BERVO gap analysis: variables needed to represent CHESS (Colorado Headwaters Ecological Spectroscopy Study) datasets

[cmungall]

Summary

A systematic analysis of the CHESS datasets on ESS-DIVE (16 datasets) reveals significant gaps in BERVO's coverage of field ecology observational data. BERVO excels at representing EcoSIM's biogeochemical model parameters but lacks terms for many variables commonly collected in field campaigns that serve as model inputs, calibration targets, or validation benchmarks.

This analysis was generated from the chess-data repo which downloads all CHESS data from ESS-DIVE and maps variables to BERVO.

---

Gap Categories

1. 🔴 Vegetation Structure (HIGH PRIORITY — core EcoSIM inputs)

These are standard field measurements that parameterize or validate plant functional type (PFT) models:

Missing Variable	CHESS Source	Why BERVO Needs It
Vegetation cover fraction	Vegetation_Cover, Cover_Percent, FractionalCover	Ground-truth for canopy coverage. Maps to PFT fractional area in EcoSIM grid cells. No BERVO term exists for this fundamental measurement.
Stem diameter (DBH)	Stem_DBH, DBH_1_CM, DBH_2_CM, DBH_Avg_CM	Standard forest inventory metric. Allometric equations use DBH to estimate biomass, carbon stocks, canopy dimensions — all EcoSIM inputs.
Crown class / canopy position	Crown_Class, Canopy_Position	Dominant/codominant/intermediate/suppressed classification. Determines light interception partitioning in canopy models.
Basal area	ba (forest structure dataset)	Stand-level metric derived from DBH. Direct input for stand density parameterization.
Tree/stem density	density, abla_density, pien_density, pico_density	Species-specific stem counts per area. Required for PFT abundance initialization.

Existing close matches: BERVO:0000695 (Pft canopy height) covers height but nothing else in this category.

2. 🔴 Leaf Area Index & Canopy Optics (HIGH PRIORITY)

CHESS has an entire LAI dataset with sophisticated measurements:

Missing Variable	CHESS Source	Why BERVO Needs It
Field-measured LAI (as distinct from modeled LAI)	L_2200, L_SCATCOR, L_WN, L_LANG, L_ELLIP, L_FV2200	Multiple LAI estimation methods (LAI-2200, scattering-corrected, Warren-Neilson, Lang, Ellipsoidal). These are observational LAI values used to calibrate/validate the modeled BERVO:8000164 (Leaf area index). BERVO should distinguish observed vs. modeled LAI.
Effective LAI (Le)	Le_2200, Le_FV2200, Le_WN	Clumping-uncorrected LAI. Distinct from true LAI. Important for radiation transfer models.
Clumping index	CII, ACF_2200, ACF_SCATCOR	Canopy clumping correction factors. BERVO:0000814 references clumping but only as a parameter, not an observed metric.
Scattering correction factors	SCATCOR_*, CHI_SCATCOR, S_SCATCOR	Wood-to-total area ratio corrections for LAI instruments.

3. 🟡 Soil Properties — Observational (MEDIUM PRIORITY)

CHESS metagenome dataset includes extensive soil characterization. Some overlap with existing BERVO soil terms, but many are missing:

Missing Variable	CHESS Source	Why BERVO Needs It
Soil moisture (field-measured)	SoilMoisture_%_1/2/3, water content, VWC_1/2	Volumetric water content field measurements. Validation target for EcoSIM hydrology.
Bulk electrical conductivity	bulk electrical conductivity	TDR-measured. Proxy for soil moisture and salinity.
Saturated hydraulic conductivity	ksat	Direct input for EcoSIM soil water flow.
Available water capacity	awc	Plant-available water. Key soil parameter.
Infiltration rate	infiltrations	Surface hydrology parameterization.
Microbial biomass C/N	microbial biomass carbon, microbial biomass nitrogen	Microbial pool initialization for biogeochemical cycling.
Soil chemical properties	ammonium nitrogen, nitrate_nitrogen, nitrite_nitrogen, organic nitrogen, total phosphorus, phosphate, aluminum saturation, manganese, zinc	Nutrient pools and chemistry. Some may exist in BERVO under different names — needs cross-referencing.

