samples

187

wavelengths

sources

targets

metadata

NIR

family

Provenance & citation

Where the original data and publications come from. Follow these links to obtain the source and cite its authors. Reading guide

Contributor	Seasonal canopy spectra and traits, Blackhawk Island, WI
Conversion status	ok
Origin · url [open]	https://data.ecosis.org/dataset/seasonal-canopy-spectra-and-traits--blackhawk-island--wi
Origin · url [open]	https://data.ecosis.org/api/3/action/package_show?id=seasonal-canopy-spectra-and-traits--blackhawk-island--wi
Origin · url [open]	https://ecosis.org/package/seasonal-canopy-spectra-and-traits--blackhawk-island--wi

Governance & integrity

Source license, redistribution conditions, content version and identity hashes. Redistribution tier and original-source availability are separate. Reading guide

Tier	public
License	PDDL-1.0
Permitted use	Research and benchmarking.
Access policy	Open per source license.
Redistribution	EcoSIS CKAN API package_show verified open license: Open Data Commons Public Domain Dedication and License (PDDL).
Content version	1.1.0
Schema / protocol	2.0
Content hash	`6c9d539ae0268e0f…`
Processing hash	`009752c2f643d8ef…`
Metadata hash	`3092d4247ce32cde…`

Dataset property explorer

Start with the profile axes to locate unusual variation, then inspect the measured values and evidence below. Scores are descriptive heuristics. Reading guide

Mean available profile score0.55

Highest axisLocal artifacts · 1.00

Diagnostics8

Sources profiled1

How to read the profile

The center is 0 and the outer ring is 1. Larger values indicate stronger anomaly or heterogeneity signals on that axis. Compare individual axes using the table; this is not an overall quality grade. Missing axes are left blank, and partial profiles have no filled polygon.

Profile axes

Dimensionless scores from 0 to 1 summarize each property; larger values indicate stronger signals. Unavailable axes are not zero.

Integrity ? Combines missing cells, zero columns and infinite values. Higher values call for checking data completeness.	0.00
Noise ? Combines roughness and its signal-to-roughness proxy. Real narrow bands may also contribute.	0.03
Local artifacts ? Combines spike, jump and clipping heuristics. Inspect flagged spectral regions before attributing a cause.	1.00
Baseline/shape ? Summarizes baseline slope and excess edge variation relative to signal scale.	0.42
PCA outliers ? Combines relative PCA residual and score-distance tails. Valid rare spectra can produce a high value.	0.97
Distance to average spectrum ? Dispersion around this dataset's mean spectrum, using distance, shape and shift summaries.	0.74
Repeatability ? Dispersion among observations sharing an ID. Interpretation depends on the documented meaning of that ID.	unavailable
Multi-regime structure ? Variation in local density in PCA space. May reflect legitimate batches or subpopulations.	0.67

Possible explanations to investigate

Bars rank heuristic signatures from 0 to 1. They do not estimate the probability of a physical fault. On narrow screens, scroll the chart horizontally. Full labels, scores, evidence coverage, possible causes and required external checks are in the table below.

Heuristic signature scores are not calibrated probabilities of physical causes: 0.78 does not mean a 78% causal probability. Missing evidence remains unavailable. RMS/SAM describe dispersion around the dataset mean, without an independent external reference.

Diagnostic hypotheses and supporting evidence
Diagnostic	Score	Heuristic strength	Evidence coverage	Signals	Possible cause	Required external checks
Signature compatible with a spectral spliceX	0.88	strong	100 %	Spike rate 1.00, Jump rate 1.00, PCA Q 0.97	Break at detector junctions, local calibration or different probe.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
Offset compatible with a different white referenceX	0.57	medium	100 %	local artifacts 1.00, PCA Q 0.97, RMS/SAM to mean spectrum 0.74	Systematic discrepancy between campaigns, instruments or white reference.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
Signature compatible with a different backgroundX	0.51	medium	100 %	PCA Q 0.97, RMS/SAM to mean spectrum 0.74, Mahalanobis / T2 0.49	Systematic effect of support, white/black, transflectance or measurement environment.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
Signature compatible with a population shiftX	0.44	medium	100 %	RMS/SAM to mean spectrum 0.74, PCA structure 0.67, Mahalanobis / T2 0.49	Different but physically plausible variety, species, batch or condition.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
High spectral roughness signatureX	0.41	weak	100 %	Spike rate 1.00, PCA Q 0.97, RMS/SAM to mean spectrum 0.74	Weak signal, detector problem, lamp or unstable acquisition.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
Signature compatible with clippingX	0.40	weak	100 %	Jump rate 1.00, PCA Q 0.97, Baseline/mean/area 0.42	Detector saturated or insufficient dynamic range.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
Flat spectrum/weak signalX	0.19	weak	100 %	PCA Q 0.97	Bad contact, missing sample, degraded measurement or very low dynamic range.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.
Weak signatures observedX	—	unknown	90 %		Weak signatures on available metrics; does not certify either physical validity or the absence of defects.	Check the white/black reference actually acquired and its metadata. Check instrument, probe, geometry, campaign and processing/resampling log. Check the repeat protocol before interpreting an intra-ID dispersion.

Spectral sources

Each block is a separate instrument or spectral source. Read its signal summary before using PCA to explore variation between observations. Reading guide

bhi2018_plot_spectra

X · NIR

Read the signal and variation summaries for this source. Axis unit: nm; signal representation: not recorded.

Spectral signal summary

The mean and median describe the typical signal at each channel; shaded bands show the 25th–75th and 5th–95th percentiles across observations. A wider band means greater variation, which can be biological or instrumental.

