Accuracy Matters in Radiosonde Measurements

Home , Barometer, Capping inversion, Radiosonde

/ WHITE PAPER

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 1 11.3.2016 10.37 Table of Contents

Introduction ...... 3

Quality of Radiosonde Measurements is Critical ...... 5

Convective Weather and Thunderstorms ...... 6

Radiosondes in Forecasting Convection ...... 6

Case Study 1: Convective Weather ...... 8

Case Study 2: Weak Convection ...... 10

Winter Weather ...... 12

Radiosondes in Forecasting Winter Precipitation ...... 13

Case Study 3: Freezing Rain and Ice Storm ...... 15

Numerical Weather Prediction ...... 17

Radiosondes in Validating NWP Models ...... 18

Case Study 4: Fog ...... 18

Meteorological Indices ...... 20

Impact of Radiosonde Data Quality ...... 21

Radiosonde Pressure Measurement ...... 23

GPS-based Pressure ...... 24

Sensor-based Pressure ...... 25

Case Study 5: Tropical Cyclone ...... 26

Radiosonde Data in Climatology ...... 27

Radiosonde Accuracy Matters ...... 28

Further reading ...... 29

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 2 11.3.2016 10.37 Introduction

Radiosondes measure critical atmospheric variables with accuracy and precision that cannot be obtained with other meteorological observations. Radiosondes are unique instruments as they provide continuous, detailed profiles from the ground to altitudes of 30 km and above. The quality and reliability of the measurements are essential. Even small inaccuracies in the profiles can prevent the forecaster from observing critical details and making correct conclusions.

The chapters in this document provide information on the importance of radiosonde measurement accuracy, and how accuracy helps to successfully forecast high impact weather. Graphs, tables, and case study summaries illustrate how correct or incorrect observations of temperature inversion layers, ice forming layers, humidity, and other atmospheric properties can change a forecast in various weather situations.

a) Deep convection d) Freezing rain

Sounding Modified sounding: profile wet-bulb error

Ice formation Shallow layer T < -10 °C ➔ Probable ice formation

Elevated warm Tmax = 1.9 °C ➔ Partial melting of layer ice ➔ Solid and liquid can occur

Surface Tsurface < 0 °C ➔ Rain will freeze on the ground ➔ Ice accumulation / sleet

Forecast Ice pellets (more probable) or freezing rain or mix

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 3 11.3.2016 10.37 4

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 4 11.3.2016 10.37 Quality of Radiosonde Measurements is Critical

The radiosonde is unique among important that radiosonde sensors or even entirely mask the feature other observation systems in work reliably in changing conditions in the profile. In summertime that it provides a complete throughout the harsh environment this may lead the meteorologist vertical profiling of the initial of the upper atmosphere. Erroneous to conclude that, for example, state of the atmospheric analysis measurements could entirely convection will start earlier in that drives numerical weather change the forecast conclusions. For the day with less energy released prediction models. Furthermore, example, the following challenging into convection, or to predict that meteorologists are interested conditions could occur in cloudy rain will start earlier and cool the in several phenomena that are weather situations: land surface, with thunderstorms visible in the radiosonde data, unlikely to form. including cloud layers, dry layers, • As the radiosonde emerges from temperature inversions, jet streams a cloud, the temperature sensor • If the humidity sensor freezes and wind shear. Radiosondes also is in danger of experiencing a while passing through clouds, have an important role in providing ‘wet bulb’ error, generated as the radiosonde may incorrectly long-term high-quality time series water or ice evaporates from the continue measuring high relative of climatology trends of various surface of the sensor. This may humidity. The sensitivity to detect parameters. lead to incorrectly measuring cloud layers is reduced. For the strength of the temperature example, erroneously predicted These applications place a high inversion. Such erroneous upper cloud layers would block demand on the accuracy and observations can reduce the the solar radiation from reaching precision of the measurement. It is magnitude of the detected layer the ground and thus inhibit convection.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 5 11.3.2016 10.37 Convective Weather and Thunderstorms

Figure 1. Model radiosonde profiles of temperature, dew-point and wind, showing characteristic features that predict various types of convection. See further explanations in Table 1.

During convection, warm air near Radiosondes in Forecasting Convection the ground starts rising until it cools and becomes balanced with In a convective situation, the Radiosonde measurements its surroundings. The following analysis of the latest radiosonde summarize the state of the factors are needed for atmospheric profiles in the region is an essential atmosphere and give the basis for convection to occur: enough step in forecasting. Convection understanding how the weather will moisture at the height from which is still poorly represented in the evolve in the next hours. Figure 1 convection initiates, instability, numerical weather prediction shows examples of three basic types and a lifting mechanism. Strong models due to inadequate spatial of convective weather and how they convection in humid atmosphere and temporal resolutions and the can be forecasted from radiosonde can lead to thunderstorms. difficulty to quantify humidity. profiles.

a) Deep convection The air near the surface is warm, moist and well-mixed. There is a strong temperature inversion and enough convective inhibition (CIN) to prevent convection from beginning too early and thus enables convective available potential energy (CAPE) to build up. A relatively rapid decrease in temperature with height in the middle troposphere results in small stability. Cloud layers affect the amount of solar warming on the surface. If air near the surface

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 6 11.3.2016 10.37 Factor Explanation Interpretation in radiosonde profile

Temperature inversion Layer of warmer air above cool A temperature inversion prevents air. ascending air from penetrating it and helps build up convective energy. A strong inversion may prevent convection entirely.

Dew-point temperature The temperature where Humid and dry layers in the profile condensation begins. Moisture indicate the amount of energy available content of air is high when in the atmosphere. the dew-point is close to air A deep dry layer can eat moisture from temperature. lower layers and thus prevent deep convection.

