Test 15 — Dams Can Emit
Supplied working report; original verdicts not validated.
Review status: Needs review. This is an archival source note, not a passed proof.
README.md
# Test 15 — Dams Can Emit ## Verdict: supported Both posted predictions hold on the Deemer et al. 2016 compilation: tropical reservoirs emit more greenhouse gas per square metre than temperate ones, and at least some tropical reservoirs exceed the fossil baseline under the pre-specified kWh conversion. ## Headline numbers | Quantity | Value | |---|---| | Reservoirs in the SI with CH4 data | 161 (of 267 total rows; 40 tropical, 121 temperate by \|lat\| ≤ 23.5°) | | Median CO2e flux, tropical | **3 666** mg CO2e m⁻² d⁻¹ (GWP100 = 28) | | Median CO2e flux, temperate | **1 016** mg CO2e m⁻² d⁻¹ | | Mann–Whitney tropical vs temperate | U = 3 478, z = 4.14, **p = 0.00003** | | Tropical reservoirs above 400 gCO2e/kWh | **19** (14 hydroelectric, 15 shallow-or-young) | | Top exceeder | Lago Olmeca (Mexico), ~13 613 gCO2e/kWh; largest hydro exceeders: Gatun 1 488, Tucuruí 1 086, Corumbá 952, Balbina 925 gCO2e/kWh | ## Conversion assumptions (stated, pre-specified) The SI gives no generation data (installed capacity is categorical), so the kWh conversion uses the pre-specified fallback: global hydro fleet average flooded area of 1.0 km² per MW and capacity factor 0.45, i.e. 3.942 GWh yr⁻¹ km⁻²; equivalently 1 mg CO2e m⁻² d⁻¹ = 0.0927 gCO2e/kWh. The fossil baseline is 400 gCO2e/kWh (CCGT lifecycle, as posted). Verdicts use GWP100 = 28 (AR5); GWP100 = 34 columns are reported as sensitivity. Where CO2 is missing in the SI, CO2e is computed from CH4 alone (a lower bound against the claim). ## What this shows and cannot show - Shows: a robust tropical–temperate gradient in reservoir surface fluxes (p = 0.00003) and many tropical hydroelectric reservoirs whose surface flux, converted with the stated area-energy assumption, exceeds a 400 gCO2e/kWh CCGT baseline several-fold (e.g. Gatun, Tucuruí, Corumbá, Balbina, Samuel, Xingó, Nam Theun 2 — all young, most shallow, all hydro). - Cannot show: measured lifecycle emissions per kWh for any specific dam. The g/kWh figures are an assumed conversion of surface flux, not a generation audit; they exclude degassing and downstream emissions (which Deemer et al. discuss but do not compile per-reservoir) and exclude carbon burial offsets. ## Limitations 1. **No measured generation.** The SI lacks energy output, so every gCO2e/kWh figure inherits the 1.0 km²/MW, 0.45 capacity-factor assumption. A reservoir with a higher energy yield per flooded area (or one that displaces more than 400 gCO2e/kWh of fossil generation) would need a different threshold; this is the principal rival explanation for the exceedance claim, and it cannot be resolved from this dataset alone. 2. **Flux method heterogeneity.** The CH4 "all estimates" column mixes diffusive-only, diffusive+ebullitive, and eddy-covariance methods; Deemer et al. show methodology materially affects reported flux. Diffusive-only estimates omit ebullition and are biased low, which acts against the claim, but the spread is real and unmeasured here. 3. **Latitude cutoff quirks.** The tropical/temperate split uses the pre-specified \|lat\| ≤ 23.5° rule; a few subtropical reservoirs (e.g. a Polish reservoir at 21.9° N) fall in the "tropical" bin by that rule. The gradient conclusion is insensitive to this (medians and p-value barely move). 4. **Gross surface flux, not net lifecycle emissions.** The comparison with fossil generation omits reservoir construction emissions, downstream degassing (up to ~0.2 Pg CO2e yr⁻¹ globally per later work) and burial; it is a flux comparison, not a full lifecycle claim, and no clinical or policy advice follows from it. ## Structure - `hypothesis.md` — claim and pre-specified criteria (written before analysis) - `run.py` — parses the cached SI workbook, computes flux and kWh conversions, tests the criteria; regenerates `results.json`, `data/*.csv`, `figures/*.png` - `results.json` — per-sub-claim verdicts and decisive numbers - `SOURCE.md` — dataset provenance, hashes, failed attempts
hypothesis.md
