How to Minimize Dew Point Error: Temperature Accuracy vs. RH Accuracy

Humidity calculations and road-weather accuracy

How to minimize dew-point and frost-point error: which sensor specification matters more?

Question

When comparing temperature-and-humidity sensors, should I prioritize better air-temperature accuracy or better relative-humidity accuracy to obtain the most accurate calculated dew point or frost point?

Answer

Neither specification should be judged alone. A temperature-and-humidity probe calculates dew point or frost point from two measured inputs: air temperature and relative humidity. The best choice is therefore the sensor whose complete temperature-and-RH accuracy pair produces the lowest output error at the conditions that matter to the application.

For equal numerical input errors, temperature has the larger effect. At 0 °C and 90%RH, ±0.1 °C produces about 0.0988 °C of dew-point error, while ±0.1%RH produces about 0.0151 °C. Real RH specifications are usually several times larger than 0.1%RH, however, so RH can still become the largest part of a product’s complete dew-point or frost-point error.

Practical selection rule: convert both sensor accuracy limits into the same output unit—degrees Celsius of dew-point or frost-point error—at the required temperature and humidity. Then check whether hysteresis, repeatability, drift, and other terms are included or listed separately before ranking products.

Below freezing, this article reports frost point because frost and ice deposition are governed by saturation relative to ice. At and above 0 °C, it reports dew point, which is referenced to liquid water.

Why both temperature and RH matter

A combined probe does not directly sense dew point. It measures air temperature T and relative humidity RH, uses them to calculate the amount of water vapor present, and then finds the temperature at which that vapor would become saturated.

Water-vapor pressure: e = (RH / 100) × e_s,w(T) Dew point T_d: e_s,w(T_d) = e Frost point T_f: e_s,i(T_f) = e

Plain-language translation: the measured RH tells the calculation what fraction of the possible water vapor is present, while the measured temperature tells it how much vapor the air could hold. An error in either input moves the calculated dew point or frost point.

The calculations use the Murphy–Koop saturation-vapor-pressure equations and numerical inversion. The RH input is treated as relative to liquid water, which is the usual convention for meteorological RH sensors.

How the sensor limits are propagated

If a sensor is specified as ±u_T for temperature and ±u_RH for humidity, the calculation checks all four combinations: both high, both low, temperature high with RH low, and temperature low with RH high. The reported result is the largest change in dew point or frost point.

E = max |Point(T ± u_T, RH ± u_RH) − Point(T, RH)|

This is a conservative limit calculation, not a prediction of the most likely everyday error. It asks: how far could the calculated point move if both input channels reached their stated limits in the least favorable combination?

How %RH accuracy is written: ±1%RH is an absolute accuracy limit. At a true value of 90%RH, it means the indicated value may be approximately 89%RH to 91%RH. It does not mean one percent of 90.

Which input is more influential per equal increment?

Local dew-point and frost-point sensitivity at 90 percent relative humidity.

Air condition	Output	Change per 1 °C temperature error	Change per 1%RH error	Error from ±0.1 °C	Error from ±0.1%RH
0 °C, 90%RH	Dew point	0.988 °C/°C	0.151 °C/%RH	0.0988 °C	0.0151 °C
−10 °C, 90%RH	Frost point	0.889 °C/°C	0.125 °C/%RH	0.0890 °C	0.0125 °C

6.54×At 0 °C, ±0.1 °C causes about 6.54 times as much dew-point error as ±0.1%RH.

7.11×At −10 °C, ±0.1 °C causes about 7.11 times as much frost-point error as ±0.1%RH.

Do not turn this into a product ranking. Figure 1 compares equal-size input increments. A real product might have ±0.1 °C temperature accuracy but ±2%RH or ±3%RH humidity accuracy. The product comparison must use each channel’s full specified limit.

Figure 1. What a small error in each sensor channel does to calculated dew point or frost point at 90%RH. The blue line is the maximum output error caused by ±0.1 °C of air-temperature error. The orange line is the output error caused by ±0.1%RH. Because the blue line remains much higher, temperature has the larger effect when the two numerical input errors are equal. The visible step at 0 °C is not a sudden change in sensor behavior: it is caused by changing the calculation reference from saturation over ice below 0 °C to saturation over liquid water at and above 0 °C. This figure explains sensitivity only; it does not compare complete product specifications.

