Evaluation of measurement errors of temperature and

Environ Monit Assess (2015) 187:236 DOI 10.1007/s10661-015-4458-x

Evaluation of measurement errors of temperature and relative humidity from HOBO data logger under different conditions of exposure to solar radiation Antonio Ribeiro da Cunha

Received: 28 May 2014 / Accepted: 18 March 2015 # Springer International Publishing Switzerland 2015

Abstract This study aimed to assess measurements of temperature and relative humidity obtained with HOBO a data logger, under various conditions of exposure to solar radiation, comparing them with those obtained through the use of a temperature/relative humidity probe and a copper-constantan thermocouple psychrometer, which are considered the standards for obtaining such measurements. Data were collected over a 6-day period (from 25 March to 1 April, 2010), during which the equipment was monitored continuously and simultaneously. We employed the following combinations of equipment and conditions: a HOBO data logger in full sunlight; a HOBO data logger shielded within a white plastic cup with windows for air circulation; a HOBO data logger shielded within a gill-type shelter (multiplate prototype plastic); a copper-constantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight; and a temperature/relative humidity probe under a commercial, multi-plate radiation shield. Comparisons between the measurements obtained with the various devices were made on the basis of statistical indicators: linear regression, with coefficient of determination; index of agreement; maximum absolute error; and mean absolute error. The prototype multiplate shelter (gill-type) used in order to protect the HOBO data logger was found to provide the best protection against the effects of solar radiation on measurements of temperature and relative humidity. The A. R. da Cunha (*) Unesp, São Paulo State University, School of Agronomic Sciences, Botucatu, SP, Brazil e-mail: [email protected]

precision and accuracy of a device that measures temperature and relative humidity depend on an efficient shelter that minimizes the interference caused by solar radiation, thereby avoiding erroneous analysis of the data obtained. Keywords Multi-plate radiation shield . Efficient shelter . Solar radiation . Temperature and relative humidity

Introduction The automation of equipment has provided benefits to various areas by facilitating the obtainment of large volumes of data in real time, as well as the transfer and storage of those data. Automating the data collection process, because of the shorter scan times that implies, creates a need for a more representative measurement, which facilitates obtaining measurements that are more consistent and allows for greater quality and standardization in data collection, reducing errors in instrument reading, data entry, and interpretation of the data obtained. Inappropriate equipment can provide low quality data (WMO 2008). According to Danni-Oliveira (2002), the monitoring of meteorological data presents difficulties in relation to the equipment used in measuring variables, the team responsible for the equipment, the standardization of procedures adopted to measure the data, and the measuring equipment employed.

236

Environ Monit Assess (2015) 187:236

Page 2 of 11

The WMO (2008) recommends the calibration of a sensor as the first step in ensuring the validity and quality of data, as well as specifying the type of exposure, placement, and procedures for the reading of the sensor. As an example, air temperature is defined as the temperature indicated by a thermometer exposed to air and protected from exposure to direct sunlight. To avoid uncertainty in the measurement of air temperature, it must be ensured that the thermometer element is in thermal equilibrium with the air when used in a shelter, because its efficiency relies on the air in the shelter being the same temperature as the air immediately surrounding it (WMO 2008). Many types of shelters are in use worldwide, each with its specific characteristics (Van der Meulen and Brandsma 2008). To avoid misinterpretations caused by inappropriate uses of shelters (Spetalen et al. 2000; Sun and Baker 2004), it is important to evaluate the efficiency of the shelters employed. HOBO data loggers are among the most widely used instruments for obtaining air temperature. There are many models and manufacturers of data loggers, and it is therefore important to evaluate their accuracy, precision, memory capacity, durability, and programming (Dunham et al. 2005). One example is the HOBO data logger (HOBO 2002), a low cost, durable, compact device that consumes little power (Whiteman et al. 2000; Nakamura and Mahrt 2005; Mauder et al. 2008; Palmieiri 2009), provides acceptable precision and has a storage capacity that is sufficient for multiple meteorological and climatological investigations (Whiteman et al. 2000). However, in some cases, HOBO data loggers are being used without criteria for their protection against the effects of solar radiation. Therefore, the

present study aimed to evaluate the measurements of temperature and relative humidity obtained with HOBO data loggers under different conditions of exposure to solar radiation, comparing them with those obtained through the use of standard sensors—a temperature/ relative humidity probe and a copper-constantan thermocouple psychrometer.