CAPE and CIN Meteorological indices for Calculated from radiosonde profile. convective available potential Indicate the stability of the atmosphere. energy and convective inhibition.

Vertical wind shear Change in wind speed and Vertical wind shear is needed for direction over height. thunderstorms to become severe and long-lasting.

Lifted parcel path In free convection the air A deep convective layer indicates a will become warmer than its strong chance for thunderstorms. The surroundings and experience top of convection shows the height that accelerating ascent along the thundercloud tops will reach. lifted parcel path until it becomes of equal temperature with its surroundings.

Table 1. Factors affecting convective weather and examples of their interpretation in radiosonde profiles for weather forecasting.

is being lifted, for example, by b) Weak convection c) Elevated convection solar heating and penetrates the inversion layer, the air will become Only weak convection and Even if the air near the surface is warmer than its surroundings and small cumulus clouds occur if very stable, thunderstorms can experience accelerating ascent until the boundary layer is not moist develop by means of elevated it becomes of equal temperature enough, there is not enough convection. If the air above the with its surroundings. This is where convective inhibition to help build stable layer has the required cloud tops form. Vertical wind shear up convective energy, and the characteristics, convection can is a main ingredient for severe and mid-troposphere is too stable for occur if the unstable air becomes long-lasting thunderstorms. thunderstorms to appear. lifted, for example, mechanically by a weather front.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 7 11.3.2016 10.37 Case Study 1: Convective Weather

A real world example can show how even small details in the radiosonde profile can be critical for understanding convective weather.

On Tuesday August 9, 2005, a weather front approached the Helsinki capital area from southeast Finland and caused one of the most dramatic moments during the course of the 2005 World Championships in Athletics. The historic open-air Helsinki Olympic Stadium was hit by an intense thunderstorm with pouring rain and severe wind gusts. Emergency services issued an evacuation of the stadium and electrical power was interrupted.

The storm also caused over thunderstorms forming during the What if the radiosonde 200 emergency incidents near day. The most interesting details in profile cannot be the capital area, including fires, the graph show how temperature trusted? traffic accidents and injuries from and humidity (described by the fallen trees. Railway services dew point) behave through the The correct interpretation of and ship traffic from Helsinki atmosphere. convective weather requires precise harbor were affected, and and accurate observations. The thousands of households were left • The deep convective layer (925 to example radiosonde measurement without electricity. Despite fans’ 250 hPa) indicates a strong chance has several details where a disappointment, everyone was able for thunderstorms, with cloud radiosonde providing erroneous to feel safe and in good hands. The tops reaching heights of around profile details could have changed authorities had been working hard 10 km. the forecast entirely. since the early morning. The Finnish Meteorological Institute had issued • There is a temperature inversion Incorrectly measuring the several thunder alerts and kept the at above 700 hPa. The layer helps depth of the temperature emergency organizations up to date build up convective energy, inversion (Figure 3) of the changing weather. however it is not going to stop the convection. An evaporative cooling error of What does the profile 1.0 °C when the radiosonde emerges tell us? • The meteorological index CAPE from a cloud reduces the depth of of 1,121 J/kg in this case indicates the observed inversion layer. The At noon on August 9, 2005, the an unstable atmosphere with a temperature sensor is measuring Jokioinen sounding station in potential for moderate to strong too low values (red curve) while southwest Finland provided convection in a cold climate. water is evaporating from the the observation depicted on surface of the sensor. The estimated the thermodynamic diagram in • Humid and dry layers in the weak inversion layer may lead the Figure 2. This graph is from the profile indicate the energy forecaster to predict the start of Finnish Meteorological Institute available in the atmosphere. convection earlier in the day, with meteorologist’s workstation. The less energy and without forming radiosonde profile shows clear • Cloud layers affect the amount of thunderstorms. evidence of the possibility of solar warming on the surface.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 8 11.3.2016 10.37 (1) Radiosonde measured temperature profile.

(2) Radiosonde measured dew point profile. Dew point is the temperature where condensation begins, and describes the (1) moisture content of air. (3) Moist Adiabatic Lapse Rate. Shows the rate Top of convection (EL) at which air packet which has reached 100% (4) humidity is cooling when ascending.

(3) (4) Stability indices such as CAPE and CIN describe the likelihood of thunderstorms forming. (2)

Temperature Convective Condensation Level (CCL) represents inversion the height where a lifting air packet becomes saturated. It can be used in estimating CAPE.

Dry/moist layer CCL Figure 2. Radiosonde observation shown on a thermodynamic diagram.

Figure 3. A detail of the temperature Figure 4. A detail of the dew point Figure 5. A detail of the temperature inversion. The red curve depicts an profile. inversion, depicting a capping incorrect measurement. inversion.

Incorrectly detecting Incorrectly detecting multiple cloud layers a capping inversion (Figure 4) (Figure 5)

If the humidity sensor freezes while Only a one-degree error in the dew create enough lift and energy to passing through clouds, or is not point measurement (about 3% RH) break the air parcel through the fast enough, the radiosonde may fail near the surface turns the inversion warmer inversion. This prevents to detect fine structures in upper layer at 925 hPa into a capping the release of convective energy in cloud layers. These cloud layers inversion. A similar effect could upward drafts and the formation of may be relevant in blocking solar occur if the temperature profile has thunderstorms. In these weather radiation from reaching the ground, offset due to incorrect calibration. conditions, it is important to and preventing the start of the The resulting thermodynamic measure accurately the temperature convective process. diagram indicates that solar and especially the humidity in the warming of the surface will not lowest kilometer.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 9 11.3.2016 10.37 Case Study 2: Weak Convection

This example compares two radiosondes launched simultaneously from a site representing summer conditions of a humid continental climate zone. In this case the atmosphere was weakly unstable and only light showers were reported later in the day. The compared radiosondes were Vaisala Radiosonde RS41 and a radiosonde with less advanced technology, missing features such as hydrophobic protection for temperature sensor or active heating of the humidity sensor.