What this means when selecting a sensor: start by recognizing that very good temperature accuracy is valuable, especially near freezing. Then convert the sensor’s full RH limit into dew-point or frost-point degrees as well. A much larger RH limit can outweigh the lower per-unit RH sensitivity.

Contour maps: combine both specifications on one chart

The next maps answer the practical buying question: what dew-point or frost-point error results from this sensor’s temperature accuracy and RH accuracy together?

1. Choose the temperature accuracy on the horizontal axis. A probe specified at ±0.20 °C is located at 0.20.
2. Choose the RH accuracy on the vertical axis. A probe specified at ±2%RH is located at 2.
3. Find where the two values meet. The color at that intersection is the largest resulting dew-point or frost-point error.

Reading the direction of improvement: farther left means better temperature accuracy; farther down means better RH accuracy; greener colors mean lower resulting error. If two products land in the same color band, their calculated point accuracy is similar enough that response time, shielding, durability, calibration, power, communications, and price may become more important.

Every map uses the same axes and the same 0.1–1.4 °C green-to-red scale. Each color band represents 0.1 °C of calculated output error, so maps at different air temperatures can be compared visually.

The maps stop at ±4%RH. Any product limit above that range must be calculated numerically rather than read from the picture.

Keep hysteresis treatment consistent. The MeteoTemp / MeteoHelix IoT Pro data-sheet DPw/FPi values exclude their separately stated short-term hysteresis allowance. HMP155 includes hysteresis and repeatability in its RH accuracy. Figures 6 and 7 show the BARANI allowance separately so the difference is visible rather than hidden.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on dew-point error at +50 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined dew-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on dew-point error at +40 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined dew-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on dew-point error at +30 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined dew-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on dew-point error at +20 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined dew-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on dew-point error at +10 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined dew-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on dew-point error at 0 °C and 90% RH using the liquid-water convention. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined dew-point error: greener bands indicate lower error and redder bands indicate higher error.

Figure 2. Dew-point sensor-selection maps from 0 °C to +50 °C at 90%RH. Each panel represents one air temperature. Select a sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis; the color where the two values meet is the largest combined dew-point error. Moving left or down improves the result. Green bands indicate lower error, while yellow, orange, and red indicate progressively higher error. Because every panel uses the same color scale, the increasing red area at warmer temperatures shows that the same RH error creates more dew-point error as the air becomes warmer.

Why the warmer maps become redder

A fixed ±2%RH limit does not produce the same dew-point error at every temperature. Warm air can contain much more water vapor, so the dew-point calculation becomes more sensitive to the same RH uncertainty. The redder warm maps therefore describe a property of the calculation—not proof that the physical sensor becomes less accurate in warm air.

Example: a sensor with ±0.2 °C temperature accuracy and ±2%RH humidity accuracy is read at x = 0.2 and y = 2.0 on every panel. Its color is greener at 0 °C and redder at +50 °C because the same RH error produces a larger dew-point change at the warmer condition.

Below freezing, use frost point

Dew point is referenced to liquid water. Below freezing, frost deposition and road icing are better evaluated with frost point, which is referenced to ice. Figure 3 therefore uses the same sensor-accuracy axes but changes the output calculation to frost point.

A separate 0 °C frost-point panel is included to show the ice-side convention at the phase boundary. The −7 °C panel is included because approximately −7 °C is a practical ordinary-NaCl road-treatment landmark; it is not the freezing point of concentrated brine.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on frost-point error at 0 °C and 90% RH using the ice-side convention. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined frost-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Practical NaCl road-salt operating landmark. Combined effect of temperature and relative-humidity sensor accuracy on frost-point error at −7 °C and 90%RH. Select the sensor’s temperature accuracy on the horizontal axis and its absolute RH-accuracy limit in ±%RH on the vertical axis. The color where the two values meet shows the resulting combined frost-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on frost-point error at −10 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined frost-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on frost-point error at −20 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined frost-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on frost-point error at −30 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined frost-point error: greener bands indicate lower error and redder bands indicate higher error.

