Rain gauge accuracy rankings explained based datasheets: tipping buckets, siphons, and weighing gauges

Rain gauge uncertainty article

Rain gauge rankings explained: tipping buckets, siphons, and weighing gauges

This article translates the corrected rain-gauge ranking table into plain language. The main idea is simple: a rain gauge should be ranked by the data product it produces—instant intensity, 1-minute intensity, delayed accumulation, daily total—not by brand name alone.

Prepared for the low-power rain-gauge uncertainty project. Updated 2026-04-28.

Summary Key terms Ranking table Worked calculations Weighing and Lambrecht rain[e] Dynamic high-B condition Assumptions Sources

Plain-language summary

Best short-window floor

MeteoRain 400 Aero (B=2) ranks first when a two-bit boundary-state or equivalent phase-aware estimate is available. It directly reduces the start/stop water-state uncertainty that hurts short rainfall windows.

Best raw B=0 group

MeteoRain 533 0.075 mm, the large 0.1 mm-class gauges, and other high-area small-tip TBRs rank well because they combine small depth increments with large collection areas.

Siphon penalty

Geolux RG200 and RG400 are siphon TBRs in this model. Their siphons may improve high-rate hydraulics, but an unreported siphon fill state adds a short-window storage term.

Do not collapse products

OTT Pluvio² S RT and NRT must be separate rows. RT intensity is a current estimate; Accu NRT is a delayed accumulation product.

Lambrecht nuance

Lambrecht rain[e] products are best described here as hybrid self-emptying weighing/TBR products. Their 0.001 mm serial amount state can behave like a high-B fractional state, but only when the weighing state is timely enough.

Ranking rule

Lower required rainfall depth is better. A threshold depth such as 3.333 mm is the same for 1 minute, 10 minutes, 1 hour, or 1 day; only the equivalent mean intensity changes.

Key terms any reader needs

Tipping-bucket rain gauge (TBR): water fills a small bucket; when the bucket tips, the logger counts one pulse. A 0.1 mm bucket means each tip represents 0.1 mm of rain.

B=0: the logger knows only the completed tips. It does not know how much water is sitting inside the bucket at the start or end of a reporting window.

B=2: the logger has two bits of boundary-state information, or an equivalent fractional estimate. This does not change the bucket itself; it reduces the uncertainty of the leftover water at the window boundaries.

Siphon TBR: a siphon regulates flow into the tipping mechanism. That can help high-intensity mechanics, but if the siphon fill state is not reported, it adds another start/stop storage uncertainty.

Weighing gauge: a load cell weighs collected water. It can have excellent amount resolution, but the public output may be filtered, delayed, or split into real-time and non-real-time products.

Threshold depth: the minimum rain amount needed before the modeled uncertainty becomes smaller than a chosen target such as 1%, 2%, or 3%.

Main corrected ranking table

The table is sorted by the 3% mean-intensity threshold. A smaller Pmin 3% means the instrument reaches the target with less rainfall in the reporting window. The ranking is a model-floor ranking, not a complete field-catch or siting ranking.