Temperature and humidity profiles from the two radiosondes are shown in Figures 6 and 7. The observations are rather similar close to the ground, however differences appear at above the lowest 1.5 kilometers. In this case, the temperature predicted by several meteorological below. differences are of special interest. indices (see chapter Meteorological The RS41 radiosonde detected two Indices for more information) In this case a forecast based on the temperature inversions, while the is different between the two RS41 sounding shows a relatively other radiosonde detected none. radiosondes, as shown in Table 2 Temperature inversions block the ascending air motion and thus inhibit convection. According to the RS41 measurements, the ascending air would not likely penetrate the

higher inversion layer and cloud RS41 tops would form already at the Less advanced height of 4 km, indicating a weak technology possibility for thunderstorms.

According to the radiosonde with less advanced technology, there are no temperature inversions to inhibit convection. This radiosonde also observes a drier lower troposphere which indicates higher solar radiation reaching the surface, and thus a stronger lifting mechanism for convection. The measurement profile predicts that cloud tops could form at the height of 6.7 km, and could produce lightning as the Figure 6. Radiosonde observations of temperature profiles measured cloud top temperatures are well simultaneously with RS41 and a radiosonde with less advanced technology (left), below -20 °C. The thunder potential and differences between the radiosondes (right).

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 10 11.3.2016 10.37 weak potential for thunderstorms. Only light showers developed during the next few hours. A RS41 Less forecast based on the less advanced advanced technology radiosonde shows a high likelihood of thunderstorms.

This case demonstrates that the detection of temperature inversions and accurate humidity observations are important when forecasting development of convection. The lack of temperature inversions in the profile of the radiosonde with less advanced technology was possibly the result of a wet bulb cooling error in the measurement.

Figure 7. Radiosonde observations of humidity profiles measured simultaneously with RS41 and a radiosonde with less advanced technology (left), and differences between the radiosondes (right).

Radiosonde RS41 technology Less advanced technology CAPE + CIN Weak Strong

LI Weak-moderate Moderate

TT Moderate Strong

KI Moderate Moderate

Cloud top 620 hPa (4 km) 430 hPa (6.7 km)

Forecast Unlikely thunderstorm development Probable thunderstorm development

Observed weather Light rain showers, no thunder

Table 2. Forecasts based on RS41 and a radiosonde with less advanced technology, showing thunderstorm potentials (from meteorological indices) and cloud top heights calculated from the sounding profiles, example forecast decisions, and the weather observation at the sounding site later in the day.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 11 11.3.2016 10.37 Winter Weather

Winter weather in cold climates Challenges in winter Quantitative precipitation often highly impacts the society. weather forecasting forecasting (QPF) is generally Accumulated ice due to freezing regarded as the least certain aspect rain and heavy snow packs can Winter weather forecasting is in NWP models since very small damage energy transfer systems mainly focused on predicting the changes in temperature or moisture and cause cut-offs for many days. track and intensity of synoptic can have a large effect on the type Transportation is heavily affected scale low-pressure systems, surface of precipitation. Temperature by winter precipitation and icing. temperatures, and precipitation variations of ±0.5 °C can change the Snowfall decreases road surface rate and type at the ground. Low- precipitation type to another. friction and visibility, and thus pressure systems are currently worsens driving conditions. For relatively well forecasted, whereas Accurate precipitation type example, in the United States numerical weather prediction forecasting is especially difficult winter precipitation was shown to (NWP) models have difficulty when temperatures near and at the increase traffic collisions by 19 % especially in forecasting small surface are close to freezing. In such and injuries by 13 %, compared with scale surface phenomena and conditions, everything between dry conditions (Journal of Transport shallow layers of air where snow and rain can occur within a Geography, Issue 48). Incorrect temperature and humidity deviate small area. In particular, discerning weather forecasts can become costly from the surroundings. This stems between ice pellets and freezing rain if roads are salted too early or when mainly from a limited vertical can be difficult. Another example is not necessary. The same applies resolution used in NWP models. the conversion of forecasted rainfall to runway maintenance and de- As an example, nocturnal surface rate (mm h-1) to snow depth (cm), icing decisions at airports, causing temperatures in the winter are which depends on the density of unnecessarily delayed or canceled highly dependent on cloudiness, snow, which in turn is drawn from flights. Furthermore, the risk of and the type and amount of the vertical profiles of temperature avalanches is highly related to the precipitation may vary notably due and moisture. precipitation type. to an elevated warm layer.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 12 11.3.2016 10.37 Radiosondes in Forecasting Winter Precipitation

Radiosonde soundings have an predictions. In addition, they help type. In the following we discuss important role since they catch all the forecaster understand situations the six basic types of winter the important atmospheric features where NWP models are likely to be precipitation and their prediction which, when assimilated into NWP incorrect and misleading, and help from radiosonde profiles. models, help produce more accurate interpret the upcoming precipitation

Figure 8. Model radiosonde profiles of temperature and dew-point, showing characteristic features that correspond to various types of winter precipitation.