View fullsize

Combined effect of temperature and relative-humidity sensor accuracy on frost-point error at −40 °C and 90% RH. Select the sensor’s temperature accuracy on the horizontal axis and RH accuracy on the vertical axis. The color where the two values meet shows the resulting combined frost-point error: greener bands indicate lower error and redder bands indicate higher error.

Figure 3. Frost-point sensor-selection maps from 0 °C to −40 °C at 90%RH. Use these panels exactly like the dew-point maps: choose temperature accuracy on the horizontal axis, choose RH accuracy on the vertical axis, and read the largest combined frost-point error from the color at their intersection. Left, down, and green are better. The common scale makes it possible to compare cold conditions directly. The −7 °C panel is marked as a practical NaCl road-treatment reference, not as the freezing point of concentrated road brine.

How to minimize dew-point and frost-point error in practice

1. Compare the complete temperature-and-RH pair. A probe with excellent temperature accuracy can still have a larger final point error if its RH limit is broad, and the reverse can also be true.
2. Use specifications that apply at the actual temperature and RH. A value stated only at +20 or +25 °C should not automatically be carried into a subzero road-weather calculation.
3. Check what the number includes. “Accuracy” may or may not include nonlinearity, repeatability, hysteresis, drift, calibration uncertainty, or field exposure. Product numbers are comparable only when their scopes are understood.
4. Use dew point at and above freezing and frost point below freezing. The reference phase matters near 0 °C.
5. Make temperature and RH represent the same parcel of air. Different response times, poor airflow, or electronics heating can pair a new temperature value with an older RH value and create a false point calculation.
6. Keep the sensing surface dry and clean. Fog droplets, condensation, road spray, salt, dust, and a clogged filter can create errors much larger than the laboratory sensor limit.
7. Calibrate where the decision matters. A road-weather probe should be checked at high RH near and below freezing, not only at comfortable indoor laboratory conditions.
8. Do not confuse air frost point with pavement condition. Road icing also depends on pavement temperature, residual salt concentration, precipitation, radiation, drainage, and traffic.

The best-value sensor is the one whose complete and correctly interpreted specification meets the required point accuracy in the real installation—not automatically the one with the smallest single headline number.

Case study: MeteoTemp / MeteoHelix IoT Pro vs HMP155 and HMP45A

The BARANI MeteoTemp probe and the wireless MeteoHelix IoT Pro micro-weather station publish the same temperature, RH, dew-point, and frost-point accuracy over the MeteoHelix operating range. The article therefore treats them as one accuracy class and calls them MeteoTemp / MeteoHelix IoT Pro. MeteoTemp extends temperature operation to +105 °C; MeteoHelix IoT Pro operates to +65 °C.

Two conditions are used for the comparison:

• −7 °C and 90%RH: a practical cold road-treatment condition; the calculated output is frost point.
• +25 °C and 90%RH: a warm high-humidity reference condition; the calculated output is dew point.

How the BARANI RH specification is built

Base accuracy from 0 to 80%RH:
−45 to 0 °C: ±3.95 → ±2.05%RH
0 to +25 °C: ±2.05 → ±1.30%RH
+25 to +50 °C: ±1.30 → ±2.05%RH

High-RH allowance from 80 to 100%RH: add 0.025%RH for each 1%RH above 80%RH, up to +0.50%RH at 100%RH. At 90%RH, the addition is +0.25%RH.

Short-term hysteresis: ≤±1.0%RH additional.

What “hysteresis” means: at the same true RH, a polymer sensor can read slightly differently depending on whether humidity approached that point from a wetter condition or a drier condition. BARANI lists this possible direction-dependent difference separately instead of hiding it inside the base accuracy.

The base RH accuracy includes nonlinearity, repeatability, and temperature dependence. Short-term hysteresis, long-term drift, and recovery after condensation or frosting are additional. Accuracy applies only after thermal and humidity equilibrium, with no liquid water, frost, or ice on the sensing surface.

Data-sheet abbreviations: DPw means dew point calculated relative to liquid water; FPi means frost point calculated relative to ice.

