ECSS Space Radiation — Biological-Effects Uncertainty (space-systems/ecss/e1012-bio-unc)
Use when the task is to evaluate and propagate uncertainties in a space-radiation
biological-effects assessment under ECSS-E-ST-10-12C §11.6 — verifying that all
required uncertainty sources are identified and quantified, combining them in the
lognormal quadrature model, deriving confidence bounds around the risk point
estimate, and confirming no required source has been omitted.
Domain quick reference
- §11.6 identifies five independent uncertainty sources that must each be
accounted for in a biological-effects risk assessment. Radiation quality
factor uncertainty reflects the limited experimental basis for Q(L) at high
LET (HZE ions above roughly 100 keV/µm carry the largest uncertainty in
biological effectiveness). Dose-and-dose-rate effectiveness factor (DDREF)
uncertainty arises from extrapolating high-dose, high-dose-rate laboratory
exposures to the low-dose-rate chronic irradiation of the space environment;
the DDREF is treated as a range (typically 1.5–2.5) rather than a fixed
value. Cancer-risk coefficient uncertainty is the statistical uncertainty in
the risk-per-unit-dose-equivalent derived from epidemiological cohorts,
primarily A-bomb survivor data. Transfer model uncertainty reflects the
systematic error in mapping risk estimates from the reference Japanese cohort
to the astronaut population. Dosimetry and transport uncertainty captures
errors in the computed or measured absorbed dose, including transport code
accuracy and shielding geometry approximations.
- Each source is treated as lognormally distributed. Its magnitude is
expressed as a sigma in natural-log space (sigma_log), which is the natural
logarithm of the geometric standard deviation. A sigma_log of ln(2)/1.645
means the 95th-percentile upper bound is a factor of 2 above the median.
- Independent uncertainty sources combine in quadrature in log space: the
combined sigma_log is the square root of the sum of the squared individual
sigma_log values. This preserves the lognormal model and avoids the
under-estimation that results from arithmetic addition of asymmetric
uncertainties.
- The 95% confidence interval is obtained by multiplying and dividing the
point-estimate risk by exp(z_0.95 × combined_sigma_log), where z_0.95 ≈
1.645 is the one-sided standard normal quantile at the 95th percentile.
The resulting interval is asymmetric in linear space (the upper bound is
farther from the median than the lower bound) and symmetric in log space.
- Qualitative uncertainty levels are assigned based on the 95th-percentile
CI factor: less than ×2 is low; ×2 to ×4 is moderate; ×4 to ×8 is high;
×8 or more is very high.
Workflow
- Enumerate the five required uncertainty sources for the assessment: quality
factor, DDREF, risk coefficient, transfer model, and dosimetry. Confirm
that each source appears in the uncertainty budget; flag any source that is
absent, because its omission invalidates the combined confidence interval.
- For each present source, assign a sigma_log value derived from the
supporting evidence (experimental spread, published uncertainty analysis, or
expert elicitation). Document the basis for each assigned sigma_log; a
sigma_log that is simply assumed without basis must be flagged as
unquantified.
- Validate each component: reject a component whose sigma_log is not a
positive finite number, and reject a component whose source type is not one
of the five required identifiers. An unknown source type indicates either a
labelling error or an undocumented additional source that requires explicit
engineering justification before it enters the combined budget.
- Combine the validated components: compute the combined sigma_log as the
square root of the sum of the squared individual sigma_log values. Do not
add the sigma_log values directly (linear addition overstates the
combination for independent lognormal components).
- Derive the 95% confidence interval: multiply the point-estimate risk by
exp(z × combined_sigma_log) for the upper bound and divide by the same
factor for the lower bound, where z is the standard normal quantile at the
chosen percentile (1.645 for 95%).
- Label the combined uncertainty level from the 95th-percentile CI factor: if
upper_CI / point_estimate < 2, the level is low; 2–4 is moderate; 4–8 is
high; above 8 is very high.
- Report the combined sigma_log, the lower and upper confidence bounds, the
qualitative level, and the list of any missing required sources. A result
with missing sources must not be accepted as a complete assessment.
Pitfalls
- Combining sigma_log values by arithmetic sum instead of quadrature: the
arithmetic sum is the correct rule only when the components are perfectly
correlated (worst-case combination); for independent sources it
over-estimates the combined uncertainty by treating every component as if it
points in the same direction simultaneously.