Sampling

How many observations and spectral channels were exported, and how the wavelength axis is spaced. Reading guide

Wavelengths ? Number of spectral channels. A channel can represent wavelength, wavenumber or another feature axis.	187
Axis range ? Lowest and highest recorded spectral-axis coordinates, in the stated unit.	407.1–2389 nm
Mean spacing ? Average difference between adjacent spectral-axis coordinates; compare only within the same unit.	10.7 nm
Grid ? Uniform means adjacent axis spacings are approximately equal. An irregular grid may need care when applying derivatives.	irregular
Observations ? Exported spectral rows in this source, including repeated measurements when present.	80

Signal & quality

Signal level and roughness summaries. These depend on the signal representation and are not instrument acceptance tests. Reading guide

Value range ? Minimum and maximum recorded spectral values, in the original signal representation.	0.386 – 60.2
Mean range ? Minimum and maximum of the mean spectrum across spectral channels.	1.04 – 44.7
Mean level ? Average signal over observations and channels. The historical key mean_reflectance may also store other signal representations.	20
Area ? Trapezoidal integral of the mean spectrum over the sorted recorded axis; its unit is signal multiplied by axis unit.	3.882e+04
PTP ? Peak-to-peak: maximum minus minimum of the mean spectrum. Low values indicate little variation along the mean spectrum.	43.66
Noise RMS ? Historical roughness sigma proxy from robust second differences. Narrow genuine bands can also raise it; this is not repeat-measured instrument noise.	0.069947
Roughness S/N proxy ? Average absolute signal divided by the roughness proxy. Higher means smoother signal relative to its amplitude, not certified instrument SNR.	2.9e+02
Roughness S/N proxy dB ? 20 log10 of the roughness signal-to-noise proxy. Units are decibels; compare only compatible signal representations.	5e+01 dB
Dynamic range ? Maximum minus minimum of the mean spectrum. Its magnitude depends on the signal representation.	43.7
Smoothness ? Mean per-spectrum RMS of second differences, in signal units. Smaller values indicate smoother spectra under this protocol.	3.18
Saturated ? Fraction at the observed global maximum; a heuristic indicator, not a comparison with known detector limits.	0.0%
X-outliers ? Number flagged by the qualification outlier method. An unusual spectrum can be valid and should not be discarded automatically.	29

Integrity & artefacts

Missing values, narrow spikes and abrupt jumps to investigate in the original spectra. Reading guide

NaN ratio ? Missing spectral cells divided by all spectral cells, displayed as a percentage.	0.00%
Inf count ? Number of positive or negative infinite spectral values; investigate invalid calculations or export.	0
Zero ratio ? Exactly zero spectral cells divided by all cells. Zero can be a valid measurement and is not automatically missing.	0.00%
Spike count ? Local narrow peaks flagged by the second-difference heuristic; real sharp spectral bands may also be flagged.	620
Spike rate ? Fraction of tested second-difference positions flagged as spikes, displayed as a percentage.	4.19%
Jump count ? Adjacent-channel changes flagged by the first-difference heuristic, potentially including detector joins.	588
Jump rate ? Fraction of tested adjacent-channel differences flagged as jumps, displayed as a percentage.	3.95%
Clip fraction ? Fraction of finite cells near the observed global minimum or maximum; inspect detector limits before concluding saturation.	0.01%

Shape & reference

Variation relative to this dataset's mean spectrum; the mean is not an independently measured calibration reference. Reading guide

Baseline slope ? Linear slope fitted to the mean spectrum over an axis normalized to 0–1. It describes total baseline change, not change per physical axis unit.	-9.1263
Curvature RMS ? Median per-spectrum RMS of second differences, summarizing curvature and high-frequency variation.	3.2219
D1 RMS ? Median per-spectrum RMS of adjacent-channel first differences; reflects both real spectral structure and artefacts.	2.4997
RMS to mean ? Root mean square distance to this dataset's mean spectrum, in signal units.	3.7983
RMS p95 ? 95th percentile of distances to the mean spectrum: 95% of observations have a distance at or below this value.	8.059
SAM to mean ? Spectral Angle Mapper: angle to the mean spectrum, in radians. Smaller angles indicate more similar shapes.	0.058524
SAM p95 ? 95th percentile of spectral angles to the dataset mean, in radians. It is less sensitive to overall amplitude than RMS.	0.18459
Affine offset p95 ? 95th percentile of absolute fitted offsets when matching each spectrum to the dataset mean.	4.4922
Affine gain p95 Δ ? 95th percentile of the absolute gain deviation from 1 in the fit to the mean spectrum.	0.36908
Affine residual p95 ? 95th percentile of residual RMS after gain and offset correction to the mean spectrum.	2.7983
Xcorr lag p95 ? 95th percentile of the absolute cross-correlation shift relative to the mean spectrum, in spectral channels.	0

Outliers & repeatability

Unusual spectra and within-ID dispersion. Repeated IDs do not by themselves establish technical repetitions. Reading guide

PCA Q p95/median ? Q (squared prediction error) is variation left outside the retained PCA subspace. The displayed ratio compares its 95th percentile with its median.	7.7
Hotelling T2 p95/median ? Hotelling T² measures distance within the retained PCA subspace, scaled by component variance. Ratio of the 95th percentile to the median.	3.9
Mahalanobis H p95/median ? Distance from the PCA-score center after variance scaling. Ratio of the 95th percentile to the median; large values identify relatively extreme spectra.	2
Repeat groups ? Number of IDs with several observations. Source documentation is needed to decide whether these are technical repeats, positions or time points.	0

Dimensionality (PCA)

Principal component analysis (PCA) compresses correlated wavelengths into components. This describes spectral variation, without using prediction targets. Reading guide

Effective rank ? Entropy-based summary of the variance distribution across components. Larger values indicate variation spread across more dimensions.	1.7
PCs → 95% var ? Number of principal components needed to explain 95% of spectral variance, when reached by the computed projection.	2
PCs → 99% var ? Number of principal components needed to explain 99% of spectral variance, when reached by the computed projection.	4
Top-10 cum. var ? Fraction of spectral variance explained by up to the first ten computed principal components, displayed as a percentage.	99.9%

Computed metric scores 30highest alert 1.00

Value is the measured quantity; alert score is its heuristic mapping to 0–1. Levels are weak below 0.42, medium from 0.42, and strong from 0.72. These are display bands, not calibrated acceptance thresholds. A dash means unavailable, never zero. Possible causes are suggestions to investigate. Use each metric’s ? control for its meaning and scoring: hover for a hint, or expand it with a tap, Enter or Space.