Rank	Gauge / output product	A cm²	d or δ mm	Model	Vs mL	Pmin 1% mm	Pmin 2% mm	Pmin 3% mm	3% @1 min mm/h	3% @1 day mm/h	Note
1	TBR (B=2) MeteoRain 400 Aero 0.1 mm — phase-aware model	400.0	0.100	B=2	0	2.378	1.102	0.716	43.0	0.030	Best stochastic-floor row; only valid if boundary state is transmitted or reconstructed.
2	TBR (B=0) MeteoRain 533 Classic 0.075 mm	533.0	0.075	B=0	0	6.363	3.121	2.067	124.0	0.086	Corrected: 533 row is 0.075 mm only.
3	TBR (B=0) Texas/Campbell TR-525M / TE525MM 245 mm 0.1 mm	471.4	0.100	B=0	0	8.393	4.139	2.747	164.8	0.114	Large collector and 0.1 mm tip.
4	TBR (B=0) MeteoRain 400 Aero 0.1 mm; CAE PG4i 400 cm² 0.1 mm; Casella TBRG 400 cm² 0.1 mm	400.0	0.100	B=0	0	8.483	4.161	2.757	165.4	0.115	Same stochastic inputs: 400 cm², 0.1 mm, no siphon.
5	TBR (B=0) Texas TR-525-W2 200 mm 0.1 mm	314.2	0.100	B=0	0	8.687	4.211	2.779	166.7	0.116	WMO 200 mm collector class.
6	TBR (B=0) Lambrecht 15189 0.1 mm; MicroStep/Meteoservis MR2 0.1 mm option	200.0	0.100	B=0	0	9.510	4.407	2.864	171.8	0.119	Small 200 cm² 0.1 mm class.
7	Hybrid self-emptying weighing/TBR Lambrecht rain[e] / rain[e]LP / rain[e]H3 / rain[e]314 / rain[e]400 — public amount envelope	—	—	public fixed floor	—	10.000	5.000	3.333	200.0	0.139	Static public amount envelope; see high-B section before judging 1-minute real-time intensity.
8	Weighing OTT Pluvio² S Accu NRT / Accu total NRT	—	—	public fixed floor	—	10.000	5.000	3.333	200.0	0.139	Best documented delayed accumulation benchmark; split from Intensity RT.
9	TBR (B=0) Vaisala QMR102 / EML ARG100 / Campbell ARG100 500 cm² 0.2 mm	500.0	0.200	B=0	0	16.430	8.190	5.454	327.3	0.227	Large-area 0.2 mm ARG100 lineage.
10	TBR (B=0) Casella TBRG 400 cm² 0.2 mm	400.0	0.200	B=0	0	16.487	8.204	5.461	327.6	0.228	400 cm² 0.2 mm class.
11	TBR (B=0) 8 in / 324 cm² 0.2 mm class: Texas TR-525USW, Met One 370D, confirmed 8 in metric variants	324.3	0.200	B=0	0	16.569	8.225	5.470	328.2	0.228	8-inch metric class.
12	TBR (B=0) Texas TR-525-W2 200 mm 0.2 mm	314.2	0.200	B=0	0	16.585	8.229	5.472	328.3	0.228	WMO 200 mm collector class.
13	TBR (B=0) Davis 6466/6466M AeroCone 214 cm² 0.2 mm	214.0	0.200	B=0	0	16.885	8.303	5.504	330.3	0.229	Davis metric 0.2 mm row.
14	TBR (B=0) MeteoRain 200 Pro 0.2 mm; Vaisala QMR101 / Pronamic Professional 0.2 mm; Lambrecht 15189 0.2 mm; MicroStep MR2 0.2 mm; Seven Sensor Solutions 3S-RG 0.2 mm — no siphon	200.0	0.200	B=0	0	16.967	8.323	5.513	330.8	0.230	3S-RG corrected: no siphon.
15	TBR (B=0) Onset HOBO RG3-M 15.39 cm 0.2 mm	186.0	0.200	B=0	0	17.068	8.348	5.524	331.5	0.230	Small 0.2 mm class.
16	Siphon TBR (B=0) KISTERS / HyQuest TB4 Series II 0.1 mm — Vs=12 mL	628.3	0.100	B=0 + siphon	12	17.730	8.833	5.882	352.9	0.245	Siphon improves high-rate flow, but unresolved siphon state hurts short-window floor.
17	TBR (B=0) EML ARG100 optional 0.25 mm	500.0	0.250	B=0	0	20.493	10.226	6.813	408.8	0.284	Optional coarser ARG100 configuration.
18	TBR (B=0) 8 in / 0.01 in class: Texas/Campbell TR-525WS / TE525WS, RainWise Rainew, AEM/FTS 2408	324.3	0.254	B=0	0	20.927	10.416	6.934	416.0	0.289	0.01 in class.
19	TBR (B=0) Davis 6466/6466M AeroCone 214 cm² 0.01 in	214.0	0.254	B=0	0	21.173	10.478	6.961	417.7	0.290	Davis imperial row.
20	TBR (B=0) MeteoRain 200 Pro 0.254 mm / 0.01 in	200.0	0.254	B=0	0	21.237	10.493	6.968	418.1	0.290	Corrected MeteoRain 200 imperial row.
21	TBR (B=0) Onset HOBO RG3 15.39 cm 0.01 in	186.0	0.254	B=0	0	21.316	10.513	6.976	418.6	0.291	Small 0.01 in row.
22	TBR (B=0) Texas/Campbell TR-525 / TE525 6 in 0.01 in	182.4	0.254	B=0	0	21.339	10.518	6.979	418.7	0.291	Small 6-inch 0.01 in row.
23	Siphon TBR (B=0) Geolux RG400 0.1 mm — Vs≈16 mL, 4 mL/tip	400.0	0.100	B=0 + siphon	16	33.979	16.911	11.256	675.4	0.469	Corrected siphon volume and 4 mL/tip bucket volume.
24	Siphon TBR (B=0) KISTERS / HyQuest TB4 Series II 0.2 mm — Vs=12 mL	314.2	0.200	B=0 + siphon	12	35.459	17.666	11.763	705.8	0.490	Documented siphon capacity.
25	Siphon TBR (B=0) Geolux RG200 0.2 mm — Vs≈8 mL, 4 mL/tip	200.0	0.200	B=0 + siphon	8	37.145	18.414	12.241	734.5	0.510	Corrected siphon volume and 4 mL/tip bucket volume.
26	Siphon TBR (B=0) KISTERS / HyQuest TB4 Series II 0.01 in — Vs=12 mL	314.2	0.254	B=0 + siphon	12	37.654	18.777	12.507	750.4	0.521	Documented siphon capacity.
27	TBR (B=0) Casella TBRG 400 cm² 0.5 mm	400.0	0.500	B=0	0	40.887	20.428	13.615	816.9	0.567	Coarse 0.5 mm row.
28	TBR (B=0) MicroStep/Meteoservis MR2 200 cm² 0.5 mm option	200.0	0.500	B=0	0	41.076	20.475	13.636	818.2	0.568	Coarse 0.5 mm row.