Materials and methods We collected data over a 6-day period (from 25 March to 1 April, 2010), during which the equipment was monitored continuously and simultaneously. Measurements were obtained in a grassy area on the campus of the São Paulo State University School of Agronomic Sciences, in the city of Botucatu, Brazil (22° 51′S; 48° 26′ W; altitude, 786 m). We measured mean air temperature (°C) and mean relative humidity (%) at a height of 2 m. Data were stored once every 15 min. We employed the following combinations of equipment and conditions, which are depicted in Fig. 1: a HOBO data logger (Onset Computer Corp., Pocasset, MA) in full sunlight (Hsun); a HOBO data logger shielded within a white plastic cup with windows for air circulation (Hcup); a HOBO data logger shielded within a gill-type shelter (multi-plate prototype plastic—Hgill); a copper-constantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight (Psy), as described by Cunha et al. (2001); and a temperature/relative humidity probe (HMP45C; Vaisala, Helsinki, Finland) under a multi-plate radiation

Psy

Hgill

Vais Hcup

Hsun

Fig. 1 Equipment used for measurements of air temperature and relative humidity. Hgill, HOBO data logger shielded within a gilltype shelter (multi-plate prototype plastic); Hsun, HOBO data logger in full sunlight; Hcup, HOBO data logger shielded within

a white plastic cup with windows for air circulation; Psy, copperconstantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight; and Vais, temperature/relative humidity probe under a commercial, multi-plate radiation shield


shield (41002; RM Young Co., Traverse City, MI) to protect it from sunlight (Vais). The multi-plate prototype plastic shelter used in the Hgill combination consists of six white overlapping plastic plates that facilitate the passage of air while protecting the device from direct sunlight and rain, as well as functioning as a thermal insulator, in accordance with the recommendations of the WMO (2008). In the case of HOBO data loggers, two replicates were used for each condition. Therefore, we employed a total of six HOBO data loggers, all of which were previously calibrated. The HOBO data loggers have built-in temperature and relative humidity sensors, collecting and storing their own measurements, whereas the sensors of the thermocouple psychrometer and Vaisala probe had to be connected to a separate data logger (CR21XL Micrologger; Campbell Scientific, Inc., North Logan, UT) to collect and store their measurements. Table 1 presents the characteristics of the two types of data loggers employed.

Page 3 of 11 236

To compare the mean values of temperature and relative humidity obtained in two repetitions by HOBO data loggers under the various exposure conditions (Hsun, Hcup and Hgill), as well as to compare the mean values among the Hsun, Hcup, Hgill, Vais, and Psy combinations, we used t tests at a probability of 5 %. The t test shows the differences between the means of two distinct populations—independent samples. To calculate the hidden outlier means over the course of a day, the temperature and relative humidity readings were separated into diurnal data (those collected between 0630 and 1830 hours) and nocturnal data (those collected between 1845 and 0615 hours). Comparisons were made between the measurements derived from each data logger or sensor device-condition combination and those obtained with the Psy combination, which is considered the standard for measurements of temperature and relative humidity and suitable for the calibration of other sensors (WMO 2008). Comparisons were made on the basis of the following statistical indicators:

Table 1 Characteristics of the data loggers used for obtaining of temperature and relative humidity Characteristic

Onset HOBO data logger

Campbell Scientific CR21XL micrologger

Data storage

8 GB; stored data is not lost when battery dies

4 GB; stored data is not lost when battery dies

Reason sampling

0.5–32,400 s

0.0125–6553 s

Operating range

−20 °C to 70 °C 0–95 % non-condensing

−25 °C to 50 °C 0–100 % non-condensing in protective housing

Types of entries

Internal measurements of temperature and relative 16 analog inputs in single-ended or 8 differential; 4 pulse humidity; two external inputs for additional count inputs; 4 excitation outputs; 2 analog continuous parameters outputs; and 6 digital controls

Type of battery

3 V lithium (CR-2032; shelf life: 1 year)

7 AHr rechargeable via AC or solar panel photovoltaic (lifespan: 5 years) 1800 mAHr lithium - clock and RAM (lifetime: 10 years)

