD soil + VIC climate produce publication-quality GHG estimates without calibration.

## LDNDC vs DNDC

Same biogeochemistry science, different architecture:
- **DNDC**: Windows GUI, Delphi, single-site, manual setup — cannot be automated
- **LDNDC**: Linux CLI, C++11, XML config, MPI parallel, modular — fully agentic

## Working Pipeline Tool

`tools/run_ldndc_bengbu_ghg.py` (1,032 lines) — complete end-to-end:
```bash
python run_ldndc_bengbu_ghg.py --lat 32.94 --lon 117.35 \
  --forcing_file outputs/bengbu_.../forcing_final/bengbu_0.25deg_32.8750_117.3750
```
Steps: VIC forcing → climate.txt, HWSD → site.xml, management XML, setup XML, airchemistry, project → run LDNDC → parse JSON results.

## Critical Species Names (verified)

| Crop | LDNDC Name | DSSAT Name | SWAT+ Name |
|------|-----------|-----------|------------|
| Maize | **CORN** | MZCER | agrl |
| Winter wheat | **WIWH** | WHCER | wwht |
| Rice | **RICE** | RICER | rice |
| Spring wheat | **SWHE** | WHCER | swht |
| Soybean | **SOYB** | SBGRO | soyb |
| Rapeseed | **RAPE** | — | canp |

## Critical Unit Conversions (verified)

| Variable | HWSD/VIC Unit | LDNDC Unit | Conversion |
|----------|:---:|:---:|:---:|
| corg (SOC) | % | fraction | ÷ 100 |
| sks (Ksat) | mm/hr | cm/min | ÷ 600 |
| clay | % | fraction | ÷ 100 |
| temperature | °C (VIC) | °C | none |
| radiation | W/m² (VIC) | W/m² | none |
| precipitation | mm/3hr (VIC) | mm/day | sum 8 steps |

Key capabilities:
- **Greenhouse gas emissions**: N2O, CO2, CH4 from soils (denitrification, nitrification, decomposition, methanogenesis)
- **Nutrient leaching**: NO3, NH4, dissolved organic N leaching to groundwater
- **Soil carbon dynamics**: Organic matter decomposition, humification, C pool changes
- **Crop/vegetation growth**: GPP, NPP, yield, LAI, biomass allocation via PlaMox or PSIM modules
- **Water cycle**: Infiltration, percolation, ET partitioning, runoff, snow dynamics
- **Management effects**: Tillage, fertilization, irrigation, crop rotation impacts on biogeochemistry

## 4. Build Instructions

See `docs/s1_installation_skill.md` for compilation from source and dependency setup. For HydroCraft, the pre-built binary is at `KISSPATH_HOME/LDNDC/bin/ldndc`. The LDNDC Docker image is available at `codebase.helmholtz.cloud/landscapedndc/ldndc-docker` for containerized deployment.

## 5. Execution

```bash
# LDNDC binary location
LDNDC_BIN="KISSPATH_HOME/LDNDC/bin/ldndc"

# Run LDNDC
$LDNDC_BIN project.xml
# or with config file:
$LDNDC_BIN -c ldndc.conf project.xml

# Python tools require the HydroCraft venv
source KISSPATH_PYTHON_ENV/bin/activate
```

Always run `python preflight_check.py` in this KI directory before debugging model execution.

## 6. Output Description

**Source: `dag.yaml`.** The dag is the model identity for outputs. If this section ever disagrees with `dag.yaml`, the dag wins and this section is the bug.

**Headline output** (the dag's `validation_rank: 1` variable; the output this model is judged by):

> `evapotranspiration` -- Daily actual evapotranspiration (transpiration + soil + interception evaporation). (mm)

| Output variable (dag `var`) | Rank | Unit | Description |
|-----------------------------|------|------|-------------|
| `evapotranspiration` | 1 | mm | Daily actual evapotranspiration (transpiration + soil + interception evaporation). |

Other dag outputs: `dN_n2o_emis`, `dN_no_emis`, `dC_co2_emis_hetero`, `dC_ch4_emis`, `dN_no3_leach`, `yield`, `lai`.

## 7. Tool Inventory

Use the stage table and tool reference below as the operational inventory. Read each tool's argparse (`--help`) before composing a command.