BARANI values used at 90%RH

Condition	Base RH limit	Addition because RH is 90%	RH limit used in DPw/FPi calculation	Temperature limit	Published point bound
−7 °C, 90%RH — frost point	±2.346%RH	+0.25%RH	±2.596%RH	±0.116 °C	±0.44 °C FPi
+25 °C, 90%RH — dew point	±1.30%RH	+0.25%RH	±1.55%RH	±0.100 °C	±0.39 °C DPw

The listed BARANI DPw/FPi bounds include the +0.25%RH high-RH allowance applicable at 90%RH but exclude the separate short-term hysteresis term. Figures 6 and 7 show that separate term as a conservative additional screen.

Temperature accuracy used in the case study

Sensor	At −7 °C	At +25 °C	Where the value comes from
MeteoTemp / MeteoHelix IoT Pro	±0.116 °C	±0.100 °C	Published BARANI linear temperature-accuracy rule.
Vaisala HMP155 RS-485	±0.196 °C	±0.133 °C	Published digital-output equations.
Vaisala HMP45A	±0.335 °C	±0.225 °C	Read by piecewise-linear interpolation of the published graph.

MeteoTemp / MeteoHelix IoT Pro has the tightest temperature specification at both conditions. Figures 4 and 5 show that advantage by itself before RH is added.

Figure 4. Temperature accuracy only: MeteoTemp / MeteoHelix IoT Pro compared with HMP45A. The shaded band shows the maximum positive and negative temperature error allowed by each specification at every air temperature. A narrower band means the reported air temperature is constrained more tightly around the true value. At −7 °C, the BARANI limit is approximately ±0.116 °C versus ±0.335 °C for HMP45A. At +25 °C, the values are ±0.100 °C and ±0.225 °C. This figure compares only the temperature channel; it does not yet include RH or calculate dew/frost point.

Published BARANI temperature rule: ±0.10 °C from 0 to +65 °C. Below 0 °C, the limit increases linearly to ±0.20 °C at −45 °C. The data sheet also states ±0.20 °C maximum over the full MeteoTemp range.

MeteoTemp / MeteoHelix IoT Pro versus HMP155 temperature accuracy

Figure 5 repeats the temperature-only comparison against HMP155. The solid HMP155 band is the RS-485 digital-output specification used in the case study. The dashed lines show the wider voltage-output limits, which should not be mixed into the digital comparison.

At −7 °C: BARANI ±0.116 °C; HMP155 RS-485 ±0.196 °C.

At +25 °C: BARANI ±0.100 °C; HMP155 RS-485 ±0.133 °C.

Temperature accuracy alone favors MeteoTemp / MeteoHelix IoT Pro at both selected conditions. The next section adds the RH channel and explains why the treatment of hysteresis can change the apparent ranking.

Figure 5. Temperature accuracy only: MeteoTemp / MeteoHelix IoT Pro compared with HMP155. The narrower BARANI band means a tighter temperature specification across the shared operating range. The solid HMP155 band is its RS-485 digital-output accuracy; the dashed limits apply to the voltage output. At −7 °C, the digital limits are ±0.116 °C for BARANI and ±0.196 °C for HMP155. At +25 °C, they are ±0.100 °C and ±0.133 °C. RH is not included in this figure.

Now add the RH channel

The three products do not present RH accuracy in exactly the same way. That is why simply copying the smallest headline number can be misleading.

Nonlinearity means the response curve is not perfectly straight across the RH range.

Repeatability means repeated measurements at the same condition are not perfectly identical.

Hysteresis means the reading can depend slightly on whether humidity is rising or falling.

MeteoTemp / MeteoHelix IoT Pro at 90%RH

At 90%RH, the reading is 10%RH above the 80%RH threshold. The high-RH rule therefore adds 10 × 0.025%RH = +0.25%RH to the temperature-dependent base limit.

−7 °C: 2.346 + 0.25 = ±2.596%RH +25 °C: 1.30 + 0.25 = ±1.55%RH

These are the RH limits used to derive the published BARANI FPi and DPw bounds. The separate ≤±1.0%RH short-term hysteresis term is not included in those listed point bounds.

HMP155 at the same conditions

At −7 °C, HMP155 uses its −20 to +40 °C equation: ±(1.0 + 0.008 × 90) = ±1.72%RH. At +25 °C and exactly 90%RH, the article conservatively uses the near-saturation room-temperature band of ±1.70%RH. Vaisala states that its published HMP155 RH accuracy includes nonlinearity, hysteresis, and repeatability.