# Test 15 — Dams Can Emit (reservoir methane) ## Posted claim (Vol XXXII, reservoir-methane), verbatim "Warm, shallow reservoirs can emit more greenhouse gas than the fossil generation they displace." ## Posted prediction Tropical reservoirs' CH4 (and CO2) fluxes, converted to gCO2-eq per kWh, exceed the emissions intensity of fossil generation for at least some reservoirs. ## Posted refutation No reservoir exceeds the fossil equivalent. ## Source link Deemer, Harrison, Li, Beaulieu, DelSontro, Barros, Bezerra-Neto, Powers, dos Santos & Vonk 2016, "Greenhouse Gas Emissions from Reservoir Water Surfaces: A New Global Synthesis", BioScience 66(11):949–964, doi:10.1093/biosci/biw117. Supplementary data table (per-reservoir CH4/CO2 fluxes and attributes) obtained from the publisher or a documented fallback; see `SOURCE.md`. ## Pre-specified criteria for this test (written before analysis) Conversions (fixed before analysis): - Primary CH4 flux per reservoir: the SI column "mg CH4-C m-2 d-1 All Estimates Combined" (each reservoir's best available estimate, diffusive + ebullitive where measured). Secondary check: "Diffusive + Ebullitive" only, where reported. Both converted from CH4-C to CH4 by ×(16/12). - CO2-equivalent flux per reservoir = CO2-C × (44/12) + 28 × CH4 (GWP100, AR5, no climate-carbon feedbacks); a sensitivity column with GWP100 = 34 (as used by Deemer et al.) is reported but verdicts use 28. Where CO2 is missing in the SI, the CO2e flux is computed from CH4 alone — a conservative lower bound against the claim, since most reservoirs are CO2 sources. - Units: mg CO2e m⁻² d⁻¹ as compiled; for the kWh conversion, 1 mg m⁻² d⁻¹ = 1 g km⁻² d⁻¹, annualised (×365.25) over reservoir area and divided by annual electricity generation where the SI provides it. - Where the SI does not provide generation, pre-specified fallback conversion (stated as an assumption, not data): global hydro fleet average flooded area ≈ 1.0 km² per MW of installed capacity and capacity factor 0.45 (≈ 3.94 GWh yr⁻¹ per km²). The fossil baseline is 400 gCO2e/kWh (CCGT, lifecycle, as posted). Verdict criteria: 1. **Exceedance claim** — supported if at least one documented tropical reservoir (lat ≤ 23.5°, and shallow or young by the SI attributes) exceeds 400 gCO2e/kWh under the SI-provided generation where available, or under the pre-specified fallback conversion otherwise; partly supported if some reservoirs exceed under the fallback but none with SI-provided generation (or vice versa); not supported if no reservoir exceeds the baseline under any stated conversion. 2. **Tropical gradient** — supported if median tropical flux > median temperate flux with Mann–Whitney p < 0.05 (two-sided, tie-corrected normal approximation); partly supported if the direction holds but p ≥ 0.05; not supported if no gradient. 3. **Overall verdict** — supported only if both (1) and (2) hold; partly supported if exactly one holds (or the exceedance holds only under one conversion); not supported as stated if neither holds; **not testable** if the SI table cannot be obtained or digitised after documented attempts (a real effort will be made first, including fallback digitisation of the published flux table from the PDF). Distribution summaries (median, quartiles, top reservoirs) are reported regardless of the kWh conversion, since the flux gradient is itself a posted prediction.