## Pipeline

| Stage | Name | Skill Document | Tools |
|-------|------|----------------|-------|
| S1 | Project Structure Setup | `docs/s1_project_setup_skill.md` | `create_project_structure`, `generate_project_xml` |
| S2 | Site Configuration (site.xml) | `docs/s2_site_config_skill.md` | `generate_site_xml`, `hwsd_to_ldndc_soil` |
| S3 | Module Configuration (setup.xml) | `docs/s3_setup_modules_skill.md` | `generate_setup_xml` |
| S4 | Climate Data Preparation | `docs/s4_climate_prep_skill.md` | `convert_forcing_to_ldndc_climate`, `validate_climate_file` |
| S5 | Air Chemistry Preparation | `docs/s5_airchemistry_prep_skill.md` | `generate_airchemistry_file` |
| S6 | Management Events (mana.xml) | `docs/s6_management_config_skill.md` | `generate_management_xml` |
| S7 | Species Parameters | `docs/s7_species_params_skill.md` | `validate_species_params` |
| S8 | Model Execution | `docs/s8_execution_skill.md` | `run_ldndc` |
| S9 | Output Parsing and Analysis | `docs/s9_output_parsing_skill.md` | `parse_soilchemistry_output`, `parse_watercycle_output`, `parse_physiology_output`, `aggregate_annual_budget` |
| S10 | Multi-Model Coupling | `docs/s9_coupling_skill.md` | `vic_to_ldndc_climate`, `vic_to_ldndc_soilwater`, `vic_soil_to_ldndc_site` |

## Tools Reference

| Tool ID | Script Path | Purpose |
|---------|------------|---------|
| `create_project_structure` | `tools/s1_project_setup/create_project_structure.py` | Create project directory with input/output subdirs |
| `generate_project_xml` | `tools/s1_project_setup/generate_project_xml.py` | Generate master project.xml with schedule and paths |
| `generate_site_xml` | `tools/s2_site_config/generate_site_xml.py` | Generate site.xml with soil profile and canopy |
| `hwsd_to_ldndc_soil` | `tools/s2_site_config/hwsd_to_ldndc_soil.py` | Convert HWSD global soil data to LDNDC format |
| `generate_setup_xml` | `tools/s3_setup_modules/generate_setup_xml.py` | Generate module selection config |
| `convert_forcing_to_ldndc_climate` | `tools/s4_climate_prep/convert_forcing_to_ldndc_climate.py` | Convert CMFD/MSWX/VIC forcing to climate.txt |
| `validate_climate_file` | `tools/s4_climate_prep/validate_climate_file.py` | Validate climate.txt ranges and completeness |
| `generate_airchemistry_file` | `tools/s5_airchemistry_prep/generate_airchemistry_file.py` | Generate CO2 + N deposition input |
| `generate_management_xml` | `tools/s6_management_config/generate_management_xml.py` | Generate sowing/fertilizer/harvest events |
| `validate_species_params` | `tools/s7_species_params/validate_species_params.py` | Cross-validate species params against management |
| `run_ldndc` | `tools/s8_execution/run_ldndc.py` | Execute LDNDC binary with error handling |
| `parse_soilchemistry_output` | `tools/s9_output_parsing/parse_soilchemistry_output.py` | Extract GHG fluxes and nutrient leaching |
| `parse_watercycle_output` | `tools/s9_output_parsing/parse_watercycle_output.py` | Extract ET, drainage, runoff, soil moisture |
| `parse_physiology_output` | `tools/s9_output_parsing/parse_physiology_output.py` | Extract GPP, NPP, yield, LAI, biomass |
| `aggregate_annual_budget` | `tools/s9_output_parsing/aggregate_annual_budget.py` | Compute annual C/N budgets with mass balance |
| `vic_to_ldndc_climate` | `tools/s10_vic_coupling/vic_to_ldndc_climate.py` | Convert VIC forcing to LDNDC climate format |
| `vic_to_ldndc_soilwater` | `tools/s10_vic_coupling/vic_to_ldndc_soilwater.py` | Map VIC soil moisture to LDNDC layers |
| `vic_soil_to_ldndc_site` | `tools/s10_vic_coupling/vic_soil_to_ldndc_site.py` | Convert VIC soil params to LDNDC site.xml |

Shared utilities should be used instead of raw one-off data extraction code:

```python
from ki_tools_common.load_forcing import load_daily_forcing
from ki_tools_common.netcdf_utils import load_cmfd_daily_all
from ki_tools_common.soil_utils import lookup_hwsd
from ki_tools_common.metrics import all_metrics
from ki_tools_common.validation import validate_forcing_ranges
from ki_tools_common.units import convert
```