HMP45A benchmark

The HMP45A manual gives ±1%RH at +20 °C and a temperature dependence of ±0.05%RH per °C. Applying those statements additively gives approximately ±2.35%RH at −7 °C and ±1.25%RH at +25 °C. These are useful modeled benchmarks, but they are not a directly published cold-range table.

Sensor	−7 °C, 90%RH	+25 °C, 90%RH	How hysteresis is handled	Evidence level
MeteoTemp / MeteoHelix IoT Pro	±2.596%RH	±1.55%RH	≤±1.0%RH short-term hysteresis is stated separately.	Published
HMP155	±1.72%RH	±1.70%RH	Included in the published accuracy.	Published
HMP45A	Approx. ±2.35%RH	Approx. ±1.25%RH	Not fully harmonized in the article-derived model.	Derived benchmark

Why two BARANI results are shown later: one result reproduces the DPw/FPi value listed in the data sheet. A second, more conservative screen adds the entire separate hysteresis allowance in the same direction as all other errors. That second value is useful for comparison, but it is not the published DPw/FPi bound and should not replace it.

Figure 6. RH accuracy inputs used in the −7 °C and +25 °C case studies. Lower solid bars mean a tighter RH specification. The solid BARANI bars are the values used to derive the listed DPw/FPi bounds. The hatched extensions show the separate maximum short-term hysteresis allowance; they are not part of those listed bounds. HMP155 already includes hysteresis and repeatability in its published RH accuracy. HMP45A remains an article-derived benchmark. This chart therefore shows both the numerical limits and the difference in specification scope.

Complete modeled dew-point and frost-point results

Condition	Sensor or interpretation	Temperature-only contribution	Combined point error	How to interpret it
−7 °C, 90%RH Frost point	BARANI data-sheet calculation	0.102 °C	0.438 °C published bound ±0.44 °C	Uses base RH accuracy plus the +0.25%RH high-RH allowance; separate hysteresis excluded.
	BARANI full-hysteresis screening case	0.102 °C	0.569 °C	Adds the full ≤±1.0%RH allowance conservatively.
	HMP155	0.173 °C	0.394 °C	Published temperature and RH inputs; hysteresis included in RH accuracy.
	HMP45A	0.297 °C	0.599 °C	Derived RH benchmark, not a directly published cold-range total.
+25 °C, 90%RH Dew point	BARANI data-sheet calculation	0.099 °C	0.386 °C published bound ±0.39 °C	Uses base RH accuracy plus the +0.25%RH high-RH allowance; separate hysteresis excluded.
	BARANI full-hysteresis screening case	0.099 °C	0.573 °C	Adds the full ≤±1.0%RH allowance conservatively.
	HMP155	0.131 °C	0.446 °C	Published temperature and RH inputs.
	HMP45A	0.222 °C	0.453 °C	Derived RH benchmark.

Why the BARANI data-sheet result and the full-hysteresis screen are both shown

The lower BARANI value reproduces the DPw/FPi row in the final data sheets. The higher value is a deliberately conservative what-if calculation in which the full separate hysteresis allowance acts in the same direction as all other limits. In real operation, the full hysteresis allowance will not necessarily coincide with every other worst-case contribution.

Use the two values for different questions: use the data-sheet value when checking the published DPw/FPi specification; use the full-hysteresis screen when asking how the comparison changes under a more conservative like-for-like treatment against an all-in specification.

How to read the direct comparison maps

1. The horizontal position is the temperature accuracy. Farther left is tighter.
2. The vertical position is the RH accuracy. Farther down is tighter.
3. The background color is the resulting combined dew-point or frost-point error. Greener is lower.
4. The BARANI diamond is the data-sheet calculation input. The triangle above it is the full-hysteresis screening input.
5. Dotted guides help the reader project each point to the two axes.

View fullsize

View fullsize

Figure 7A. Direct frost-point comparison at −7 °C and 90%RH. Each point combines one temperature limit and one RH limit. Farther left, farther down, and greener mean lower resulting frost-point error. HMP155 produces 0.394 °C. MeteoTemp / MeteoHelix IoT Pro produces 0.438 °C using the RH accuracy applied in its published ±0.44 °C FPi bound. HMP45A produces 0.599 °C using the derived RH benchmark. The BARANI triangle shows the more conservative 0.569 °C result obtained after adding the entire separate short-term hysteresis allowance.