Below is a description of the c - d) If ice particles fall through than -10 °C, freezing drizzle is weather events in Figure 8. See also an elevated warm layer, they will likely to form. the terminology explanations in melt partly or entirely. Depending Table 3. on the degree of melting and the f) In a snow seeder-feeder thickness of the near-surface cold mechanism ice particles are a) Snowfall is observed when snow layer, either ice pellets (sleet) or formed in an upper-level cloud, crystals are formed aloft in an ice freezing rain will occur. Ice pellets whereas a lower-level cloud formation layer and temperatures are solid particles, while freezing contains only super-cooled water. remain below freezing in the rain consists of super-cooled Falling ice particles will partly whole profile. liquid particles that partly freeze sublimate in a dry layer between when in contact with a surface the clouds. If any ice particles b) If the surface layer is above that is below freezing. reach the lower cloud, they will freezing, part of the flakes will start to glaciate the super-cooled melt to form melting (wet) snow. e) If the saturation layer is shallow cloud droplets and snowfall will and below freezing, but warmer be observed.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 13 11.3.2016 10.37 Interpretation from Factor Explanation radiosonde profile

Ice formation layer Formation of ice crystals and snow by Ice formation (> 50% chance): heterogeneous nucleation when air is saturated T < -10 °C and air is saturated relative to ice. relative to ice

Warm layer (T > 0 °C) Melting of ice or snow particles. Can be a surface Warm layer maximum T:

layer or an elevated layer. Complete melting: Tmax > 3 °C

Partial melting: Tmax = 1-3 °C

No melting: Tmax < 1 °C

Near surface layer Determines the precipitation type near and on the Snow: Tsurface < 1°C surface, and the possible formation of freezing rain, Ice pellets / sleet: ice pellets, or sleet. T < -6 °C in a > 750 m layer Freezing rain: T < 0 °C in a < 750 m layer and at the surface

Saturation layer A layer in which water vapor condensates or is Minimum layer depth for deposited on particles. Indicates the depth of cloud precipitation to form: layer and the type of precipitation. > 500 m

Dry layer A dry layer can block precipitation from reaching Maximum layer depth for the surface due to evaporation aloft, or change the precipitation to occur: precipitation type due to cooling. ~ 1000-1500 m

Table 3. Factors affecting winter weather and examples of their interpretation in radiosonde profiles for forecasting various precipitation types.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 14 11.3.2016 10.37 Case Study 3: Freezing Rain and Ice Storm

This case study demonstrates the importance of the accuracy of radiosonde observations in situations where NWP models are likely to be incorrect and misleading in predicting precipitation type.

In January-February of 2014 the Eastern Europe, especially parts of Slovenia and Croatia, were affected by a long-lasting freezing rain event that covered vast areas with ice. The extreme weather was caused by an encounter of cold arctic and moist subtropical air masses. Accumulation of ice damaged the power transmission network in both countries and large areas of forests were destroyed. The economic loss in both countries was estimated Figure 9. Radiosonde observations of temperature (left) and relative humidity to be hundreds of millions of with respect to ice (right), showing original measurements (solid lines) and euros. NWP models had difficulties modified profiles (dashed lines). The precipitation types for moist (green) and dry forecasting the right precipitation (yellow) layers according to the original measurements are illustrated on type during the event. the right.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 15 11.3.2016 10.37 Temperature and humidity 2.8 °C. In this case it is not obvious Table 4 compares forecast results observations from a Vaisala whether precipitation will fall in for the original sounding and for the Radiosonde RS92 launched at the liquid or partly solid form. It is modifications, using interpretation Zagreb sounding station in Croatia probable that ice formation will not rules from Table 3. In this case a on February 5, 2014, are presented be efficient, and thus clouds will wet-bulb type error would decrease in Figure 9. Light freezing rain was contain mostly super-cooled water the level of melting in the elevated observed at the station during the and freezing rain will be observed at warm layer and thus increase soundings. the ground. the probability of forecasting ice pellets instead of freezing rain. On The radiosonde profiles show In this type of borderline situation, the other hand, a -4 % RH humidity an elevated inversion layer with even small temperature and offset would further decrease the relatively dry and warm air at humidity offsets can change the efficiency of ice formation in the 900 hPa, and a saturated layer forecast towards either solid or originally shallow ice formation between 750 hPa and 640 hPa. The liquid precipitation. The impact of layer. Combined with a temperature mid-tropospheric air is dry and the measurement quality was studied offset of +0.3 °C, which indicates a surface layer is not saturated. There by introducing small offsets of surface temperature above freezing, is a shallow ice formation layer at +0.3 °C and -4 % RH, and a wet-bulb the forecasted precipitation type above 700 hPa, and the warm layer type error to the lowest dry layer, as would be rain. has a maximum temperature of shown in dashed lines in Figure 9.

Modified sounding: Sounding Modified sounding: Original sounding ΔT = +0.3 °C, ΔRH = profile wet-bulb error -4 % Ice formation Shallow layer T < -10 °C Shallow layer T < -10 °C Shallow layer T < -10 °C ➔ Probable ice formation ➔ Probable ice formation ➔ Less probable due to lower humidity

Elevated warm Tmax = 1.9 °C ➔ Partial melting of Tmax = 2.8 °C ➔ Partial melting Tmax > 3 °C ➔ Complete layer ice ➔ Solid and liquid can occur of ice ➔ Rain more probable, melting of ice ➔ Rain also sleet can occur

Surface Tsurface < 0 °C ➔ Rain will Tsurface < 0 °C ➔ Rain will Tsurface > 0 °C ➔ No freeze on the ground ➔ Ice freeze on the ground ➔ Ice freezing on the ground accumulation / sleet accumulation / sleet

Forecast Ice pellets (more probable) or Light freezing rain (more Light rain freezing rain or mix probable) or ice pellets

Observed Light freezing rain weather

Table 4. Forecasts based on the original sounding profile (middle) and two modified profiles introducing a wet-bulb error (left) and temperature and humidity offsets (right). The table shows the reasoning based on factors in the sounding profiles, example forecasts, and the weather observation at the sounding site.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 16 11.3.2016 10.37 Numerical Weather Prediction

Modern weather forecasting is strongly founded on numerical weather prediction (NWP) models. Numerical models rely on many sources of measurement data, and, for example, satellite based remote sensing has become crucial especially over the oceans. However, radiosonde observations remain among the most important model input. Their resolution, accuracy, and full vertical coverage of the atmosphere are essential for model initialization.