## 8. Unit Conversion Table

| Variable | Source unit (verified) | Model unit | Factor or operation | Type |
|----------|------------------------|------------|---------------------|------|
| `corg` soil organic carbon | % | fraction | divide by 100 | multiplicative |
| `sks` saturated hydraulic conductivity | mm/hr | cm/min | divide by 600 | multiplicative |
| `clay` | % | fraction | divide by 100 | multiplicative |
| Temperature | K | degC | subtract 273.15 | additive |
| Temperature | degC | degC | none | identity |
| Radiation | W/m2 | W/m2 | none | identity |
| Precipitation | mm/3hr | mm/day | sum 8 steps | aggregation |
| CMFD precipitation | kg/m2/s | mm/day | multiply each 3-hour step by 10800, then sum 8 steps | multiplicative + aggregation |

### 8c. Sign Conventions and Output Units

Post-processing must treat evapotranspiration as the dag headline output in `mm`. For all output variables, bind observations and metric calculations by the dag variable name, not by an inferred column label.

| Variable | Convention in this model | Common alternative | Impact if wrong |
|----------|--------------------------|--------------------|-----------------|
| `evapotranspiration` | Daily actual evapotranspiration in mm | Rate or differently signed ET flux | Magnitude or sign error in water-balance validation |
| `dN_n2o_emis` | Daily nitrogen emission output by dag variable name | Confusion with annual kgN/ha summaries | Wrong temporal aggregation or variable binding |
| `dN_no_emis` | Daily nitrogen emission output by dag variable name | Confusion with N2O emission | Wrong gas species validation |
| `dN_no3_leach` | Leaching output by dag variable name | Confusion with concentration | Wrong nutrient-loss comparison |

## Critical Domain Knowledge

### LDNDC Input File Structure

```
project_dir/
  project.xml          # Master config: schedule, source paths, output paths
  input/
    site.xml           # Soil profile, canopy, initial conditions
    setup.xml          # Module selection and output configuration
    climate.txt        # Tab-separated meteorological forcing
    airchem.txt        # CO2 concentration and N deposition
    mana.xml           # Management events (sowing, fertilizer, harvest)
    parameters_species.xml   # Crop/vegetation parameters
    parameters_site.xml      # Site-level parameters (optional)
  output/
    soilchemistry-daily.txt  # Daily GHG fluxes, leaching
    watercycle-daily.txt     # Daily water balance
    physiology-daily.txt     # Daily vegetation variables
    *.txt                    # Additional configured outputs
```

### project.xml Structure

```xml
<ldndcproject id="0" lat="33.5" lon="117.2">
  <schedule time="2000-01-01/24 -> 2010-12-31"/>
  <input>
    <sources sourceprefix="input/">
      <site source="site.xml"/>
      <event source="mana.xml"/>
      <setup source="setup.xml"/>
      <climate source="climate.txt"/>
      <airchemistry source="airchem.txt"/>
      <speciesparameters source="parameters_species.xml"/>
      <siteparameters source="parameters_site.xml"/>
    </sources>
    <attributes use="0">
      <airchemistry endless="yes"/>
    </attributes>
  </input>
  <output>
    <sinks sinkprefix="output/"/>
  </output>
</ldndcproject>
```

### climate.txt Format

Tab-separated with header row. Required columns:
- `tavg` [C] or `tmin` + `tmax` [C] -- average/min/max temperature
- `prec` [mm] -- precipitation (rain + snow)
- `wind` [m/s] -- wind speed
- `rh` [%] -- relative humidity (0-100)
- `glob` [W/m2] -- global (shortwave) radiation (optional, can be synthesized)
- `lrad` [W/m2] -- incoming longwave radiation (optional)

Example:
```
#       tavg    tmin    tmax    prec    wind    rh      glob
2000-01-01      -2.3    -5.1    0.8     0.0     2.1     78.5    85.2
2000-01-02      -1.5    -4.2    1.2     3.2     1.8     82.1    72.4
```

### Silent Error Modes (Most Dangerous)

1. **Soil organic carbon in wrong units**: LDNDC expects `corg` as mass fraction (e.g., 0.015 = 1.5%). If provided as percentage (1.5), soil C pools will be 100x too large, producing massively inflated CO2 emissions. The model runs without error.

2. **Temperature in Kelvin instead of Celsius**: LDNDC expects Celsius. Kelvin input causes no parse error but makes all temperature-dependent biogeochemistry wrong (decomposition rates, nitrification, denitrification).

3. **Mismatched crop species name**: If the species name in mana.xml does not exactly match parameters_species.xml, LDNDC may use default parameters or skip the crop entirely. No error is raised.

4. **N deposition units**: LDNDC expects kgN/ha/yr for deposition rates. Providing g/m2/yr (factor of 10 different) silently corrupts the N budget.

5. **Incompatible module combination**: Selecting a forest physiology module with cropland management events produces no error but generates nonsensical output.