Family	Computed metric	Value	Alert score	Heuristic level	Descriptive interpretation	Possible causes to verify	Calculation / scoring
Data integrity	Invalid/masked cell count (owner report) ? Count of cells reported unavailable by the native owner under its null or validity mask. This count is distinct from the historical NaN ratio; consult the recorded population and denominator. Scoring: No alert mapping is defined for this metric; an unavailable score does not mean a zero measured value. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. integrity.missing_cells	—	—	not calculable	Not enough information to score this metric on this dataset.	Missing data or validity mask; not evidence of an acquisition cause	Native Preflight count of cells unavailable under x/null and valid_mask; denominator stays in native_metric.denominators.n_cells. Not legacy NaN ratio.
Data integrity	NaN ratio ? Missing spectral cells divided by all spectral cells, displayed as a percentage. Scoring: alert = min(1, nan_ratio / 0.05). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. integrity.nan_ratio	0%	0.00	weak	Full spectrum	Acquisition/export error	count(isnan(X)) / X.sizealert = min(1, nan_ratio / 0.05)
Data integrity	Inf count ? Number of positive or negative infinite spectral values; investigate invalid calculations or export. Scoring: alert = min(1, inf_count / 1). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. integrity.inf_count	0	0.00	weak	Normal	Invalid calculations	count(isinf(X))alert = min(1, inf_count / 1)
Data integrity	Zero ratio ? Exactly zero spectral cells divided by all cells. Zero can be a valid measurement and is not automatically missing. Scoring: alert = min(1, zero_ratio / 0.05). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. integrity.zero_ratio	0%	0.00	weak	Normal	Export, saturation	count(X == 0) / count(finite X)alert = min(1, zero_ratio / 0.05)
Overall amplitude	Mean reflectance ? Average signal over observations and channels. The historical key mean_reflectance may also store other signal representations. Scoring: alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. amplitude.mean_reflectance	20.004	0.42	weak	Too dark	Background, geometry	mean(X finite)alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent
Overall amplitude	Area under curve ? Trapezoidal integral of the mean spectrum over the sorted recorded axis; its unit is signal multiplied by axis unit. Scoring: alert reuses baseline/shape drift because area scale depends on axis and units. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. amplitude.area_under_curve	38817	0.42	weak	Normal	Probe distance	trapezoid(mean_spectrum, spectral_axis)alert reuses baseline/shape drift because area scale depends on axis and units
Overall amplitude	Peak-to-peak (PTP) ? Peak-to-peak: maximum minus minimum of the mean spectrum. Low values indicate little variation along the mean spectrum. Scoring: alert increases when dynamic range is abnormally flat. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. amplitude.peak_to_peak	43.661	0.00	weak	Strong variability	Saturation	max(mean_spectrum) - min(mean_spectrum)alert increases when dynamic range is abnormally flat
Overall amplitude	Variance ? Variance of all finite spectral cells, in squared signal units. It combines variation across observations and channels; a small value can describe a flat signal. Scoring: alert increases when variance/dynamic range is abnormally flat. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. amplitude.variance	290.66	0.00	weak	Normal or heterogeneous	Bad contact	var(X finite)alert increases when variance/dynamic range is abnormally flat
Noise	Roughness sigma proxy ? Historical roughness sigma proxy from robust second differences. Narrow genuine bands can also raise it; this is not repeat-measured instrument noise. Scoring: alert = noise_rms / signal_scale, saturated at 5%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. noise.roughness_sigma_proxy	0.069947	0.03	weak	Stable	Lamp, detector	median MAD(second derivative) * 1.4826 / sqrt(6)alert = noise_rms / signal_scale, saturated at 5%
Noise	Roughness S/N proxy ? Average absolute signal divided by the roughness proxy. Higher means smoother signal relative to its amplitude, not certified instrument SNR. Scoring: alert decreases with SNR dB; >=40 dB is treated as low alert. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. noise.roughness_signal_to_noise_proxy	285.99	0.00	weak	Good signal	Acquisition	mean(abs(X)) / noise_rmsalert decreases with SNR dB; >=40 dB is treated as low alert
Noise	Bandwise roughness S/N proxy ? Minimum channel-wise ratio of absolute mean signal to local second-difference roughness. A smaller ratio identifies a weak or rough spectral region; genuine narrow bands and near-zero signal can also lower it. This is not certified instrument SNR. Scoring: alert decreases with worst-band SNR dB; >=35 dB is treated as low alert. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. noise.bandwise_roughness_signal_to_noise_proxy_min	4.0632	0.65	medium	Problem area	Detector	min(abs(mean_spectrum) / local second-derivative noise)alert decreases with worst-band SNR dB; >=35 dB is treated as low alert
Local artifacts	Spike count ? Local narrow peaks flagged by the second-difference heuristic; real sharp spectral bands may also be flagged. Scoring: alert follows spike_rate, saturated at 1%. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. artefacts.spike_count	620	1.00	strong	Artifacts	Cosmic rays, splice	count robust outliers in second derivativealert follows spike_rate, saturated at 1%
Local artifacts	Spike rate ? Fraction of tested second-difference positions flagged as spikes, displayed as a percentage. Scoring: alert = min(1, spike_rate / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. artefacts.spike_rate	4.19%	1.00	strong	Suspicious spectrum	Interpolation	spike_count / (n_samples * (n_features - 2))alert = min(1, spike_rate / 0.01)
Local artifacts	Jump count ? Adjacent-channel changes flagged by the first-difference heuristic, potentially including detector joins. Scoring: alert follows jump_rate, saturated at 1%. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. artefacts.jump_count	588	1.00	strong	Detector connection	Splice	count robust outliers in first derivativealert follows jump_rate, saturated at 1%
Local artifacts	Jump rate ? Fraction of tested adjacent-channel differences flagged as jumps, displayed as a percentage. Scoring: alert = min(1, jump_rate / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. artefacts.jump_rate	3.95%	1.00	strong	Spectral problem	Calibration	jump_count / (n_samples * (n_features - 1))alert = min(1, jump_rate / 0.01)
Local artifacts	Clip fraction ? Fraction of finite cells near the observed global minimum or maximum; inspect detector limits before concluding saturation. Scoring: alert = min(1, clip_fraction / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. artefacts.clip_fraction	0.0134%	0.01	weak	Normal	Detector saturated	fraction of finite cells equal to repeated min/max extremaalert = min(1, clip_fraction / 0.01)