The significance of radiosonde observations as input data to NWP models has been studied in several denial studies. In the studies, the NWP model is run several times, each time neglecting some observation type from the input mean impact per observation, especially short-range forecasts, data. Singh, et al. (2014) showed and with the largest impact and in reducing forecast errors for that the radiosonde is one of the on the rainfall prediction skill. wind, temperature, and humidity, most effective terrestial-based Other studies have shown that especially on the Northern instruments in reducing forecast radiosondes are among the main hemisphere. See the references error with the largest total and contributors to the quality of below for more information.

Studies on radiosonde observations impact on NWP models: Gelaro, R. & Zhu, Y., 2009. Examination of observation impacts derived from observing system experiments (OSEs) and adjoint models. Tellus A, Volume 61, pp. 179-193. Ingleby, B., Rodwell, M. & Isaksen, L., 2016: Impact of halving the number of Russian radiosonde reports. The AMS 96th Annual Meeting, New Orleans, USA. Kutty, G. & Wang, X., 2015. A Comparison of the Impacts of Radiosonde and AMSU Radiance Observations in GSI Based 3DEnsVar and 3DVar Data Assimilation Systems for NCEP GFS. Advances in Meteorology, Volume 2015, pp. 1-17. Laroche, S. & Sarrazin, R., 2010. Impact study with observations assimilated over North America and the North Pacific Ocean on the MSC global forecast system. Part I: Contribution of radiosonde, aircraft and satellite data. Atmosphere-Ocean, 48(1), pp. 10-25. Singh, R., Ojha, S. P., Kishtawal, C. M. & Pal, P. K., 2014. Impact of various observing systems on weather analysis and forecast over the Indian region. Journal of Geophysical Research: Atmospheres, Issue 119, pp. 232-246. Zapotocny, T. H., Jung, J. A., Le Marshall, J. F. & Treadon, R. E., 2008. A Two-Season Impact Study of Four Satellite Data Types and Rawinsonde Data in the NCEP Global Data Assimilation System. Weather and Forecasting, Volume 23, pp. 80-100.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 17 11.3.2016 10.37 Radiosondes in Validating NWP Models

Radiosonde observations can be of to predict, such as melting and The importance of measurement great utility for determining which freezing layers, surface winds and accuracy is emphasized when numerical weather prediction low-level temperature inversions radiosonde observations are (NWP) model performs best, and and clouds. These are important for applied in forecasting. Accuracy for detecting phenomena that predicting phenomena such as fog, of temperature measurements models fail to predict. Radiosonde low-level wind jets, and surface air is important, for example, when profiles are helpful when forecasting quality. Inspection of radiosonde determining the precipitation type phenomena that depend on the profiles may be currently underused based on the temperature of a vertical profile of the atmosphere due to the large variety of forecast cloud layer when temperatures are and exist at a resolution that is products and model analysis tools close to 0 °C. Accuracy of humidity poorly represented in NWP models. that meteorologist’s workstations measurements is important when They can also discern shallow offer. determining whether a shallow features that models often fail cloud is deep enough to produce drizzle.

Case Study 4: Fog

Fog is an important winter weather phenomenon in the British Isles. For Forecasting fog is challenging for NWP models since the formation, example, Heathrow Airport, one of evolution, and dissipation of fog depend on interactions between the busiest airports in the world, small-scale physical processes that are parametrized in the models. is notorious for flight cancellations Fog is often more shallow than the height of the lowest grid point in the model. The forecasting of fog and its dissipation remain a great due to fog-related reduced visibility financial issue for airport operations. in the winter months.

A persistent layer of thick fog appeared over the whole British Isles during the first days of November 2015. The fog was formed as a radiation fog where night-time radiative cooling decreases the temperature of moist surface air below its dew-point. The density of the fog made it long-lasting since solar radiation could not reach the surface to dissipate it during the day.

Figure 10 compares an RS92 radiosonde observation and an NWP model run profiles at 11 UTC when a dense fog reduced the visibility at the Nottingham sounding Figure 10. Comparison of radiosonde sounding (black) and NWP model (red) station to below 100 m. Visibility profiles of temperature (solid lines) and dew-point profile (dashed lines). The did not improve over the day. The sounding is from a sounding station in Nottingham, UK, at 11 UTC on November dense fog layer is well visible in 2, 2015. The NWP model run (WRF-EMS) is for 11 UTC.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 18 11.3.2016 10.37 the radiosonde observations, with dry compared with the radiosonde example shows the difficulty that a very moist surface layer with profile, with differences between models have in fog prediction for temperatures equal or close to temperatures and dew-point only a few hours ahead. Accurate their dew-point temperatures up to temperatures of several degrees sounding observations in the lower several hundreds of meters. There Celsius. The erroneous model troposphere can be of great help. is a sharp inversion layer which prediction can be explained by In this example, the sounding prevents vertical mixing of the the higher surface winds and a profile tells that the fog layer will moist surface air with the relatively less sharp temperature inversion not dissipate during the whole day, dry air above. The radiosonde which weaken the fog layer. Once in contrast to the NWP prediction. profile correctly implies a persistent initiated, the dissipation of fog by This type of information is very fog situation. solar radiation is self-feeding: a less important, for example, for airport dense fog layer permits more solar operations. In addition to accurate The NWP model forecast for 11 radiation to reach the ground and temperature and humidity profiles, UTC was done 5 hours before the to heat the surface. Thus, relative the accuracy of wind measurements observation. The model prediction humidity decreases and the fog can be crucial for correctly implies that the fog that existed layer continues dissipating. forecasting the dissipation of the during the previous night would fog layer. have dissipated by 11 UTC. The A summary of the forecast results model surface layer is relatively is presented in Table 5. This