## 9. Diagnostic Triplets (Top 5)

The full diagnostic corpus stays in `diagnostics/triplets.yaml`; check it first on any error and do not duplicate or renumber entries here.

| # | Error class | Diagnosis | Remedy |
|---|-------------|-----------|--------|
| 1 | Path resolution errors (`dt_001`) | Required binary, config, or input path is wrong | Use the triplet remedy, then re-run `python preflight_check.py` |
| 2 | XML format errors (`dt_002`, `dt_004`, `dt_008`) | LDNDC input XML is malformed or semantically incompatible | Compare with working examples and regenerate with this KI's XML tools |
| 3 | Silent unit conversion errors (`dt_005`, `dt_012`, `dt_014`) | Input values have plausible syntax but wrong units | Re-read `docs/format_spec.yaml` and the conversion table before rerunning |
| 4 | Silent biogeochemistry errors (`dt_003`, `dt_007`, `dt_009`, `dt_013`) | Model runs but crop, C/N, or module behavior is scientifically wrong | Follow the triplet remedy and inspect model-native outputs |
| 5 | Runtime crashes (`dt_010`, `dt_011`) | LDNDC terminates before producing valid output | Use the triplet remedy and preserve full stderr/log context |

## 10. Coupling Interfaces

## Coupling Points with HydroCraft Models

### VIC -> LDNDC
- **Climate forcing**: VIC forcing files (7-col ASCII) converted to LDNDC climate.txt via `vic_to_ldndc_climate`. Unit conversions: K->C (temperature), specific humidity->RH%, sub-daily->daily aggregation.
- **Soil parameters**: VIC soil param file (texture, Ksat, porosity, bulk density) mapped to LDNDC site.xml soil horizons via `vic_soil_to_ldndc_site`. VIC 3-layer to LDNDC multi-layer interpolation.
- **Soil moisture**: VIC simulated soil moisture used for LDNDC initial conditions via `vic_to_ldndc_soilwater`.

### CaMa-Flood -> LDNDC
- **Water table depth**: CaMa-Flood simulated water table depth can inform LDNDC groundwater boundary conditions, affecting anaerobic zone extent and CH4/N2O production.
- **Flood inundation**: CaMa-Flood flood fraction and depth can drive LDNDC wetland/paddy rice methane modules.

### LDNDC -> DSSAT (comparison)
- LDNDC crop yield can be compared with DSSAT yield for validation.
- LDNDC N leaching can inform DSSAT water quality assessment.

## Error Handling

See `diagnostics/triplets.yaml` for 15 diagnostic triplets covering:
- Path resolution errors (dt_001)
- XML format errors (dt_002, dt_004, dt_008)
- Silent unit conversion errors (dt_005, dt_012, dt_014)
- Silent biogeochemistry errors (dt_003, dt_007, dt_009, dt_013)
- Runtime crashes (dt_010, dt_011)
- Cross-model coupling mismatches (dt_015)

---

*This knowledge infrastructure was built using the Knowledge Dissection Toolkit v1.0 (Zhang et al., Nature, under review).*
*Part of the HydroCraft multi-model simulation platform by the Jianyun Zhang Research Group, Hohai University.*

---

## 11. Validated Results

### Test Site: Bengbu Wheat-Maize

| Property | Value |
|----------|-------|
| Location | 32.94, 117.35 |
| Period | 2000-2005 |
| Status | Validated on Bengbu wheat-maize |

### Performance Metrics -- judged against the field's bar, not intuition

**State the bar from `docs/validation_convention.yaml`; cite every band and write null bands as "no cited threshold".** The convention bars currently restated here are for `dN_n2o_emis` and `dN_no_emis`. No convention bar is stated here for the dag rank-1 `evapotranspiration`; consult `docs/validation_convention.yaml` before judging an evapotranspiration run.

| Dag variable | Metric | Direction | Satisfactory band | Good band | Very good band | Citation key |
|--------------|--------|-----------|-------------------|-----------|----------------|--------------|
| `dN_n2o_emis` | NSE | maximize | 0.0 (`ali2014`) | no cited threshold (`ali2014`) | 1.0 (`ali2014`) | `ali2014` |
| `dN_n2o_emis` | PBIAS | zero_centered | 25.0 (`ali2014`) | 15.0 (`ali2014`) | 10.0 (`ali2014`) | `ali2014` |
| `dN_no_emis` | NSE | maximize | 0.0 (`ali2014`) | no cited threshold (`ali2014`) | 1.0 (`ali2014`) | `ali2014` |

### Data Replacement Tracking

No additional component replacement statuses are restated here. Use `knowledge_infrastructure.yaml`, `docs/format_spec.yaml`, and the run artifacts for machine-readable component status; do not infer replacement status from this prose section.