Spectral form	Baseline slope ? Linear slope fitted to the mean spectrum over an axis normalized to 0–1. It describes total baseline change, not change per physical axis unit. Scoring: alert = abs(slope / signal_scale), saturated at 0.5. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. shape.baseline_slope	-9.1263	0.42	weak	Stable	Illumination	linear slope of mean_spectrum over normalized axisalert = abs(slope / signal_scale), saturated at 0.5
Spectral form	Curvature RMS ? Median per-spectrum RMS of second differences, summarizing curvature and high-frequency variation. Scoring: alert = curvature_rms / signal_scale, saturated at 1%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. shape.curvature_rms	3.2219	1.00	strong	Unusual shape	Background, splice	median RMS(second derivative per spectrum)alert = curvature_rms / signal_scale, saturated at 1%
Spectral form	D1 RMS ? Median per-spectrum RMS of adjacent-channel first differences; reflects both real spectral structure and artefacts. Scoring: alert = d1_rms / signal_scale, saturated at 5%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. shape.d1_rms	2.4997	1.00	strong	Structured spectrum	Biology or artifact	median RMS(first derivative per spectrum)alert = d1_rms / signal_scale, saturated at 5%
Multivariate outliers	PCA Q (SPE) ? Q (squared prediction error) is variation left outside the retained PCA subspace. The displayed ratio compares its 95th percentile with its median. Scoring: alert = min(1, pca_q_ratio / 8). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. outliers.pca_q_ratio	7.7226	0.97	strong	Atypical spectrum	Artifact, mixture	p95(Q/SPE residual) / median(Q/SPE residual)alert = min(1, pca_q_ratio / 8)
Multivariate outliers	Hotelling T² ? Hotelling T² measures distance within the retained PCA subspace, scaled by component variance. Ratio of the 95th percentile to the median. Scoring: alert = min(1, hotelling_t2_ratio / 8). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. outliers.hotelling_t2_ratio	3.8787	0.48	medium	Extreme but consistent	Natural variability	p95(Hotelling T2) / median(Hotelling T2)alert = min(1, hotelling_t2_ratio / 8)
Multivariate outliers	Mahalanobis H ? Distance from the PCA-score center after variance scaling. Ratio of the 95th percentile to the median; large values identify relatively extreme spectra. Scoring: alert = min(1, mahalanobis_h_ratio / 4). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. outliers.mahalanobis_h_ratio	1.9691	0.49	medium	Global outlier	Different domain	p95(sqrt(T2)) / median(sqrt(T2))alert = min(1, mahalanobis_h_ratio / 4)
Comparison to average spectrum	RMS to mean spectrum ? 95th percentile of distances to the mean spectrum: 95% of observations have a distance at or below this value. Scoring: alert = RMS_p95 / signal_scale, saturated at 25%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. reference.rms_to_mean_spectrum_p95	8.059	0.74	strong	Different spectrum	Domain shift	p95 RMS distance to dataset mean spectrumalert = RMS_p95 / signal_scale, saturated at 25%
Comparison to average spectrum	Spectral Angle Mapper (SAM) ? 95th percentile of spectral angles to the dataset mean, in radians. It is less sensitive to overall amplitude than RMS. Scoring: alert = min(1, SAM_p95 / 0.35 rad). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. reference.sam_to_mean_spectrum_p95	0.18459	0.53	medium	Different shape	Background, geometry	p95 spectral angle to dataset mean spectrumalert = min(1, SAM_p95 / 0.35 rad)
Repeatability	RMS intra-ID ? Within-ID root mean square deviation from the group's mean spectrum, in signal units. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = RMS_intra_ID / signal_scale, saturated at 10%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. repeatability.rms_intra_id	—	—	not calculable	Not enough information to score this metric on this dataset.	Positioning	median RMS distance to repeated-sample centroidalert = RMS_intra_ID / signal_scale, saturated at 10%
Repeatability	SAM intra-ID ? Within-ID spectral angle to the group's mean spectrum, in radians. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = min(1, SAM_intra_ID / 0.15 rad). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. repeatability.sam_intra_id	—	—	not calculable	Not enough information to score this metric on this dataset.	Acquisition	median SAM to repeated-sample centroidalert = min(1, SAM_intra_ID / 0.15 rad)
Repeatability	CV intra-ID ? Within-ID coefficient of variation: standard deviation relative to absolute mean; unstable near zero signal. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = min(1, CV_intra_ID / 0.25). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. repeatability.cv_intra_id	—	—	not calculable	Not enough information to score this metric on this dataset.	Operator	median within-ID band CValert = min(1, CV_intra_ID / 0.25)
Structure of the dataset	PCA score density ? Inverse median distance to nearby observations in retained PCA-score space. Larger raw values mean more closely spaced observations; scale depends on the PCA representation. The alert uses density variation, so raw density alone does not establish an anomaly. Scoring: alert follows density_cv/profile structure complexity, not raw density alone. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. structure.pca_score_density	0.06369	0.67	medium	Subpopulations	Different batches	1 / median kNN distance in PCA score spacealert follows density_cv/profile structure complexity, not raw density alone
Structure of the dataset	kNN distance ratio (not LOF) ? 95th percentile of mean k-nearest-neighbor distance divided by the population median of that distance, in PCA-score space. Values above 1 indicate an upper tail with more isolated observations; valid rare spectra can also be isolated. This distance ratio is not the Local Outlier Factor algorithm. Scoring: alert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. structure.knn_distance_ratio_p95	2.016	0.51	medium	Isolated spectrum	Rare cases	p95(mean kNN distance / median population mean kNN distance) in PCA-score space; not LOFalert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2)
Structure of the dataset	Isolation Forest score ? 95th percentile of the negated Isolation Forest score in retained PCA-score space. Larger raw scores indicate easier isolation under that model; the scale is implementation-dependent. The displayed alert reuses structure complexity rather than directly mapping this raw score. Scoring: alert follows structure complexity; raw score is implementation-dependent. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable. structure.isolation_forest_score_p95	0.59485	0.67	medium	Atypical spectrum	Various causes	p95 IsolationForest anomaly score on PCA scoresalert follows structure complexity; raw score is implementation-dependent