Interpretation: the same threshold depth applies across aggregation periods. For example, if a product has Pmin 3% = 3.333 mm, that is 200 mm/h for a 1-minute product, 20 mm/h for a 10-minute product, 3.333 mm/h for a 1-hour product, and 0.139 mm/h for a 1-day product.

Worked calculations

The following examples show exactly where the numbers come from.

Universal conversions

1 mm of rain on 1 cm² = 0.1 mL

Tip volume:
V_tip_mL = A_cm² × d_mm / 10

Equivalent siphon rainfall depth:
s_mm = 10 × V_s_mL / A_cm²

Equivalent mean intensity:
I_mm_per_h = Pmin_mm / T_h
For 1 minute, T = 1/60 h, so I = 60 × Pmin.

Example 1 — MeteoRain 400 Aero, raw pulse product (B=0)

Inputs: A = 400 cm², d = 0.1 mm, no siphon, B = 0.

sigma_tip = 0.05 / (0.1 × 400)
          = 0.00125 mm

Delta_B = d / 2^B
        = 0.1 / 1
        = 0.1 mm

sigma_B = Delta_B / sqrt(6)
        = 0.1 / sqrt(6)
        = 0.04082 mm

Solving U95 / P ≤ 0.03 with U95 = 2 × sigma_total gives:

Pmin 3% = 2.757 mm

For a 1-minute product, that is 2.757 × 60 = 165.4 mm/h. This is why MeteoRain 400 Aero (B=0) is near the top of the raw TBR table.

Example 2 — Geolux RG400 with corrected siphon storage

Inputs: A = 400 cm², d = 0.1 mm, Vs ≈ 16 mL, B = 0.

Bucket volume:
V_tip = 400 × 0.1 / 10
      = 4.0 mL/tip

Siphon rainfall-depth equivalent:
s = 10 × 16 / 400
  = 0.400 mm

sigma_B_siphon = sqrt(0.1² + 0.400²) / sqrt(6)
               = 0.1683 mm

Solving the same 3% condition gives:

Pmin 3% = 11.256 mm

For a 1-minute product, that is 11.256 × 60 = 675.4 mm/h. This does not mean the gauge is poor; it means the unresolved siphon state is a large short-window uncertainty term.

Example 3 — OTT Pluvio² S Accu NRT

The Pluvio² S amount product is not a tipping bucket, so the bucket-phase equation is not used. The public amount accuracy floor is treated as 0.1 mm or 1%.

For 1% accumulation:
Pmin = 0.1 / 0.01 = 10.000 mm

For 2% accumulation:
Pmin = 0.1 / 0.02 = 5.000 mm

For 3% amount-derived mean intensity:
Pmin = 0.1 / 0.03 = 3.333 mm

The important part is product validity: Accu NRT is the delayed accumulation product, while Intensity RT is a real-time intensity product and should not be used as a daily/monthly accumulation source.

Example 4 — Lambrecht rain[e] as a high-B boundary-state product

The rain[e] serial amount resolution is 0.001 mm. If we isolate only that state resolution, the boundary step is far smaller than a normal TBR tip.

delta = 0.001 mm

sigma_B = delta / sqrt(6)
        = 0.000408 mm

Resolution-only 3% threshold:
Pmin ≈ 2 × sigma_B / 0.03
     ≈ 0.027 mm

That looks outstanding, but it is not the final public ranking because the public amount accuracy floor is still 0.1 mm or 1% for most rain[e] products. Also, for real-time 1-minute intensity, a high-resolution weight state must be current enough to represent the rain that just happened.

