← Back to report

Methods & references

How each number in the report is calculated, and the sources behind it.

AidWeather Report · Methods

From weather observations to usable evidence

This companion explains what was measured, how summaries were calculated, and what the results can—and cannot—support.

NASA POWER via AidBio aidweather Download & Install

What this report summarizes

The report describes daily environmental conditions for Feb 18, 2022 – Jul 05, 2022 at the configured coordinates. Available variables in this run include T2M, T2M_MAX, T2M_MIN, PRECTOTCORR, RH2M, WS10M, WD10M, PS, ALLSKY_SFC_SW_DWN, ALLSKY_SFC_PAR_TOT, VPD_MEAN, VPD_MAX, VPD, GDD, CUM_GDD, ET0, KC, ETC.

Missing variables are omitted rather than estimated. The report is organised as a verdict followed by chapters (warmth, water, atmospheric demand, stress events); supporting variables are in the appendix, and the caveats and definitions behind each chart are on the Notes page (notes.html).

Location map

The interactive report shows the configured location as a single pin on a map image made when the report was generated (OpenFreeMap tiles, derived from OpenStreetMap). The pages carry no coordinates and the reader's browser contacts no map server. The map never draws a field boundary, and the weather values describe the surrounding grid cell, not the exact pin. If the map cannot be made, the report is written without it. The print report has no map.

OpenFreeMap © OpenMapTiles, data from OpenStreetMap contributors.

Data retrieval

NASA POWER gridded values are fetched through AidBio’s aidweather application[1]. The default GDD method requests a separate hourly T2M series in NASA POWER's local solar time and saves it as hourly_temperature.csv. Daily values, ET0, ETc and Kc are saved in weather_data.csv so every headline number can be reproduced from the exported data.

View aidweather on GitHub ↗

Baseline and anomalies

The baseline is the same NASA POWER grid cell and calendar window in each of 30 years (1991–2020). Percentile statements count the share of baseline seasons strictly below this season. Heat accumulation is recomputed for the season and every baseline year with the single-sine method so they are comparable.[1]

Atmospheric demand (VPD)

Saturation vapour pressure es = 0.6108·exp(17.27·T/(T+237.3)) (FAO-56 Eq. 11)[2]. VPD = max(0, es − ea). ea = es(T2M) × RH2M/100, or es(T2MDEW) when dewpoint is requested. Mean VPD uses es(T2M); peak-demand VPD uses es(Tmax) with the day's mean ea, so it estimates demand at the daily maximum rather than measuring afternoon VPD. The reference line is the crop registry's sourced VPD value when one exists; otherwise it is a generic 2 kPa reference, labelled as such on the chart, because breakpoints are species-specific[7]. VPD is atmospheric demand, not a soil-water or plant-water-status measurement.

Stress events

Dry day = precipitation under 1 mm; a dry spell is at least seven consecutive dry days; the longest dry run is the largest consecutive-dry-day count (ETCCDI definitions)[6]. Cool night = Tmin at or below the crop's sourced base temperature (10 °C without a crop). Heat day = Tmax at or above the crop registry's sourced heat threshold when one exists, otherwise a generic 35 °C reference. Thresholds without a registry source carry a "generic" label on the chart. Near-freezing values do not by themselves establish canopy-level frost. The former 0–100 stability score was removed: it had no agronomic reference.

Unit handling

Every parameter's unit is determined deterministically from NASA POWER's own API metadata and AidBio's aidweather client — never assumed from memory — before any number reaches a chart, a card, or the AI prompt. Solar radiation is canonicalized to MJ/m²/day; an additional kWh/m²/day figure is always computed explicitly (divide by 3.6), never substituted for the MJ value. A unit label can only change via an explicit numerical conversion, never a relabel.

Reference evapotranspiration (ET0) and crop water need (ETc)