PCA score plot — explore spectral variation

Source bhi2018_plot_spectra: 80 of 80 source observations are plotted. PC1 captures 85.9% and PC2 10.1% of the reported spectral variance: 96.0% together; 4.0% remains outside this view.

Principal component analysis (PCA) summarizes spectral differences without using targets. The horizontal PC1 axis captures the largest variation; vertical PC2 captures the next largest direction of variation. Each dot is one observation; repeated measurements need not be distinct samples. Scores are projection coordinates, not wavelengths or target values. Nearby dots have similar projections; separate groups or isolated dots are leads to inspect in the original spectra and acquisition metadata. Overlapping dots can differ in omitted dimensions. The dots have no class or batch labels, so this view cannot identify groups, confirm defective observations, explain a chemical cause or measure prediction accuracy.

View plotted PCA coordinates

Point numbers identify rows in this displayed coordinate list, not original observation or sample IDs. Non-integer coordinates use up to five significant digits. Original IDs, component loadings, preprocessing details and the observation selection method are not supplied with this plot.

Plotted observation positions (not original sample IDs)
Point	PC1 score	PC2 score
1	-37.922	-16.524
2	46.609	-15.28
3	-55.111	36.28
4	63.989	10.911
5	-72.486	32.519
6	39.032	13.363
7	-25.339	21.496
8	48.363	-6.267
9	-42.714	-10.761
10	111.93	22.618
11	40.347	13.737
12	-28.788	-15.726
13	49.856	1.7772
14	111.18	47.356
15	-50.689	-20.37
16	26.574	-16.303
17	-72.457	-10.408
18	-113.98	-3.6711
19	-12.479	-27.481
20	-51.01	2.0308
21	-31.518	-24.307
22	-53.052	-12.96
23	-64.117	-15.153
24	-1.0591	-9.0457
25	81.873	-23.258
26	6.0522	-19.697
27	-58.032	-8.5431
28	-45.704	-6.3795
29	-108.03	-1.2663
30	-32.506	-18.976
31	-40.167	-10.349
32	-113.85	7.0311
33	-42.657	-8.6747
34	-67.675	-7.0373
35	-38.548	-1.026
36	-18.638	-12.116
37	72.495	-33.875
38	-1.5912	-15.504
39	-30.332	-8.7252
40	-72.023	-2.2456
41	-59.329	8.9009
42	-32.536	-18.252
43	17.987	-8.1473
44	-12.063	-16.282
45	37.28	-3.006
46	-42.427	-8.6419
47	-108.73	11.88
48	-15.889	-15.548
49	-44.298	-2.5953
50	-22.565	-2.1132
51	-37.154	3.5163
52	-39.888	-5.4816
53	-18.025	-9.7514
54	20.093	-22.162
55	17.731	0.0082557
56	92.77	-2.5008
57	85.085	-22.754
58	-67.662	23.317
59	64.332	0.8136
60	45.387	-8.2818
61	94.238	-25.688
62	76.669	-25.476
63	16.219	-5.4899
64	72.465	-25.7
65	-50.521	3.3464
66	-91.076	26.147
67	-52.709	72.194
68	9.8299	28.851
69	2.7692	29.363
70	-6.1336	18.53
71	97.739	-10.928
72	90.503	-23.261
73	60.682	45.428
74	75.983	33.492
75	99.9	0.77284
76	33.019	49.221
77	67.7	11.984
78	62.78	9.0057
79	100.66	26.149
80	43.347	31.949

PCA explained variance — how many dimensions?

Each bar reports the share of spectral variance explained by a component. Concentration in the first components indicates correlated wavelengths. This is not variance explained in the prediction target. This projection may use a different observation sample from the score plot.

View explained variance values

Variance per displayed principal component
Component	Spectral variance
PC1	85.89%
PC2	10.10%
PC3	2.70%
PC4	0.48%
PC5	0.34%
PC6	0.15%
PC7	0.09%
PC8	0.07%
PC9	0.04%
PC10	0.03%

X-Y spectral correlation 3

Pearson r measures the linear association between each spectral channel (X) and a numeric target (Y), from −1 to +1. The table reports peak and average absolute correlation and the share of channels with |r| ≥ 0.5. These are in-dataset associations, not validation scores or proof of causation.

Spectral association with lignin

Read Pearson correlation (−1 to +1) along the spectral axis (nm); the solid curve shows |r| and the dashed curve shows signed r.

Spectral association with lma_g_m2

Read Pearson correlation (−1 to +1) along the spectral axis (nm); the solid curve shows |r| and the dashed curve shows signed r.

Spectral association with nitrogen

Read Pearson correlation (−1 to +1) along the spectral axis (nm); the solid curve shows |r| and the dashed curve shows signed r.

Target	max \|r\|	axis @ max	mean \|r\|	\|r\| ≥ .5
lignin	0.591	720	0.287	4.8%
lma_g_m2	0.644	541	0.311	8.6%
nitrogen	0.442	682	0.171	0.0%

Metric interpretation reference

Definitions, formulas and possible explanations for the reported values. A possible cause requires verification against acquisition records. Reading guide

Metric catalog 30

Look up each measured quantity, its calculation and the heuristic alert mapping before interpreting a high or low value.