Weighing and hybrid gauge ranking

Weighing gauges need a different ranking logic from raw tipping buckets. A TBR uncertainty is mostly bucket phase plus tip repeatability. A weighing-gauge uncertainty is amount floor, intensity floor, filtering, delay, and product validity.

Best simple rule: treat public amount/total products, real-time intensity products, and delayed NRT accumulation products as different rows.

Rank class	Product	Product type	Public fixed-floor basis	Pmin 1% accum.	Pmin 2% accum.	Pmin 3% amount/intensity	How to interpret
1T	OTT Pluvio² S Accu NRT / Accu total NRT	Weighing, delayed NRT accumulation	±0.1 mm or ±1%	10.000	5.000	3.333	Best documented delayed accumulation product; 5-minute NRT separation in project model.
1T	Lambrecht rain[e] / rain[e]LP / rain[e]H3	Hybrid self-emptying weighing/TBR	±0.1 mm or ±1% below 6 mm/min; ±2% at high rate for LP/H3 series rows	10.000*	5.000	3.333	Static amount floor ties OTT; real-time 1-minute use needs dynamic high-B check.
1T	Lambrecht rain[e]314	Hybrid self-emptying weighing/TBR	±0.1 mm or ±1% below 3.82 mm/min; ±2% above	10.000*	5.000	3.333	Same floor; lower high-rate breakpoint than 200 cm² rain[e]/H3.
1T	Lambrecht rain[e]400	Hybrid self-emptying weighing/TBR	±0.1 mm or ±1% below 3 mm/min; ±2% above	10.000*	5.000	3.333	Same floor; 400 cm² area and 600 mm/h range.
Lower for 1% envelope	Lambrecht rain[e]one Modbus	Hybrid self-emptying weighing/TBR	0.1 mm or 2%	Not supported by public 1% envelope	5.000	3.333	Still acceptable for 2% and 3% thresholds by fixed floor, but not a 1% public-envelope row.
RT intensity floor tie	OTT Pluvio² S Intensity RT and Lambrecht native 1-min intensity channels	Native intensity products	±0.1 mm/min or ±6 mm/h where stated	—	—	200 mm/h native 1-min threshold	Do not mix RT intensity with delayed Accu NRT or public amount-total rows.

What this means for Lambrecht rain[e]

The Lambrecht rain[e] family should not be dismissed as “just a weighing gauge,” and it should not be over-ranked as if 0.001 mm resolution automatically guarantees 0.001 mm real-time intensity accuracy. It is a hybrid case:

• The public pages describe weighing with automatic self-emptying and 0.001 mm amount resolution.
• Some variants also provide pulse outputs, commonly 0.01 mm.
• For delayed or settled amount products, the high-resolution serial amount state is very favorable.
• For 1-minute real-time intensity, the value of that high-resolution state depends on how much the load-cell signal and processing have aged.

Product group	A cm²	Serial amount δ	Pulse-output δ	B_eff vs 0.01 mm	B_eff vs 0.1 mm	B_eff vs 0.2 mm	Interpretation
rain[e] / rain[e]LP / rain[e]H3	200	0.001 mm	0.01 mm where applicable	3.32	6.64	7.64	Public amount envelope dominates static ranking; serial state can be high-B if timely.
rain[e]314	314	0.001 mm	0.01 mm	3.32	6.64	7.64	Same high-B state; high-rate public breakpoint is 3.82 mm/min.
rain[e]400	400	0.001 mm	0.01 mm	3.32	6.64	7.64	Same high-B state; public high-rate breakpoint is 3 mm/min.
rain[e]one Modbus	200	0.001 mm	0.01 mm	3.32	6.64	7.64	Public amount envelope is 0.1 mm or 2%, so do not claim 1% from public accuracy.

Why the public envelope still matters: a 0.001 mm display or serial increment is not the same as a 0.001 mm public accuracy claim. The high-B model isolates the state-resolution floor. The public amount envelope, such as 0.1 mm or 1%, is the safer public-facing ranking unless the time constant and filtering of the specific output are measured.

The dynamic high-B condition for Lambrecht or other weighing gauges

The high-B idea is useful only when the fractional state is timely. If the weight estimate is delayed or filtered, the gauge may know the water level very precisely—but the water level may represent a slightly older rainfall state.