- Using a symmetric (normal) confidence interval for a lognormal distribution:
the correct interval is asymmetric in linear space; a symmetric ±σ interval
allows the lower bound to become negative for large uncertainty, which is
physically meaningless for a risk quantity that must remain non-negative.
- Treating DDREF as a fixed scalar rather than an uncertain quantity: DDREF
is itself a variable derived from experimental dose-response models; locking
it to a nominal value of 2 and omitting the DDREF source from the budget
understates the combined uncertainty and typically violates §11.6
requirements.
- Declaring an assessment complete when one or more required uncertainty
sources are absent: the gap does not default to zero uncertainty — it
indicates the source has not been evaluated and the combined interval is
narrower than it should be, which may falsely suggest the risk is
well-constrained.
- Applying the combined uncertainty interval directly to a point estimate
derived from a different dose-rate regime without also applying the DDREF
correction: the risk coefficient from epidemiological data applies to acute
exposures; the DDREF adjustment and its uncertainty must both be carried
through before the interval is meaningful for the chronic space exposure.
Behavior contract (gate 3)
The uncertainty-component validation, quadrature combination in log space,
95% confidence-interval derivation, qualitative level labelling, and
completeness check are exercised by the gate 3 contract test:
scripts/test_e1012_bio_unc.py against scripts/e1012_bio_unc_logic.py
(stdlib unittest, offline). Run:
python3 scripts/test_e1012_bio_unc.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and
paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
1---2name: e1012-bio-unc3description: "Use when evaluate uncertainties in a space-radiation biological-effects4 assessment under ECSS-E-ST-10-12C §11.6: identify each required uncertainty source5 (radiation quality factor, dose-and-dose-rate effectiveness factor, cancer-risk6 coefficient from epidemiology, inter-population transfer model, and dosimetry and7 transport), combine the independent log-space components in quadrature to obtain a8 geometric standard deviation, derive the 95% confidence interval around the9 point-estimate risk value, assign a qualitative uncertainty level, and flag any10 required source type absent from the budget. Trigger: ecss, e-st-10-system-scope,11 biological-effects-uncertainty, quality-factor, ddref, cancer-risk,12 confidence-interval, dose-equivalent-uncertainty."13license: Apache-2.014compliance: STANDARDS-REF15standards:16 - id: ecss17 reference-only: true18gated: false19domain: space-systems20pack: space-systems21compatibility: "agentskills.io SKILL.md; any SKILL.md host (Claude Code, Hermes, OpenClaw)"22metadata:23 source: https://skillmd.com/skills/ashfordeou/e1012-bio-unc24 domain: space-systems25 subdomain: ecss26 tags: [ecss, e-st-10-system-scope, biological-effects-uncertainty, quality-factor, ddref, cancer-risk, confidence-interval, dose-equivalent-uncertainty]27 version: 0.1.028 author: Aero Agent Skills29---3031# ECSS Space Radiation — Biological-Effects Uncertainty (space-systems/ecss/e1012-bio-unc)3233Use when the task is to evaluate and propagate uncertainties in a space-radiation34biological-effects assessment under ECSS-E-ST-10-12C §11.6 — verifying that all35required uncertainty sources are identified and quantified, combining them in the36lognormal quadrature model, deriving confidence bounds around the risk point37estimate, and confirming no required source has been omitted.3839## Domain quick reference4041- §11.6 identifies five independent uncertainty sources that must each be42 accounted for in a biological-effects risk assessment. Radiation quality43 factor uncertainty reflects the limited experimental basis for Q(L) at high44 LET (HZE ions above roughly 100 keV/µm carry the largest uncertainty in45 biological effectiveness). Dose-and-dose-rate effectiveness factor (DDREF)46 uncertainty arises from extrapolating high-dose, high-dose-rate laboratory47 exposures to the low-dose-rate chronic irradiation of the space environment;48 the DDREF is treated as a range (typically 1.5–2.5) rather than a fixed49 value. Cancer-risk coefficient uncertainty is the statistical uncertainty in50 the risk-per-unit-dose-equivalent derived from epidemiological cohorts,51 primarily A-bomb survivor data. Transfer model uncertainty reflects the52 systematic error in mapping risk estimates from the reference Japanese cohort53 to the astronaut population. Dosimetry and transport uncertainty captures54 errors in the computed or measured absorbed dose, including transport code55 accuracy and shielding geometry approximations.56- Each source is treated as lognormally distributed. Its magnitude is57 expressed as a sigma in natural-log space (sigma_log), which is the natural58 logarithm of the geometric standard deviation. A sigma_log of ln(2)/1.64559 means the 95th-percentile upper bound is a factor of 2 above the median.60- Independent