Family	Metric	What it measures	Higher values suggest	Lower values suggest	Possible causes	Calculation / score
Data integrity	`Invalid/masked cell count (owner report)`	Null or masked cells in the evaluated population	To investigate	Complete according to the declared mask	Missing data or validity mask; not evidence of an acquisition cause	Native Preflight count of cells unavailable under x/null and valid_mask; denominator stays in native_metric.denominators.n_cells. Not legacy NaN ratio.
Data integrity	`NaN ratio`	Missing data	Corrupted spectrum	Full spectrum	Acquisition/export error	count(isnan(X)) / X.sizealert = min(1, nan_ratio / 0.05)
Data integrity	`Inf count`	Infinite values	Corruption	Normal	Invalid calculations	count(isinf(X))alert = min(1, inf_count / 1)
Data integrity	`Zero ratio`	Null columns or cells	Truncated spectrum	Normal	Export, saturation	count(X == 0) / count(finite X)alert = min(1, zero_ratio / 0.05)
Overall amplitude	`Mean reflectance`	Average level	Too light / background visible	Too dark	Background, geometry	mean(X finite)alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent
Overall amplitude	`Area under curve`	Overall intensity	Illumination difference	Normal	Probe distance	trapezoid(mean_spectrum, spectral_axis)alert reuses baseline/shape drift because area scale depends on axis and units
Overall amplitude	`Peak-to-peak (PTP)`	Dynamic range	Strong variability	Flat spectrum	Saturation	max(mean_spectrum) - min(mean_spectrum)alert increases when dynamic range is abnormally flat
Overall amplitude	`Variance`	Spectral variability	Normal or heterogeneous	Flat spectrum	Bad contact	var(X finite)alert increases when variance/dynamic range is abnormally flat
Noise	`Roughness sigma proxy`	High frequency noise	Noisy	Stable	Lamp, detector	median MAD(second derivative) * 1.4826 / sqrt(6)alert = noise_rms / signal_scale, saturated at 5%
Noise	`Roughness S/N proxy`	Signal quality	Good signal	Bad signal	Acquisition	mean(abs(X)) / noise_rmsalert decreases with SNR dB; >=40 dB is treated as low alert
Noise	`Bandwise roughness S/N proxy`	Localized noise	Reliable zone	Problem area	Detector	min(abs(mean_spectrum) / local second-derivative noise)alert decreases with worst-band SNR dB; >=35 dB is treated as low alert
Local artifacts	`Spike count`	Narrow peaks	Artifacts	Clean spectrum	Cosmic rays, splice	count robust outliers in second derivativealert follows spike_rate, saturated at 1%
Local artifacts	`Spike rate`	Peak density	Suspicious spectrum	Normal	Interpolation	spike_count / (n_samples * (n_features - 2))alert = min(1, spike_rate / 0.01)
Local artifacts	`Jump count`	Discontinuities	Detector connection	Continuous	Splice	count robust outliers in first derivativealert follows jump_rate, saturated at 1%
Local artifacts	`Jump rate`	Jump frequency	Spectral problem	Normal	Calibration	jump_count / (n_samples * (n_features - 1))alert = min(1, jump_rate / 0.01)
Local artifacts	`Clip fraction`	Saturation	Clipping	Normal	Detector saturated	fraction of finite cells equal to repeated min/max extremaalert = min(1, clip_fraction / 0.01)
Spectral form	`Baseline slope`	Overall slope	Drift	Stable	Illumination	linear slope of mean_spectrum over normalized axisalert = abs(slope / signal_scale), saturated at 0.5
Spectral form	`Curvature RMS`	Curvature	Unusual shape	Smooth	Background, splice	median RMS(second derivative per spectrum)alert = curvature_rms / signal_scale, saturated at 1%
Spectral form	`D1 RMS`	Local variability	Structured spectrum	Flat	Biology or artifact	median RMS(first derivative per spectrum)alert = d1_rms / signal_scale, saturated at 5%
Multivariate outliers	`PCA Q (SPE)`	Not explained by PCA	Atypical spectrum	Compliant	Artifact, mixture	p95(Q/SPE residual) / median(Q/SPE residual)alert = min(1, pca_q_ratio / 8)
Multivariate outliers	`Hotelling T²`	Extreme in PCA	Extreme but consistent	Central	Natural variability	p95(Hotelling T2) / median(Hotelling T2)alert = min(1, hotelling_t2_ratio / 8)
Multivariate outliers	`Mahalanobis H`	Distance to cloud	Global outlier	Normal population	Different domain	p95(sqrt(T2)) / median(sqrt(T2))alert = min(1, mahalanobis_h_ratio / 4)
Comparison to average spectrum	`RMS to mean spectrum`	Average distance	Different spectrum	Typical	Domain shift	p95 RMS distance to dataset mean spectrumalert = RMS_p95 / signal_scale, saturated at 25%
Comparison to average spectrum	`Spectral Angle Mapper (SAM)`	Difference in shape	Different shape	Similar	Background, geometry	p95 spectral angle to dataset mean spectrumalert = min(1, SAM_p95 / 0.35 rad)
Repeatability	`RMS intra-ID`	Reproducibility	Poor repeatability	Stable	Positioning	median RMS distance to repeated-sample centroidalert = RMS_intra_ID / signal_scale, saturated at 10%
Repeatability	`SAM intra-ID`	Shape variation	Unstable	Stable	Acquisition	median SAM to repeated-sample centroidalert = min(1, SAM_intra_ID / 0.15 rad)
Repeatability	`CV intra-ID`	Internal variability	Bad control	Stable	Operator	median within-ID band CValert = min(1, CV_intra_ID / 0.25)
Structure of the dataset	`PCA score density`	Clusters	Subpopulations	Homogeneous	Different batches	1 / median kNN distance in PCA score spacealert follows density_cv/profile structure complexity, not raw density alone
Structure of the dataset	`kNN distance ratio (not LOF)`	Local anomaly	Isolated spectrum	Normal population	Rare cases	p95(mean kNN distance / median population mean kNN distance) in PCA-score space; not LOFalert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2)
Structure of the dataset	`Isolation Forest score`	Global anomaly	Atypical spectrum	Normal	Various causes	p95 IsolationForest anomaly score on PCA scoresalert follows structure complexity; raw score is implementation-dependent

Technology-specific extensions

Find measurement-specific checks and acquisition context that can help interpret a source's spectral patterns.