Weight

29 g

1600 g

Dimensions

6.1 cm×4.8 cm×2.0 cm

24.1 cm×17.8 cm×9.6 cm

Warranty

1 year

3 years

Advantages

Low cost and ease of handling

Channels compatible with most sensors; accepts inputs from multiple sensors; statistics instructions; insertion of mathematical equations

Disadvantages

Limited number of sensors

Higher cost; programming careful, specific and complex

Program support

BoxCar Pro 4.3 for Windows - support for telecommunication, adjustments, manipulation and data transfer Type: thermistor; measurement range: −20 °C to 70 °C; accuracy: ±0.7 °C

PC208W for Windows - support for telecommunication, programming, processing and data transfer

Temperature sensor

Relative humidity sensor Type: polymer film semiconductor; measurement range: 0–95 % (non-condensing); accuracy (at 27 °C): ±5 % (condensation damage sensor)

Type: platinum resistance temperature detector; measurement range: −40 °C to +60 °C; accuracy (at 20 °C): ±0.4 °C Type: capacitor (slim polymer film); measurement range: 0–100 % (non-condensing); accuracy (at 20 °C): ±2 % (0–90 %) and ±3 % (90–100 %)


Page 4 of 11

if properly calibrated should provide identical. The discrepancy was due to the decalibration of Hcup2, which consequently underestimated the relative humidity (Fig. 3). This problem, which has previously been addressed (Cunha and Martins 2004), illustrates the fact that the relative humidity sensors need to be calibrated often and occasionally must be replaced, because they are subject to damage from condensation or contaminated air (Campbell 1990). For Hcup1 and Hcup2, the amplitude of variation of those measurements was greater during the day than during the night, due to the effect of irradiance on each sensor. At night, some irradiation originates from the soil, which releases the heat stored during the day. However, without the influence of solar irradiance, there was less amplitude of variation (Fig. 3). Using the shelter recommended by the WMO (2008) had a notable effect on the measurements of temperature, which were quite similar throughout the day between the Hgill1 and Hgill2 combinations. Because of the atmospheric stability surrounding the sensor, there was no diurnal variation in amplitude caused by the effect of solar irradiance. As occurred in the case of Hcup2 (Fig. 3), the data logger used in the Hgill1 combination became decalibrated and underestimated the relative humidity (Fig. 4). Measurements of temperature are subject to errors due to heating of the sensor by exposure to sunlight even when the principle of measurement is precise and accurate (Lin et al. 2001; Huwald et al. 2009), making it necessary and important to evaluate the temperature data in view of the type of shelter used (Hubbard et al. 2001).

linear regression (y=bx), coefficient of determination (R2); index of agreement (d); maximum absolute error (MAE); and mean absolute error (ME) (Willmott et al. 1985). These statistical indicators were used to evaluate the accuracy and precision of the measures. When the measures from each data logger or sensor devicecondition combination were equal to obtained with the Psy combination (ME=MBE=0 and R2 =d=1) indicates elevated accuracy and precision the measures.

Results and discussion

50

100

45

90

40

80

35

70

30

60

25

50

20

40

15 10 5

JD84

JD85

30

JD88

T Hsun1 = 25.04 ºC (a) T Hsun2 = 24.39 ºC (a) RH Hsun1 = 68.79% (A) RH Hsun2 = 68.29% (A)

JD89

JD90

JD91

0

20 10 0

13:0 0 18:0 0 23:0 0 04:0 0 09:0 0 14:0 0 10:0 0 15:0 0 20:0 0 01:0 0 06:0 0 11:0 0 16:0 0 21:0 0 02:0 0 07:0 0 12:0 0 17:0 0 22:0 0 03:0 0 08:0 0 13:0 0

Fig. 2 Measurements of air temperature (T, °C) and relative humidity (RH, %) obtained with the two HOBO data loggers exposed to full sunlight (Hsun1 and Hsun2). Values followed by the same letter did not differ significantly on the t test at a significance level of 5 %. JD, Julian day

T (ºC)