4. 🟡 Terrain & Topography (MEDIUM PRIORITY)

Missing Variable	CHESS Source	Why BERVO Needs It
Elevation	Elevation, Elevation_m, Topographical_Elevation	BERVO has slope-related terms but no explicit elevation term.
Topographic wetness index (TWI)	twi	Hydrological connectivity metric. Important for subsurface flow modeling.
Topographic position index (TPI)	tpi	Ridge/valley classification. Affects soil depth, moisture.
Curvature	curvature	Surface curvature. Controls flow convergence/divergence.
Heat load index	heat_load	Aspect-derived radiation metric.
Folded aspect	folded_aspect_205	Southwest-normalized aspect. Common in ecological modeling.

Existing matches: BERVO:0000685 (Aspect), BERVO:8000031 (Slope) — these are good, but the derived indices are missing.

5. 🟡 Climate / Meteorological (MEDIUM PRIORITY)

Missing Variable	CHESS Source	Why BERVO Needs It
Actual evapotranspiration (AET)	aet	Water balance component. Key validation target.
Snow water equivalent (SWE)	swe, delta_swe	Snowpack hydrology. Critical for mountain ecosystem modeling.
Climatic water deficit (CWD)	cwd	Drought stress metric. Drives vegetation response.
Mean annual/seasonal temperature	mean annual temperature, mean seasonal temperature	Site characterization. May exist in BERVO climate force terms — needs checking.
Mean annual/seasonal precipitation	mean annual precipitation, average seasonal precipitation	Site characterization.

6. 🟢 Spectroscopy & Remote Sensing (LOW PRIORITY for EcoSIM, but important for BioEPIC)

CHESS includes imaging spectroscopy (NEON AOP), field spectra, reflectance mosaics, LiDAR. These are primarily used for spatial scaling and are less directly tied to EcoSIM inputs, but are central to the BioEPIC AI transduction workflow.

Missing Variable	Notes
Hyperspectral reflectance	NEON AOP 426-band reflectance. Not a single variable but a data type.
Canopy water content	Derived from spectroscopy. Relevant to plant stress.
Shade fraction	Canopy structural metric from LiDAR/spectroscopy.
LiDAR-derived canopy height model (CHM)	Spatial canopy height. Validates BERVO:0000708.
Digital terrain/surface models (DTM/DSM)	Gridded elevation products.

7. 🟢 Taxonomy & Biodiversity (OUT OF SCOPE for BERVO, but worth noting)

CHESS has extensive taxonomic data (Taxon_binomial, Taxon_family, GBIF_Taxon_ID, species lists). These are outside BERVO's scope as a biogeochemical variable ontology, but the link between species identity and PFT assignment is a critical workflow gap. Consider cross-referencing GBIF/NCBI taxonomy to BERVO PFT categories.

---

Recommendations

1. Immediate (for CHESS integration): Add terms for vegetation cover fraction, stem DBH, field-measured LAI (multiple methods), soil moisture, and basic topographic indices. These are the most commonly measured variables in field ecology and the most direct EcoSIM inputs.

2. Distinguish observed vs. modeled quantities. BERVO currently treats LAI as a single concept, but field measurements (LAI-2200, hemispherical photo, etc.) are fundamentally different from modeled LAI. Consider an "observation method" axis or parallel observed/modeled term pairs.

3. Add a "field measurement" category. Many BERVO categories are model-centric ("Plant growth parameters", "Canopy data type"). A "Field observation" or "Measurement" category would naturally house the ground-truth variables that validate model outputs.

4. Consider extending to derived terrain indices (TWI, TPI, heat load). These are standard inputs for spatially distributed models and increasingly important for CONUS-scale parameterization.

5. Cross-reference with ESS-DIVE Community Reporting Formats. ESS-DIVE has standardized reporting formats for CSV headers, sample metadata, etc. Aligning BERVO with these would enable automatic mapping.