ET0 uses the FAO-56 Penman–Monteith method[2]. Daily mean temperature is (Tmax + Tmin)/2 (Eq. 9), not the 24 h NASA mean. Actual vapour pressure uses mean relative humidity (Eq. 19). Surface pressure is the measured NASA POWER value when present; otherwise Eq. 7 from the supplied elevation; otherwise the 101.3 kPa sea-level standard (this run: measured). Wind is converted from 10 m to 2 m (Eq. 47); net radiation uses Eq. 38–40 with Rso from elevation when given, otherwise 0.75·Ra. ETc = Kc × ET0 with the staged single-Kc curve (constant initial, linear development, constant mid-season, linear late-season decline; Eq. 66)[5]. Packaged maize stage lengths are 20/35/40/30 days (FAO-56 Table 11, Nigeria humid, June planting), scaled to the planting–harvest window unless explicit lengths are supplied; they are indicative, not locally calibrated. Maize Kc_end is 0.60 (Table 12, harvest at high grain moisture); it is 0.35 only for harvest after complete field drying, which must be stated through water.kc.late. FAO-56 staged single Kc: 0.30 → 1.20 → 0.60; default stage lengths scaled to supplied planting–harvest window; initial 22.08 days, development 38.64 days, mid 44.16 days, late 33.12 days. FAO-56 Table 11, maize grain, Nigeria humid, June; indicative timing only, scaled to supplied planting–harvest window, not local calibration. FAO-56 Table 12, field maize (grain), Kc end 0.60 for harvest at high grain moisture ETc is an estimate of crop water need under standard conditions, not measured use. The rain-minus-ETc balance is a climatic comparison; it excludes soil storage, runoff, irrigation, rooting depth and stress-driven reduction of evapotranspiration. See docs/scientific_interpretation.md ↗.

Crop registry and crop-aware analysis

Crop-specific values (base temperature, upper cutoff, crop coefficients) come from a versioned registry, not from the AI model. Each numeric value carries a source, confidence level and citation, or is left unset when no defensible general value exists: a documented gap, not an invented number. Only --crop activates crop-aware analysis; cultivar, planting date, harvest date and irrigation default to unknown. A custom registry can be supplied with --crop-config.

Thermal time (GDD)

GDD method: NASA POWER modeled hourly T2M in local solar time; daily GDD = sum of max(0, min(T2M, 30 °C) − base) over all 24 hourly values divided by 24. Incomplete days stop report generation; the hourly inputs are saved in hourly_temperature.csv.

With no crop, GDD uses a generic 10 °C base and the report says so. With --crop, GDD uses the crop's sourced base temperature (and upper cutoff, unless you set one explicitly); a crop without a sourced base stays generic. With a planting date, GDD accumulates from planting; with a harvest date, it stops at harvest. With neither, a window shorter than a year accumulates over the whole window, and a longer window resets each 1 January. Thermal time does not by itself establish crop stage or harvest timing[3]. The hourly and single-sine methods[4] are not interchangeable without crop-model calibration.

AI summaries and limitations

Headlines, chapter findings, the purpose and how-to-read text of every chart, and every number are deterministic. When a model is configured it adds a labelled interpretation under a chart and a labelled summary under the verdict, under an explicit evidence contract: units, missing-value handling and calibrated language are specified in every prompt, and the model is told never to rescale or invent values. If the local model is unavailable, the deterministic text is used.

Data source limitations

NASA POWER is a roughly 0.5° gridded product (resolution varies by variable). It smooths local extremes of rainfall and temperature and is not a substitute for a station measurement.

Point values represent the surrounding grid cell, not necessarily conditions at the exact coordinates.

For technical interpretation

Use the report to identify patterns, compare the period with the baseline, and decide what deserves field follow-up. Treat anomalies as departures from a reference distribution, not as automatic agronomic recommendations. Interpret every result alongside crop, soil, management and observation context.

References

  1. NASA POWER Project. Data methodology: MERRA-2 grid-cell averages, solar-local-time days. source ↗
  2. Allen, R.G., Pereira, L.S., Raes, D., Smith, M. (1998). Crop evapotranspiration: guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56. source ↗
  3. McMaster, G.S., Wilhelm, W.W. (1997). Growing degree-days: one equation, two interpretations. Agricultural and Forest Meteorology 87(4):291-300. source ↗
  4. Baskerville, G.L., Emin, P. (1969). Rapid estimation of heat accumulation from maximum and minimum temperatures. Ecology 50:514-517. source ↗
  5. FAO-56 Tables 11 and 12: crop growth-stage lengths and single crop coefficients (Kc ini, mid, end). source ↗
  6. Zhang, X. et al. (2011). Indices for monitoring changes in extremes based on daily temperature and precipitation data. WIREs Climate Change 2:851-870 (ETCCDI indices). source ↗
  7. Grossiord, C. et al. (2020). Plant responses to rising vapor pressure deficit. New Phytologist 226:1550-1566. source ↗
  8. Pezzopane, J.R.M., Pedro Júnior, M.J., Camargo, M.B.P., Fazuoli, L.C. (2008). Exigência térmica do café arábica cv. Mundo Novo no subperíodo florescimento-colheita. Ciência e Agrotecnologia 32(6):1781-1786. source ↗