Data source Radiosonde WRF-EMS numerical model Surface layer Saturated Relatively dry

Low level winds Very weak Weak / moderate

Temperature Sharp inversion Moderate inversion profile

Forecast Persistent fog Dissipation of fog

Observed Persistent fog weather

Table 5. Forecasts based on a radiosonde sounding and an NWP model, showing the reasoning from factors in the atmospheric profiles, the forecasts, and the weather observation at the sounding site later in the day.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 19 11.3.2016 10.37 Meteorological Indices

Forecasting in severe weather forecasting of deep convection and situations requires making the right thunderstorms. conclusions from a large amount of information sources in a short time. Some of the most commonly used Meteorological indices can help the indices for predicting severe forecaster make faster conclusions weather are described in Table 6. from radiosonde observations when The indices show the potential for estimating the possibility of severe severe weather, typically expressed weather. Meteorological indices are by the categories of “weak”, calculated from sounding profiles. “moderate”, or “strong”. Threshold An index describes in a single value values for the different categories some aspect of the state of the are approximate and depend on the atmosphere, such as the amount of season and climate. Some indices energy a parcel of air would have if have different calculation options; lifted a certain distance vertically as an example the calculation of through the atmosphere. This index CAPE can start from the surface is known as the convective available or from a higher level, where the potential energy (CAPE). convection updraft is expected to initiate. Therefore, the optimal use Most typical indices are so-called of indices requires expertise of the stability indices which have been local weather and of how to best specifically invented to improve apply the indices.

INDEX TYPE OF USE UNIT WEAK MODERATE STRONG BRN Storm cell type - < 10 (pulse type > 50 (multicells) 10-50 Bulk convection) (supercells) Richardson Number CAPE Deep moist convection J kg-1 USA: < 500 USA: 500-2000 USA: > 2000 Convective (thunder) Europe: < 100 Europe: 100-1000 Europe: > 1000 Available Potential Energy CIN Likelihood of convection J kg-1 < -50 > -50 Convective to form Inhibition SWEAT Severe weather potential - < 200 200 – 300 > 300 Severe Weather Threat DCAPE Downdraft strength J kg-1 < 500 500 – 800 > 800 Downdraft CAPE KI Instability °C (or K) < 20 20 – 30 > 30 K-index LI Instability and thunder °C (or K) USA: > -2 USA: -2 … -4 USA: < -4 Lifted index potential Europe: > 0 Europe: 0 … -2 Europe: < -2 SI Showalter Instability °C (or K) > 0 (showers) 0 … -3 (thunder) < -3 (severe Index thunder) TT Instability and storm °C (or K) < 44 44 – 50 > 50 (isolated Total Totals strength (thunder likely) severe storms) Index Table 6. Summary of common meteorological indices.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 20 11.3.2016 10.37 Impact of Radiosonde Data Quality

Figure 11. The geographic areas covered in the study.

In the following we will look at the original profiles, and from profiles results indicate how radiosonde impact of radiosonde measurement with small offsets: -4 to +2 % RH for measurement errors are propagated accuracy on meteorological indices relative humidity and ±0.2 °C for into errors in meteorological and the forecasts based on them. temperature. These offset values indices. are rather conservative, and To evaluate the impact of demonstrate the sensitivity of The indices were more sensitive to radiosonde data quality and its the indices to small uncertainties humidity offsets than temperature significance to meteorological in radiosonde observations. offsets. When all 56 soundings indices, we studied a set of 56 Wind speed and direction are are considered (upper graph), a soundings from three geographical also significant components in 4 % RH negative bias caused mean areas in conditions that resulted in calculating indices. They were not relative changes of -19 % in BRN, severe weather. The areas studied modified in this study as the wind -29 % in CAPE, +21 % in LI, 5 % in were North America, Central measurement accuracy is typically DCAPE, and 6 % in KI indices. A Europe, and Northern Europe. The very good in radiosondes. ±0.2 °C temperature bias caused soundings were done with either small changes of less than 7 % in Vaisala Radiosonde RS92 or RS41. Results all indices. Case examples from individual flights indicated that The impact of measurement quality Figure 12 shows the mean relative the impact of temperature may be was simulated by introducing change (%) in five meteorological more considerate in wet-bulb error artificial errors to the original indices as a result of small humidity situations. Complex indices which sounding profiles, which were used or temperature offset errors. BRN, are calculated by integrating the as the reference. Meteorological CAPE, and LI were calculated using values in the sounding profile (for indices were calculated from the surface as the start level. The example, CAPE) changed more,

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 21 11.3.2016 10.37 while simple indices using only a few heights (DCAPE, KI) were less sensitive.

The changes are more pronounced when we inspect borderline situations where the evolution of convection and the possible initiation of thunderstorms have more uncertainty. The lower graph shows the results for weakly convective cases with CAPE < 1000 J kg-1. A 4 % RH negative bias caused larger mean relative changes in CAPE, BRN and LI indices. The changes were -43 % in BRN, -49 % in CAPE, and +34 % in LI. A ±0.2 °C temperature bias caused 2 – 20 % changes in these indices.

Summary The results show that radiosonde measurement accuracy is important for the correct estimation of the meteorological indices. In borderline situations even small errors can change the indices considerably and may lead the meteorologist to underestimate or overestimate the arrival of convective weather.