uncertainty sources combine in quadrature in log space: the61 combined sigma_log is the square root of the sum of the squared individual62 sigma_log values. This preserves the lognormal model and avoids the63 under-estimation that results from arithmetic addition of asymmetric64 uncertainties.65- The 95% confidence interval is obtained by multiplying and dividing the66 point-estimate risk by exp(z_0.95 × combined_sigma_log), where z_0.95 ≈67 1.645 is the one-sided standard normal quantile at the 95th percentile.68 The resulting interval is asymmetric in linear space (the upper bound is69 farther from the median than the lower bound) and symmetric in log space.70- Qualitative uncertainty levels are assigned based on the 95th-percentile71 CI factor: less than ×2 is low; ×2 to ×4 is moderate; ×4 to ×8 is high;72 ×8 or more is very high.7374## Workflow75761. Enumerate the five required uncertainty sources for the assessment: quality77 factor, DDREF, risk coefficient, transfer model, and dosimetry. Confirm78 that each source appears in the uncertainty budget; flag any source that is79 absent, because its omission invalidates the combined confidence interval.802. For each present source, assign a sigma_log value derived from the81 supporting evidence (experimental spread, published uncertainty analysis, or82 expert elicitation). Document the basis for each assigned sigma_log; a83 sigma_log that is simply assumed without basis must be flagged as84 unquantified.853. Validate each component: reject a component whose sigma_log is not a86 positive finite number, and reject a component whose source type is not one87 of the five required identifiers. An unknown source type indicates either a88 labelling error or an undocumented additional source that requires explicit89 engineering justification before it enters the combined budget.904. Combine the validated components: compute the combined sigma_log as the91 square root of the sum of the squared individual sigma_log values. Do not92 add the sigma_log values directly (linear addition overstates the93 combination for independent lognormal components).945. Derive the 95% confidence interval: multiply the point-estimate risk by95 exp(z × combined_sigma_log) for the upper bound and divide by the same96 factor for the lower bound, where z is the standard normal quantile at the97 chosen percentile (1.645 for 95%).986. Label the combined uncertainty level from the 95th-percentile CI factor: if99 upper_CI / point_estimate < 2, the level is low; 2–4 is moderate; 4–8 is100 high; above 8 is very high.1017. Report the combined sigma_log, the lower and upper confidence bounds, the102 qualitative level, and the list of any missing required sources. A result103 with missing sources must not be accepted as a complete assessment.104105## Pitfalls106107- Combining sigma_log values by arithmetic sum instead of quadrature: the108 arithmetic sum is the correct rule only when the components are perfectly109 correlated (worst-case combination); for independent sources it110 over-estimates the combined uncertainty by treating every component as if it111 points in the same direction simultaneously.112- Using a symmetric (normal) confidence interval for a lognormal distribution:113 the correct interval is asymmetric in linear space; a symmetric ±σ interval114 allows the lower bound to become negative for large uncertainty, which is115 physically meaningless for a risk quantity that must remain non-negative.116- Treating DDREF as a fixed scalar rather than an uncertain quantity: DDREF117 is itself a variable derived from experimental dose-response models; locking118 it to a nominal value of 2 and omitting the DDREF source from the budget119 understates the combined uncertainty and typically violates §11.6120 requirements.121- Declaring an assessment complete when one or more required uncertainty122 sources are absent: the gap does not default to zero uncertainty — it123 indicates the source has not been evaluated and the combined interval is124 narrower than it should be, which may falsely suggest the risk is125 well-constrained.126- Applying the combined uncertainty interval directly to a point estimate127 derived from a different dose-rate regime without also applying the DDREF128 correction: the risk coefficient from epidemiological data applies to acute129 exposures; the DDREF adjustment and its uncertainty must both be carried130 through before the interval is meaningful for the chronic space exposure.131132## Behavior contract (gate 3)133134The uncertainty-component validation, quadrature combination in log space,13595% confidence-interval derivation, qualitative level labelling, and136completeness check are exercised by the gate 3 contract test:137scripts/test_e1012_bio_unc.py against scripts/e1012_bio_unc_logic.py138(stdlib unittest, offline). Run:139python3 scripts/test_e1012_bio_unc.py140141## Compliance142143- ECSS standards are freely downloadable (ESA); cite the source and144 paraphrase per standards-map.yaml.145- compliance: STANDARDS-REF, gated: false.