Let τ + L be the effective age of the weighing estimate in seconds. Here, τ is the filtering/time-constant effect and L is output delay or dead time. At rainfall intensity I, the approximate depth-age error is:

Delta_lag ≈ I × (τ + L) / 3600

Delta_dyn ≈ sqrt(delta² + Delta_lag²)

B_eff(I) = log2(d_ref / Delta_dyn)

The high-B state is better than B=2 only when:

Delta_dyn < d_ref / 4

Approximately:
I < 3600 × (d_ref / 4) / (τ + L)

Maximum intensity for the high-B state to remain better than B=2

Values are in mm/h. Example: with a 0.1 mm reference bucket and an effective age of 10 seconds, the high-resolution weighing state is better than B=2 only below about 9 mm/h.

τ+L (s)	d_ref=0.01 mm	d_ref=0.1 mm	d_ref=0.2 mm
1	9.00	90.00	180.00
2	4.50	45.00	90.00
5	1.80	18.00	36.00
10	0.90	9.00	18.00
30	0.30	3.00	6.00
60	0.15	1.50	3.00

Example effective B for d_ref = 0.1 mm

Negative or low B values mean the delayed/filtered state is no longer helpful for short-window real-time intensity, even though the static resolution is excellent.

τ+L (s)	I=5	I=10	I=25	I=50	I=100	I=200
1	5.87	5.08	3.83	2.84	1.85	0.85
2	5.08	4.15	2.84	1.85	0.85	-0.15
5	3.83	2.84	1.53	0.53	-0.47	-1.47
10	2.84	1.85	0.53	-0.47	-1.47	-2.47
30	1.26	0.26	-1.06	-2.06	-3.06	-4.06

Important practical consequence: the Lambrecht rain[e] high-B interpretation is strongest for amount, total, delayed, or settled products. It is conditional for real-time 1-minute intensity unless the output time constant and delay are known.

Practical interpretation by application

Instant / alarm intensity

Event-locked TBR timing is strong because each tip is tied to real water arrival. Weighing RT can be useful, but only if its delay, filtering, and threshold logic are acceptable for the alarm.

1-minute intensity

This is the hardest product. B=2 or high-quality fractional boundary state helps TBR-style products. Weighing or hybrid products need a measured time constant before they can be ranked above event-locked TBR intensity.

10-minute intensity

The longer window dilutes bucket-boundary error and reduces the penalty of weighing-gauge filtering. TBR and delayed accumulation products become more comparable.

Hourly to daily accumulation

Delayed weighing products such as OTT Accu NRT are strong point-reference totals. Low-power TBR networks remain attractive when spatial density matters more than one perfect point.

Assumptions used in the article

1. The corrected stochastic model uses U95 = 2 × sigma_total.
2. The standard-drop term uses 0.05 mL.
3. For ordinary TBR rows, sigma_B = Delta_B / sqrt(6).
4. For siphon TBR rows, sigma_B = sqrt(Delta_B² + s²) / sqrt(6), with s = 10 × Vs / A.
5. Geolux RG200 uses Vs ≈ 8 mL and Geolux RG400 uses Vs ≈ 16 mL. Both are modeled as 4 mL/tip gauges.
6. Seven Sensor Solutions 3S-RG is modeled as a non-siphon B=0 TBR.
7. MeteoRain 200 is only 0.2 mm or 0.254 mm / 0.01 in; MeteoRain 400 is only 0.1 mm; MeteoRain 533 is only 0.075 mm.
8. Vaisala QMR101 is listed as Vaisala QMR101 / Pronamic Professional. Vaisala QMR102 is listed as Vaisala QMR102 / EML ARG100 / Campbell ARG100.
9. Lambrecht rain[e] high-B rows use serial amount resolution δ = 0.001 mm. Pulse-output-only operation should use the pulse output resolution instead.
10. Public manufacturer envelopes and model-floor calculations are separate evidence layers.

Source basis

1. Project v17 correction note: corrected U95 = 2 × sigma_total, siphon model, and OTT RT/NRT split.
2. Project siphon reference notes: Geolux RG200 ≈ 8 mL, RG400 ≈ 16 mL, both 4 mL/tip; 3S-RG has no siphon.
3. Lambrecht rain[e] official product page.
4. Lambrecht rain[e]LP official product page.
5. Lambrecht rain[e]H3 official product page.
6. Lambrecht rain[e]314 official product page.
7. Lambrecht rain[e]400 official product page.
8. Lambrecht rain[e]one Modbus official product page.
9. OTT Pluvio² S official product page.
10. OTT Hydromet Pluvio² S product listing.
11. Seven Sensor Solutions 3S-RG datasheet.
12. Geolux RG tipping-bucket product page.