Technology	Adaptations / metrics	Target anomalies	Practical guidance
UV-Vis 300-1000 nm	Baseline, overall slope, drift at edges 300-350 and 900-1000; metrics by zones	Stray light, bad white, saturation, weak signal at the ends	The edges are often unstable; also calculate edge/middle scores.
UV-Vis 300-1000 nm	Saturation / clipping near max absorbance or max reflectance	Signal clipped	Very important if strong absorption.
UV-Vis 300-1000 nm	Red-edge, peak position, band ratios for vegetation	Biological shift or optical artifact	Helps distinguish real change from acquisition problem.
UV-Vis 300-1000 nm	Smoothness / roughness index	High frequency noise	Often more informative than the SNR alone.
MIR / ATR-FTIR	ATR contact quality index: overall intensity, total area, depth of key bands	Poor crystal-sample contact	Crucial: many anomalies come from the ATR contact.
MIR / ATR-FTIR	CO2 / H2O atmospheric bands	Bad atmospheric correction	Frequent spurious peaks.
MIR / ATR-FTIR	Baseline curvature / rubber-band residual	Diffusion, contact, baseline drift	Very useful before PCA.
MIR / ATR-FTIR	Peak position shift	Bad spectral alignment/calibration	Important in FTIR because small shifts count.
MIR / ATR-FTIR	Band area ratios on known bands	Chemically incoherent spectrum	To be adapted by matrix: polysaccharides, proteins, lipids, etc.
HS-MS	Total Ion Current (TIC), Base Peak Intensity (BPI)	Weak injection, unstable ionization	MS equivalent of the global spectral level.
HS-MS	Number of peaks detected	Poor or too noisy spectrum	Too little = bad signal; too much = noise/contamination.
HS-MS	Mass accuracy / m/z drift	Mass calibration problem	Fundamental in HRMS.
HS-MS	Retention time drift if LC/GC-MS	Chromatographic drift	Monitor on standards/QC pools.
HS-MS	Blank contamination score	Contaminants / carry-over	Compare samples vs blanks.
HS-MS	Internal standard CV	Instrumental variability	Very robust if standards available.
HS-MS	Missingness by feature	Detection instability	Crucial for filtering variables.
With repetitions	Intra-sample RMS	Overall repeatability	Applicable to all technologies.
With repetitions	SAM / intra-sample correlation	Shape repeatability	Very useful for spectra.
With repetitions	Intra-sample CV per band / feature	Local repeatability	Detects unstable areas.
With repetitions	ICC or variance components	Share of sample vs. technical variance	Very useful if several repetitions per sample.
With repetitions	Distance to intra-ID centroid	Aberrant repetition	Allows you to flag the bad repetition rather than the entire sample.

Bug-hunting / supervised audits

Review suggested follow-up methods for suspected data issues and whether their results are available in this explorer.

Potential issue family	Suggested methods	What it detects	Explorer availability
Global spectral shift	Inter-dataset average spectrum correlation, DTW, cross-correlation, comparison of peak positions	Wavelength shift, misalignment, different interpolation	Partially calculated: cross-correlation lag and dispersion of peak positions vs. average spectrum.
Baseline / offset / gain	Regression each spectrum vs average spectrum: x = a + b ref + residual; followed by a, b, residual RMS	Additive offset, multiplicative effect, baseline drift	Calculated in reference.affine_*.
Row mixing / X-M-Y mismatch	Index check, row hash, ID duplication, intra-ID spectral distance, inconsistent labels	Mixed up rows, misaligned metadata, Y assigned to wrong spectrum	Partially covered by intra-ID repeatability; checks index/hash to add to the canonical pipeline.
Information leak / poorly split repetitions	GroupKFold by sample_id vs StratifiedKFold random; auditing partitions by sample_id	Artificially good performance due to repetitions	Requires splits and benchmark model; not calculated by the descriptive card.
Label bugs	Close samples in X but different Y, confident learning, systematic FP/FN errors	Inverted Y, input errors, ambiguous classes	Requires Y and/or model; recommended for supervised explorer.
Hidden subdomains	PCA/UMAP/t-SNE + unsupervised clustering + association with dataset/Y/date/operator	Batches, campaigns, probes, backgrounds not provided	Partially calculated by PCA and distance ratio kNN (not LOF); UMAP/t-SNE outside the static card.
Unknown localized artifacts	Wavelength x dataset map: mean difference, variance difference, KS per wavelength	Unanticipated anomalous spectral regions	To be calculated at the bank level when several datasets share a spectral axis.
Instrumental breaks	Discontinuities in derivatives, changepoint detection	Splice, detector connection, local jump not provided	Calculated by jump/spike rates; more advanced changepoint to add.
Spectral mixing/contamination	NMF / unmixing / reconstruction by convex hull	External component: background, plastic, soil	Not calculated automatically; requires component assumptions or large library.
Unstable but predictive features	Model importance vs. QC instability by variable	Model that learns an artifact rather than a biological signal	Requires supervised model; recommended for benchmark reporting.

Curated scientific variables

Scientific roles derived from the source contract. M denotes experimental metadata; Y denotes informative response variables. Reading guide

9 experimental metadata fields (M) · 3 informative targets (Y).

Variable	Role	Distinct values	Complete	Range / balance
plot__plot_num	M	80	100%	2 – 235
sample_date	M	29	100%	IR 5
image_date	M	8	100%	IR 1.375
genus	M	7	100%	IR 6.5
species	M	10	100%	IR 7
usda_species_code	M	10	100%	IR 7
common_name	M	10	100%	IR 7
latitude	M	80	100%	43.65 – 43.65
longitude	M	80	100%	-89.8 – -89.79
lignin	Y	80	100%	2.445 – 23.22
lma_g_m2	Y	80	100%	38.37 – 147.7
nitrogen	Y	80	100%	0.6096 – 4.69

Identifiers and sampling groups remain available separately. Statistics count exported observations.

Variables

Declared prediction targets and recorded metadata, with distributions and missing-value counts. Identifiers are not automatically prediction targets. Reading guide

Targets 3

lignin

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	12.31 ± 4.62
Median ? 50th percentile: half the non-missing values lie at or below it.	11.71
Range ? Smallest and largest recorded non-missing values.	2.445 – 23.22
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.375
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.13 / -0.32
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	yes

lma_g_m2

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	89.44 ± 20.5
Median ? 50th percentile: half the non-missing values lie at or below it.	89.27
Range ? Smallest and largest recorded non-missing values.	38.37 – 147.7
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.23
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.047 / 0.48
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	yes

nitrogen

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	2.281 ± 0.704
Median ? 50th percentile: half the non-missing values lie at or below it.	2.344
Range ? Smallest and largest recorded non-missing values.	0.6096 – 4.69
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.309
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.059 / 1
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	yes