Figure 7B. Direct dew-point comparison at +25 °C and 90%RH. Read the map in the same way: left and down are tighter sensor inputs, and greener is lower dew-point error. MeteoTemp / MeteoHelix IoT Pro produces 0.386 °C using the RH accuracy applied in its published ±0.39 °C DPw bound. HMP155 produces 0.446 °C and the modeled HMP45A benchmark produces 0.453 °C. Adding the full separate BARANI hysteresis allowance gives the higher 0.573 °C screening result.

What the case study actually shows

• Temperature channel: MeteoTemp / MeteoHelix IoT Pro has the tightest temperature specification at both selected conditions.
• −7 °C data-sheet comparison: HMP155 has the lowest modeled frost-point error. The BARANI published ±0.44 °C FPi bound is lower than the modeled HMP45A benchmark.
• +25 °C data-sheet comparison: the BARANI published ±0.39 °C DPw bound is lower than the modeled HMP155 and HMP45A results.
• Conservative hysteresis screen: adding the full separate BARANI hysteresis allowance raises the results to 0.569 °C at −7 °C and 0.573 °C at +25 °C. This demonstrates why specification scope must be visible.
• HMP45A evidence limitation: its RH values are derived from the manual rather than taken from a directly published cold-range equation, so the modeled differences should not be described as independent laboratory proof.

A technically fair comparison does not hide how each accuracy number is constructed. It separates equilibrium accuracy, high-RH allowances, hysteresis, drift, and field effects before drawing a conclusion.

Road-weather limitations beyond the sensing element

Actual roadside uncertainty also depends on radiation-shield performance, solar loading, self-heating, airflow, filter condition, response mismatch, salt contamination, wetting, calibration drift, installation height, and spatial differences between air and pavement.

Temperature and RH must describe the same air at approximately the same time. A fast temperature channel paired with a slower RH channel can create a transient dew-point or frost-point error even when both individual sensors eventually settle within specification.

The BARANI accuracy applies after thermal and humidity equilibrium with no liquid water, frost, or ice on the sensing surface. Long-term drift and recovery after condensation or frosting are additional.

A calculated dew point or frost point is not a substitute for pavement temperature. Road icing also depends on precipitation, residual salt, treatment history, traffic, wind, radiation, drainage, and the pavement energy balance.

Bottom line

Sensor-selection conclusion

Choose a temperature-and-humidity sensor by comparing the complete resulting dew-point or frost-point error at the required condition—and compare like with like. A number that excludes hysteresis should not be ranked as though it includes the same components as an all-in competitor specification.

• Use dew point at and above freezing and frost point below freezing.
• Treat ±%RH as an absolute accuracy limit.
• Apply the high-RH allowance at the actual RH value.
• Identify whether hysteresis, repeatability, and drift are included or additional.
• Use data-sheet output bounds for specification compliance and separate conservative screens for sensitivity analysis.
• Include shielding, response, contamination, calibration, and installation in the field uncertainty budget.

Calculation summary: the MeteoTemp / MeteoHelix IoT Pro DPw/FPi bounds use the temperature-dependent 0–80%RH limit plus the +0.25%RH allowance applicable at 90%RH. They exclude the separate ≤±1.0%RH short-term hysteresis term. Long-term drift and recovery after condensation or frosting are additional.

Appendix A: case-study calculation traceability

This appendix shows the exact numbers entered into the calculation so that an instrument user can reproduce the two case-study results. It is a specification comparison, not an independent calibration certificate or a complete field uncertainty budget.

Plain-language calculation sequence

1. Start with the nominal air temperature and 90%RH.
2. Calculate the nominal dew point or frost point.
3. Move temperature to both ends of its accuracy limit.
4. Move RH to both ends of its accuracy limit.
5. Calculate all four combinations and retain the largest output change.