---

Data Source

Analysis based on 16 CHESS datasets (ESS-DIVE), spanning:

• Field vegetation surveys (meadow, shrub, tree plots)
• Leaf Area Index (LAI-2200 instrument, multiple correction methods)
• Spectroscopy (field ASD spectrometer, NEON AOP imaging spectroscopy)
• LiDAR (discrete return + waveform)
• Geophysics (TDR soil moisture, EMI survey)
• Metagenomics (soil microbiome + soil properties)
• Forest structure (census data, individual tree detection)

Code and curated mapping: https://github.com/bioepic-data/chess-data

[cmungall]

Correction: Soil Properties Section Was Inaccurate

After a more thorough search of BERVO's 2076 terms, many of the soil properties I flagged as missing do exist in BERVO. I apologize for the sloppy initial analysis. Here's the corrected mapping:

✅ Soil properties that ARE in BERVO (incorrectly flagged as missing):

CHESS Variable	BERVO Match	BERVO ID
SoilMoisture_%, water content, VWC	Total soil water content	BERVO:0000301
bulk electrical conductivity	Soil electrical conductivity / Electrical conductivity	BERVO:0000916 / BERVO:0001703
ksat	Soil vertical/horizontal saturated hydraulic conductivity	BERVO:0001501 / BERVO:0001502
awc	Water contents at field capacity / wilting point (AWC derived)	BERVO:0001499 / BERVO:0001500
ammonium nitrogen	Total soil ammonium content	BERVO:0000306
nitrate_nitrogen	Total soil nitrate content	BERVO:0000307
organic nitrogen	Total organic nitrogen amendment	BERVO:0000312
total phosphorus, phosphate	Total phosphate net mineraln	BERVO:0000015 (+ many P chemistry terms)
aluminum saturation	Soil aluminum content	BERVO:0001677
microbial biomass carbon	Microbial biomass chemical element	BERVO:0000995
pH	Soil pH	BERVO:0000882
cec	Soil cation exchange capacity	BERVO:0000883
carbon/nitrogen ratio	Carbon to nitrogen ratio	BERVO:8000109
organic matter	Total soil particulate organic matter + humus carbon content	BERVO:0000305
sand, silt, clay	Soil sand/silt/clay content	BERVO:0001526/1527/1528
soil temperature	Soil temperature	BERVO:0001477
total carbon	Total soil carbon dioxide / subsurface carbon flux	BERVO:0000323 / 0000319
total nitrogen content	Total soil nitrogen	BERVO:0000324
Soil bulk density	Soil bulk density	BERVO:0001535

✅ Climate/terrain that ARE in BERVO:

CHESS Variable	BERVO Match	BERVO ID
Evapotranspiration	Evapotranspiration	BERVO:0001809
Snowfall / snow cover	Snowfall / Fraction of snow cover	BERVO:0001422 / BERVO:0001439
Mean annual soil temperature	Mean annual soil temperature	BERVO:0001418

Remaining genuine gaps (Section 3 of original issue should be replaced with):

Still Missing	Notes
infiltrations (field-measured infiltration rate)	BERVO has conductivity but not a specific infiltration rate measurement term
microbial biomass nitrogen (as distinct quantity)	BERVO:0000995 is generic "microbial biomass chemical element" — no specific MBN term
nitrite_nitrogen	BERVO has ammonium and nitrate but not nitrite specifically
manganese, zinc	No specific soil Mn or Zn terms (Al is covered)
lime buffer capacity	Not in BERVO
non-microbial biomass	Not in BERVO

Also confirmed in BERVO for climate/terrain:

Still Missing	Notes
Snow water equivalent (SWE)	BERVO has snowfall and snow cover fraction but not SWE as a stored quantity
Climatic water deficit (CWD)	Not in BERVO
Topographic wetness index (TWI), TPI, curvature, heat load	Derived terrain indices — still genuinely missing
Elevation	Still appears to be missing as an explicit term

Bottom line: Section 3 (Soil Properties) of the original issue was mostly wrong. BERVO's soil coverage is much better than I initially reported. The real gaps are concentrated in vegetation structure (Section 1), field-measured LAI methods (Section 2), and derived terrain indices (Section 4).