Metadata 16

plot_id

metadata · categorical

Compare recorded category counts and class balance. No distribution plot is available; use the recorded statistics below.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	80
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	1
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	1
Top class ? Most frequent recorded category and its observation count.	bhi2018_p002 (1)

plot_num

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	80.3 ± 68.4
Median ? 50th percentile: half the non-missing values lie at or below it.	65.5
Range ? Smallest and largest recorded non-missing values.	2 – 235
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.852
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	1.3 / 0.75
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	no

sample_date

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	29
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	0.97
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	5
Top class ? Most frequent recorded category and its observation count.	20180518 (5)

dbh_m

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 2
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	0.6181 ± 0.171
Median ? 50th percentile: half the non-missing values lie at or below it.	0.605
Range ? Smallest and largest recorded non-missing values.	0.319 – 1.168
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.276
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.64 / 0.43
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	no

image_date

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	8
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	1
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	1
Top class ? Most frequent recorded category and its observation count.	20180604 (11)

calcium

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	0.9321 ± 0.678
Median ? 50th percentile: half the non-missing values lie at or below it.	0.7981
Range ? Smallest and largest recorded non-missing values.	0.02536 – 3.262
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.727
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	1.2 / 1.9
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	no

phosphorus

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	0.1865 ± 0.0691
Median ? 50th percentile: half the non-missing values lie at or below it.	0.1746
Range ? Smallest and largest recorded non-missing values.	0.07227 – 0.4436
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.37
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.93 / 1.6
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	no

potassium

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	0.8451 ± 0.485
Median ? 50th percentile: half the non-missing values lie at or below it.	0.777
Range ? Smallest and largest recorded non-missing values.	0.004003 – 2.337
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.573
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.6 / 0.48
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	yes

tot_phenolics

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	8.803 ± 4.74
Median ? 50th percentile: half the non-missing values lie at or below it.	7.264
Range ? Smallest and largest recorded non-missing values.	2.456 – 20.29
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.539
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	0.89 / -0.26
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	no

chl_a_mmol_m2

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	340 ± 134
Median ? 50th percentile: half the non-missing values lie at or below it.	382.2
Range ? Smallest and largest recorded non-missing values.	24.19 – 556.2
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	0.394
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	-0.64 / -0.54
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	no

genus

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	7
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	0.85
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	6
Top class ? Most frequent recorded category and its observation count.	Quercus (26)

species

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	10
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	0.92
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	7
Top class ? Most frequent recorded category and its observation count.	rubra (14)

usda_code

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	10
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	0.92
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	7
Top class ? Most frequent recorded category and its observation count.	QURU (14)

common_name

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Classes ? Number of distinct recorded category values.	10
Balance (entropy) ? Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.	0.92
Imbalance ratio ? Largest class count divided by smallest class count; 1 indicates equal counts.	7
Top class ? Most frequent recorded category and its observation count.	Red oak (14)

latitude

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	43.65 ± 0.00226
Median ? 50th percentile: half the non-missing values lie at or below it.	43.65
Range ? Smallest and largest recorded non-missing values.	43.65 – 43.65
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	5.18e-05
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	-0.42 / -0.7
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	yes

longitude

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n / missing ? Recorded observation count and number of missing values for this variable.	80 / 0
Mean ± SD ? Arithmetic mean and standard deviation in the variable's stated unit.	-89.79 ± 0.00256
Median ? 50th percentile: half the non-missing values lie at or below it.	-89.79
Range ? Smallest and largest recorded non-missing values.	-89.8 – -89.79
CV ? Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.	2.85e-05
Skew / kurtosis ? Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.	-0.1 / -0.63
Normal? ? Descriptive normality-test result. A non-normal target is not automatically a defective variable.	yes

Constant metadata 14

Fields with one recorded value provide context for every observation; they have no distribution to compare.

SpectralRep1
veg_typetree
datasetEcoSIS
n_points_original187
dimensions1D
axis_notewavelengths_as_provided_by_EcoSIS_resource
official_data_urlhttps://data.ecosis.org/dataset/seasonal-canopy-spectra-and-traits--blackhawk-island--wi
official_package_urlhttps://ecosis.org/package/seasonal-canopy-spectra-and-traits--blackhawk-island--wi
licenseOpen Data Commons Public Domain Dedication and License (PDDL)
license_urlhttp://www.opendefinition.org/licenses/odc-pddl
rights_statusexplicit_open
usage_scopepublic_reuse_possible
public_release_allowedTrue
notesTask-specific EcoSIS target not present in active parent Y table, wavelengths kept as provided by EcoSIS resource, no project interpolation.

Alignment

How observations join to physical sample IDs. Multiple exported rows are not necessarily independent samples or technical repetitions. Reading guide

Alignment level	sample
Sample id available	yes
Samples	80
Observations (total)	80
Exported observations per unit	min 1 · mean 1 · max 1
Identity status	available
Documented technical repetitions	ND
Source unit	Multiplicity of the declared identifier; protocol not established

Splits

Partitions supplied by the original dataset. They are documented here and are not applied automatically when loading. Reading guide

No original partition summaries are recorded. Loading does not create a train/test split.

Load this dataset

Use the appropriate route below: an assembled dataset, original raw resources, or the documented access procedure. Reading guide

# pip install nirs4all-datasets
from nirs4all_datasets import get

ds = get("ecosis_seasonal_canopy_spectra_and_traits_blackhawk_island_wi_reflectance_nirs")            # DOI-pinned, checksum-verified, cached
X, y = ds.x(), ds.y()
print(X.shape, y.shape)

card.json croissant.jsonIdentity metadata only — the dataset bytes live at the origin / DOI.

EcoSIS seasonal canopy spectra and traits, Blackhawk Island, WI

Provenance & citation

Governance & integrity

Dataset property explorer

How to read the profile

Profile axes

Possible explanations to investigate

Spectral sources

bhi2018_plot_spectra

Spectral signal summary

Sampling

Signal & quality

Integrity & artefacts

Shape & reference

Outliers & repeatability

Dimensionality (PCA)

PCA score plot — explore spectral variation

PCA explained variance — how many dimensions?

Spectral association with lignin

Spectral association with lma_g_m2

Spectral association with nitrogen

Metric interpretation reference

Curated scientific variables

Variables

Targets 3

lignin

lma_g_m2

nitrogen

Metadata 16

plot_id

plot_num

sample_date

dbh_m

image_date

calcium

phosphorus

potassium

tot_phenolics

chl_a_mmol_m2

genus

species

usda_code

common_name

latitude

longitude

Alignment

Splits

Load this dataset