E_max = max |Point(T + δT, RH + δRH) − Point(T, RH)| δT ∈ {−u_T, +u_T} δRH ∈ {−u_RH, +u_RH}

Check of the four BARANI data-sheet DPw/FPi bounds

Published condition	Temperature limit	RH limit used at 90%RH	Exact calculated result	Conservative data-sheet bound
+25 °C dew point (DPw)	±0.100 °C	±1.550%RH	0.385696 °C	±0.39 °C
0 °C dew point (DPw)	±0.100 °C	±2.300%RH	0.450014 °C	±0.46 °C
0 °C frost point (FPi)	±0.100 °C	±2.300%RH	0.398280 °C	±0.40 °C
−7 °C frost point (FPi)	±0.116 °C	±2.596%RH	0.437702 °C	±0.44 °C

Case-study audit table

Condition	Sensor or interpretation	Temperature limit	RH limit	Combined point error	Evidence meaning
−7 °C, 90%RH Frost point	BARANI data-sheet calculation	±0.116 °C	±2.596%RH	0.437702 °C	Published accuracy; separate hysteresis excluded.
	BARANI full-hysteresis screen	±0.116 °C	≤±3.596%RH	0.569369 °C	Full separate hysteresis added conservatively.
	HMP155	±0.196 °C	±1.72%RH	0.394274 °C	Published equations.
	HMP45A	±0.335 °C	±2.35%RH	0.599392 °C	Derived RH benchmark.
+25 °C, 90%RH Dew point	BARANI data-sheet calculation	±0.100 °C	±1.55%RH	0.385696 °C	Published accuracy; separate hysteresis excluded.
	BARANI full-hysteresis screen	±0.100 °C	≤±2.55%RH	0.573215 °C	Full separate hysteresis added conservatively.
	HMP155	±0.133 °C	±1.70%RH	0.445695 °C	Published equations.
	HMP45A	±0.225 °C	±1.25%RH	0.452917 °C	Derived RH benchmark.

Appendix B: full-range accuracy lookup table

This table lets a reader look up the temperature and RH limits used in the article at each temperature. The BARANI columns separate the base 0–80%RH limit, the continuous high-RH allowance at 90%RH and 100%RH, and the additional short-term hysteresis ceiling.

How to read a row: first find the air temperature. Then read the product’s temperature limit and the RH limit that applies to the measured humidity. For the article’s 90%RH calculations, use the “At 90%RH” BARANI column or evaluate the exact HMP155 formula at 90%RH. Do not automatically add the BARANI hysteresis column unless a conservative hysteresis screen is specifically required.