Figure 12. Mean relative change (%) in meteorological indices calculated from radiosonde observations when humidity or temperature offset is added to the profiles.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 22 11.3.2016 10.38 Radiosonde Pressure Measurement

Accurate observations of the It is needed for calculating other atmospheric pressure are among variables, such as height and mixing the most important radiosonde ratio, and for various corrections. observations for weather Heights of standard pressure levels forecasting. In the following we will are important in forecasting the look at the two currently available strength and evolution of weather pressure measurement technologies systems. for the radiosonde and discuss how to conduct the pressure It is essential to verify that pressure measurement to obtain the required measurements are accurate. A low measurement accuracy for different quality measurement associates the application purposes. temperature and humidity values with wrong pressure levels. This Pressure information is used can produce errors in numerical in many ways. It is the vertical weather and climate models that are coordinate for temperature, larger than the actual measurement humidity, and wind profiles from uncertainty of the temperature or the radiosonde when they are humidity sensor in the radiosonde. assimilated into numerical models.

WMO accuracy requirements for upper-air measurements for synoptic meteorology are 1 hPa near the surface, increasing to 2 hPa near 100 hPa, and 2 % from 100 to 10 hPa (WMO-No. 8, CIMO Guide, Annex 12.A, 2008 Edition).

Two pressure observation methods

Atmospheric pressure and height pressure sensor. Similarly, height measurements are closely related. can be measured directly from the As the height increases, the GPS satellite navigation system, pressure decreases, following or indirectly from pressure, a nearly logarithmic profile. temperature, and humidity sensors. Pressure and height profiles can The two procedures are depicted in be derived from each other with Figure 13. small corrections, taking into account the air density variations. The two independent methods Consequently, there are two main give useful options suitable for principles available for radiosondes different radiosonde applications. for determining the atmospheric In the following we will describe pressure. their respective strengths and weaknesses in terms of precision Pressure can be measured indirectly and accuracy of measurement. using the radiosonde measurements Choice of the optimal observation of GPS height, temperature, method depends on the application and humidity, or directly with a where the measurement is used.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 23 11.3.2016 10.38 Measured Calculated

GPS- GPS Surface pressure based + Temperature height Pressure method Humidity

Pressure Sensor Surface height sensor + Temperature pressure Height method Humidity

Figure 13. Comparison of GPS-based and sensor-based pressure measurement methods.

GPS-based Pressure

The use of GPS height, temperature, and humidity for estimating pressure is the most common method used at upper air sounding stations. Omitting the pressure sensor allows simpler radiosonde design and pre-flight preparations. The ground level value of the pressure profile is obtained from a barometer at the sounding station. The barometer observation also calibrates all values in the profile. The change in pressure between each measurement point is solved using the height, temperature, and humidity information. The method assumes a hydrostatic balance, which is a valid model for most situations.

Accuracy of GPS-based pressure measurements The accuracy of GPS height measurement is fairly constant through the range of heights used A sounding system with GPS components. in a radiosounding. As a result, the GPS-based method provides a very high precision of pressure in the stratosphere where the atmosphere where an error of a few designed to fulfill the requirements pressure change with increasing meters in height will lead to several of the radiosonde application. height is small. With this method, tenths of hPa error in pressure. the measurement quality is critical The GPS antenna hardware and the An important factor affecting the in the lowest kilometers of the GPS location algorithms must be accuracy of the measurement is

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 24 11.3.2016 10.38 the surface pressure. The station When to select the GPS- barometer must be properly based method? calibrated, and the height settings, such as the local antenna and The GPS-based pressure method barometer heights, correctly is the WMO recommended choice configured in the sounding software. (WMO-No. 8, CIMO Guide, Chapter 12.3) for synoptic soundings. This The quality of the radiosonde method provides sufficient accuracy temperature measurement has for the lower atmosphere and the a moderate impact on pressure benefit of very high accuracy in accuracy, while the humidity the upper atmosphere, where the measurement has a very minor quality of long-term time series of impact. As an example, a persistent quantities such as temperature and temperature bias of 0.1 – 0.2 °C ozone require accurate pressure and causes an error of up to 0.15 – 0.3 height coordinates. The GPS-based hPa in the GPS-derived pressure method is also suitable for climate profile. observations and atmospheric research campaigns.

Sensor-based Pressure

Upper electrode, Vacuum chamber A pressure sensor observes the pliable silicon atmospheric pressure directly by diaphragm Ambient measuring the force produced by pressure the column of the atmosphere above chamber the radiosonde. This is the “true” atmospheric pressure, in contrast to the GPS-based method which assumes a hydrostatic balance in the atmosphere.

The inclusion of the pressure sensor to the radiosonde design leads Base electrode to some additional requirements, such as a calibration step during Figure 14. Construction of a capacitive pressure sensor. ground preparations. An adjustment against a well calibrated reference constant, but as pressure levels are redundancy option by measuring barometer takes into account very low, the relative errors become both sensor pressure and GPS-based sensor drifting during storage and larger. When pressure is used as pressure from the same radiosonde transport. the vertical coordinate for other sounding can be useful for quantities, such as temperature and researchers for understanding the Accuracy of pressure ozone, the pressure measurement atmospheric conditions. The sensor sensor measurements uncertainty may produce inaccurate measurement may also be preferred observations of those quantities in for continuity of the measurement A well calibrated pressure sensor the stratosphere. method for long time series of data provides very high accuracy and for climate research, or when very precision in the lowest kilometers When to select the high precision is needed for the of the atmosphere. The accuracy of pressure sensor method? lowest kilometers of the atmosphere. the sensor measurement is limited In highly non-hydrostatic conditions by the sensor’s dynamic range in The sensor-based pressure the pressure sensor may give more the upper atmosphere. The absolute measurement is suitable for accurate results, as indicated by the accuracy (hPa) remains fairly synoptic soundings. The data following case study.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 25 11.3.2016 10.38 Case Study 5: Tropical Cyclone

Figure 15. . Map of 1000 hPa pressure level heights in South-East Asia at 12 UTC on July 22, 2014.