## 12. Parameter Selection by Region

Use the region guidance below as physically informed starting points, not calibration. When no site-specific calibration exists, prefer documented crop calendars, species names, and module stacks already represented in this KI.

## Rice (Paddy) Configuration — Critical (Validated 2026-03-20)

LDNDC rice simulation requires THREE separate fixes that are NOT documented in the standard user guide:

### 1. Species Name: Use PADR, NOT RICE
```xml
<!-- WRONG — RICE is abstract base class, produces zero output silently -->
<plant type="RICE" />

<!-- CORRECT — PADR = paddy rice (flooded) -->
<plant type="PADR" />

<!-- Also valid: UPLR = upland rice (rainfed) -->
```

### 2. Module Stack: Use paddy-specific modules
Standard dryland setup.xml (MeTrX) crashes immediately for paddies. Rice needs anaerobic soil chemistry and ponded water handling.

### 3. Management Events: Explicit flood/drain required
```xml
<event type="flood" doy="135" depth_mm="80"/>   <!-- transplanting flood -->
<event type="drain" doy="195"/>                   <!-- mid-season drain -->
<event type="flood" doy="205" depth_mm="50"/>    <!-- re-flood -->
<event type="drain" doy="250"/>                   <!-- pre-harvest drain -->
```

Without all three fixes, rice either crashes (fix 2 missing) or runs silently with zero output (fix 1 or 3 missing).

### HWSD Water Body Cells
If HWSD ISSOIL=0 for a cell (water body), use nearest-neighbor soil donor (0.25° search). 4 cells in Harbin basin required this fix.

---

## Current Limitations & Future Work (Assessed 2026-03-21)

### N₂O Underestimate — Root Cause Identified

LDNDC N₂O emissions for Harbin maize are **0.3-0.8 kgN/ha/yr** vs literature **1-3 kgN/ha/yr**. Three coupling approaches were tested:

| Approach | N₂O Result | Why it didn't work |
|----------|-----------|-------------------|
| DSSAT residue via `<event type="residue">` | 0.34 (no change) | Event type silently ignored by LDNDC |
| High `initialbiomass` + `remains=0.85` | 0.15 (worse) | More crop N uptake → less mineral N |
| Increased fertilizer 180→500 kgN/ha | 0.33→0.81 (improves) | Still below literature even at extreme |

**Root cause**: plamox phenology — DVS only reaches 0.4-0.65 in ~50% of years (crop doesn't mature). This means:
- No consistent root exudates → less rhizosphere denitrification
- Variable residue → depleted soil organic C in bad years
- Different moisture regime → fewer anaerobic microsites

**Correct fix (future work)**: Calibrate plamox GDD parameters for NE China short-season maize. Reduce thermal time from ~1560 to ~1100-1200 GDD in `parameters_species.xml`.

### What IS Usable

| Output | Usable? | Note |
|--------|---------|------|
| N₂O **relative change** (future/baseline ratio) | **Yes** | Direction of change is valid |
| CO₂ heterotrophic respiration | **Yes** | 1,537 kgC/ha/yr is reasonable |
| CH₄ from rice paddies | **Partial** | High variance, needs paddy module calibration |
| **Absolute N₂O values** | **No** | 5-10x too low, needs plamox calibration |
| Crop yields | **No** | Use DSSAT instead |

### LDNDC Execution — Critical Flag

Always run with `--conf`:
```bash
ldndc --conf /path/to/ldndc.conf ldndc_project.xml
```
Without `--conf`, LDNDC crashes with "premature termination" — no log file, no error message. The `.conf` must specify `resources_path` pointing to the LDNDC installation directory.

---

## Crop Calendar Reference (China)

| Region | Latitude | Winter Wheat | Summer Maize | Rice |
|--------|----------|-------------|-------------|------|
| Northeast | >40°N | — | May-Sep | — |
| North China | 35-40°N | Oct-Jun | Jun-Sep | — |
| Huang-Huai | 32-35°N | Oct-Jun | Jun-Oct | — |
| Yangtze | 28-32°N | Nov-May | — | Apr-Oct |
| South | <28°N | — | — | Mar-Jul, Jul-Nov |

**Data sources on server:**
- GGCMI Crop Calendar: `KISSPATH_HOME/Crop_model_dataset/GGCMI_phase3_crop_calendar/`
- China Phenology GeoTIFF: `KISSPATH_HOME/Crop_model_dataset/8313530/`
- SPAM crop distribution: `KISSPATH_HOME/Crop_model_dataset/dataverse_files/`