Air temperature	Vaisala HMP155 RS-485			BARANI MeteoTemp / MeteoHelix IoT Pro					Vaisala HMP45A
	T limit	Max 0–89.9%RH	Max 90–100%RH	T limit	0–80%RH base	At 90%RH	At 100%RH	Short-term hysteresis additional	T limit	Derived at 90%RH
-80 °C	±0.400 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
-70 °C	±0.372 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
-60 °C	±0.344 °C	±4.28%RH	±4.60%RH	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
-50 °C	±0.316 °C	±4.28%RH	±4.60%RH	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
-45 °C BARANI RH and temperature lower endpoint	±0.302 °C	±4.28%RH	±4.60%RH	±0.200 °C	±3.95%RH	±4.20%RH	±4.45%RH	±1.00%RH	Not specified / outside range	Not specified / outside range
-40 °C	±0.288 °C	±4.28%RH	±4.60%RH	±0.189 °C	±3.74%RH	±3.99%RH	±4.24%RH	±1.00%RH	±0.500 °C	±4.00%RH
-30 °C	±0.260 °C	±2.28%RH	±2.40%RH	±0.167 °C	±3.32%RH	±3.57%RH	±3.82%RH	±1.00%RH	±0.450 °C	±3.50%RH
-20 °C	±0.232 °C	±2.28%RH	±2.40%RH	±0.144 °C	±2.89%RH	±3.14%RH	±3.39%RH	±1.00%RH	±0.400 °C	±3.00%RH
-10 °C	±0.204 °C	±1.72%RH	±1.80%RH	±0.122 °C	±2.47%RH	±2.72%RH	±2.97%RH	±1.00%RH	±0.350 °C	±2.50%RH
-7 °C Road-weather case	±0.196 °C	±1.72%RH	±1.80%RH	±0.116 °C	±2.35%RH	±2.60%RH	±2.85%RH	±1.00%RH	±0.335 °C	±2.35%RH
+0 °C	±0.176 °C	±1.72%RH	±1.80%RH	±0.100 °C	±2.05%RH	±2.30%RH	±2.55%RH	±1.00%RH	±0.300 °C	±2.00%RH
+10 °C	±0.148 °C	±1.72%RH	±1.80%RH	±0.100 °C	±1.75%RH	±2.00%RH	±2.25%RH	±1.00%RH	±0.250 °C	±1.50%RH
+15 °C HMP155 room-temperature band begins	±0.134 °C	±1.00%RH	±1.70%RH	±0.100 °C	±1.60%RH	±1.85%RH	±2.10%RH	±1.00%RH	±0.225 °C	±1.25%RH
+20 °C	±0.120 °C	±1.00%RH	±1.70%RH	±0.100 °C	±1.45%RH	±1.70%RH	±1.95%RH	±1.00%RH	±0.200 °C	±1.00%RH
+25 °C Warm comparison	±0.133 °C	±1.00%RH	±1.70%RH	±0.100 °C	±1.30%RH	±1.55%RH	±1.80%RH	±1.00%RH	±0.225 °C	±1.25%RH
+30 °C	±0.145 °C	±1.72%RH	±1.80%RH	±0.100 °C	±1.45%RH	±1.70%RH	±1.95%RH	±1.00%RH	±0.250 °C	±1.50%RH
+40 °C	±0.170 °C	±2.28%RH	±2.40%RH	±0.100 °C	±1.75%RH	±2.00%RH	±2.25%RH	±1.00%RH	±0.300 °C	±2.00%RH
+50 °C BARANI RH upper endpoint	±0.195 °C	±2.28%RH	±2.40%RH	±0.100 °C	±2.05%RH	±2.30%RH	±2.55%RH	±1.00%RH	±0.350 °C	±2.50%RH
+60 °C	±0.220 °C	±2.28%RH	±2.40%RH	±0.100 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.400 °C	±3.00%RH
+65 °C MeteoHelix range endpoint; MeteoTemp continues	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.100 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
+70 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.200 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
+80 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.200 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
+90 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.200 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
+100 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.200 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range
+105 °C MeteoTemp full-range endpoint	Not specified / outside range	Not specified / outside range	Not specified / outside range	±0.200 °C	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range	Not specified / outside range

Important interpretation notes

1. The BARANI 90%RH and 100%RH columns include the continuous high-RH allowance but not the separate short-term hysteresis ceiling.
2. MeteoHelix IoT Pro operates from −45 to +65 °C. MeteoTemp extends to +105 °C; above +65 °C the table shows the MeteoTemp maximum temperature limit only.
3. HMP155 columns are maxima for broad RH intervals. The exact value at a particular RH, such as 90%, can be smaller and should be calculated from the applicable equation.
4. HMP45A RH values are article-derived benchmarks, not a manufacturer-published full-range lookup table.
5. Long-term drift and recovery after condensation or frosting are additional to the BARANI limits.

---

References

1. Murphy, D. M. and Koop, T. (2005), Review of the vapour pressures of ice and supercooled water for atmospheric applications. DOI 10.1256/qj.04.94.
2. BARANI DESIGN Technologies, MeteoTemp RH+T + Pressure Datasheet, release date 2026 July 22. Open product and data-sheet page.
3. BARANI DESIGN Technologies, MeteoHelix IoT Pro Datasheet, release date 2026 July 22. Open product and data-sheet page.
4. Vaisala, HMP45A and HMP45D Operating Manual, U274EN-1.2. Open manual.
5. Vaisala, HMP155 User Guide — Technical Data. Open specifications.
6. Vaisala, Dew-point and frost-point temperature selection. Open technical note.
7. Devon County Council, Winter Service and Emergency Plan. Open winter-service guidance.

Final calculation note: the MeteoTemp / MeteoHelix IoT Pro DPw/FPi values use the 0–80%RH temperature-dependent accuracy plus the high-RH allowance applicable at 90%RH. They exclude the separate ≤±1.0%RH short-term hysteresis term. Long-term drift and recovery after condensation or frosting are additional.