Highly non-hydrostatic and Typhoon Matmo passed over maximum differences correlated with horizontally non-homogeneous Taiwan in July 2014, and tropical maximum wind velocities during the environments, such as those storm Fung-wong in September 2014. flights. observed near the eye of a tropical Radiosonde soundings using Vaisala cyclone, are interesting when Radiosondes RS41 and RS92 from a Smaller differences of 1 – 2 hPa comparing the two pressure sounding station located in Taiwan were measured after the eye of the measurement methods. Some were inspected from those time storm had passed the area, and also differences between the results can periods. A comparison of GPS and during the lower category tropical be expected as the assumptions sensor pressure profiles from RS92 storm Fung-wong. Although a more used in the GPS-based method showed some unusual differences comprehensive study is needed for may not describe the state of the when typhoon Matmo was close to verification, these results indicate atmosphere accurately. the sounding site, see Figures 15 that in highly non-hydrostatic and 16. Largest observed pressure situations a high quality pressure difference was 5.5 hPa at near sensor measurement gives a more 400 hPa level. The heights of the accurate result than the GPS-method.

Figure 16. Differences between GPS-based and sensor pressure during three soundings on July 22, 2014. Pressure levels are shown in red.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 26 11.3.2016 10.38 Radiosonde Data in Climatology

Climate research and monitoring several regional and international cirrus clouds. set very high requirements for the organizations and programs, such • Studying environmental observation instruments. Changes as the GCOS Reference Upper-Air characteristics, such as air in the trends of atmospheric Network (GRUAN). quality, pollutant dispersion, and variables should be detected early urban climate. and understood quickly. This calls It is important to monitor • Validation of reanalysis data used for an optimal mix of independent the atmosphere with several in climate studies. observations with strict accuracy independent observation systems. • Verification and development of and data continuity requirements. High measurement accuracy climate models. and data continuity are essential • Radiosondes are used as a Radiosonde measurements have requirements. Radiosonde platform for measuring additional several strengths compared with profiles are used in many ways in variables, such as atmospheric other observation systems. Data climatology: ozone. is available from the surface to • Radiosonde measurements form the stratosphere and the height • Monitoring temperature profiles the baseline of performance for resolution is much better than and trends from all heights. validating and calibrating satellite with satellite data. The time • Monitoring upper air humidity data, and help mitigate systematic series of radiosonde data span content. Mid-tropospheric to errors that might otherwise go several decades; however, the stratospheric water vapor impacts unnoticed. homogenization of such long time the thermal energy radiated into series is challenging. Continuous space and thus affects the global Radiosonde profiles also help to monitoring of the Earth’s greenhouse effect. understand and to prepare for the climate is managed by the World • Understanding the vertical consequences of climate change, Meteorological Organization distribution of clouds, and the such as the increasing occurrence of (WMO) in collaboration with initialization and maintenance of extreme weather events.

Networks for upper-air climate observations: GCOS Upper-Air Network (GUAN) www.wmo.int/pages/prog/gcos/ GCOS Reference Upper-Air Network (GRUAN) www.gruan.org

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 27 11.3.2016 10.38 Radiosonde Accuracy Matters

Radiosonde measurements are the most accurate information available for understanding weather. Their information is relevant for both long-range forecasts and for estimating how weather will change in the next few hours. An erroneous measurement could lead to wrong decisions. It does not make sense to measure with a non-optimal instrument.

Vaisala Radiosonde RS41

Usability and error preventive design of the radiosonde and MW41 sounding system bring efficiency and reliability to operations.

Platinum Resistive Temperature Ground Check Device RI41 is an Sensor with excellent stability, important part of reliable operations. very fast response time, small Each radiosonde is checked prior to solar radiation error and launch to detect possible injuries e.g. effective protection from wet during transportation and to remove bulb cooling error. contaminants from the humidity sensor. • Humidity: integrated heating elements of RS41 provide physical zero reference Vaisala Humicap® Humidity level check without desiccants or other Sensor with active de-icing external references method to prevent freezing • Temperature: stabile, yet verified with and on-chip temperature the secondary temperature sensor measurement to eliminate solar integrated into RS41 humidity sensor radiation error.

Calibration against SI traceable references, comprehensively analyzed and verified regularly. Proven measurement performance tested thoroughly in laboratory and in several test campaigns in different climates.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 28 11.3.2016 10.38 Further reading

Vaisala Radiosonde RS41 Measurement Performance White Paper

Vaisala Radiosonde RS41 Calibration Traceability and Uncertainty White Paper

Impact of Radiosonde Measurement Accuracy on Meteorological Indices, Vaisala 2015

GPS-Based Measurement of Height and Pressure with Vaisala Radiosonde RS41 White Paper

WMO Intercomparison of High Quality Radiosonde Systems, Yangjiang, China, 12 July – 3 August 2010

D. Edwards, G. Anderson, T. Oakley, P. Gault: UK MET Office Intercomparison of Vaisala RS92 and RS41 Radiosondes, 2014

www.vaisala.com > Sounding Systems and Radiosondes > Sounding Data Continuity

For more information on Vaisala sounding products, visit www.vaisala.com

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 29 11.3.2016 10.38 Please contact us at Ref. B211548EN-A ©Vaisala 2016 This material is subject to copyright protection, with all www.vaisala.com/requestinfo copyrights retained by Vaisala and its individual partners. All rights reserved. Any logos and/or product names are trademarks of Vaisala or its individual partners. The reproduction, transfer, distribution or storage of information contained in this brochure www.vaisala.com in any form without the prior written consent of Vaisala is strictly Scan the code for prohibited. All specifications — technical included — are subject more information to change without notice.

WEA-MET-G-Accuracy-Matters-in-Radiosonde-Measurements-White-Paper-B211548EN-A.indd 30 11.3.2016 10.38