The two HOBO data loggers exposed to solar radiation (Hsun1 and Hsun2) showed differences in terms of the measurements of temperature and relative humidity (Fig. 2). The amplitude of variation of those measurements was greater during the day than during the night, due to the effect of irradiance on each sensor. This indicates that, in order to maintain measurement stability, the HOBO data logger should be surrounded by a shelter (Fig. 1), which will protect the sensor from direct exposure to solar radiation, ensuring the free circulation of air and maintaining equilibrium with the surrounding atmosphere, thus avoiding erroneous measurements. The two HOBO data loggers shielded within a plastic cup (Hcup1 and Hcup2), showed no significant differences in terms of the values for temperature but differed in terms of the relative humidity values obtained. This should not have occurred, given that the two data loggers were installed side by side (quite close to each other) and

Time of day

RH (%)

236


45

100 JD84

JD85

JD88

JD89

JD90

JD91

90

40

80

35

70

30

60

25

50

20

40

15

30

10

20

T Hcup1 = 24.09 ºC (a) T Hcup2 = 24.06 ºC (a) RH Hcup1 = 67.52% (A) RH Hcup2 = 51.92% (B)

5

RH (%)

50

T (ºC)

Fig. 3 Measurements of air temperature (T, °C) and relative humidity (RH, %) obtained with the two HOBO data loggers within white plastic cups with windows for air circulation (Hcup1 and Hcup2). Values followed by the same letter did not differ significantly on the t test at a significance level of 5 %. JD, Julian day

Page 5 of 11 236

10 0

13:0 0 18:0 0 23:0 0 04:0 0 09:0 0 14:0 0 10:0 0 15:0 0 20:0 0 01:0 0 06:0 0 11:0 0 16:0 0 21:0 0 02:0 0 07:0 0 12:0 0 17:0 0 22:0 0 03:0 0 08:0 0 13:0 0

0

Time of day

Some microclimate studies employ simplified shelters that prevent direct sunlight from striking the instrument but do very little to limit the exchange of radiation, especially long wave radiation, among the instrument, the surrounding surfaces, and the atmosphere (Azevedo 2001). According to Brandsma and Van Der Meulen (2008), it is important to develop specific shelters for specific climatic conditions, in order to reduce the impact they have on the measured air temperature. The Vais and Psy combinations, which were used as references for comparison in this study, showed no differences in terms of the measurements of temperature 50

100 JD88

JD84

JD89

JD90

JD91

90

40

80

35

70

30

JD85

60

25

50

20

40

15

30

10 5

20

T Hgill1 = 22.24 ºC (a) T Hgill2 = 22.72 ºC (a) RH Hgill1 = 41.82% (A) RH Hgill2 = 75.77% (B)

10 0

13:0 0 18:0 0 23:0 0 04:0 0 09:0 0 14:0 0 10:0 0 15:0 0 20:0 0 01:0 0 06:0 0 11:0 0 16:0 0 21:0 0 02:0 0 07:0 0 12:0 0 17:0 0 22:0 0 03:0 0 08:0 0 13:0 0

0

Time of day

RH (%)

45

T (ºC)

Fig. 4 Measurements of air temperature (T, °C) and relative humidity (RH, %) obtained with the two HOBO data loggers shielded within a gill-type shelter (multi-plate prototype plastic—Hgill1 and Hgill2). Values followed by the same letter did not differ significantly on the t test at a significance level of 5 %. JD, Julian day

and relative humidity. This suggests that the sensor was Vaisala probe was properly calibrated, given that the copper-constantan thermocouple psychrometer is a device most commonly used for determining the water vapor content of the atmosphere (Fritschen and Gay 1979; White and Ross 1991), because it provides acceptable precision and accuracy when it is made of copper-constantan (Cunha et al. 2001; Marin et al. 2001) and is considered the standard for the measurement of temperature and relative humidity (WMO 2008), being recommended for use as a benchmark in the calibration of other types of sensors (Fig. 5).

Fig. 5 Measurements of air temperature (T, °C) and relative humidity (RH, %) obtained with the copper-constantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight (Psy) and with the temperature/relative humidity probe under a commercial, multi-plate radiation shield (Vais). Values followed by the same letter did not differ significantly on the t test at a significance level of 5 %. JD, Julian day

50

100

45

90

40

80

35

70

30

60

25

50

20 15

40 JD84

JD85

JD88

JD89

JD90

JD91

T Vais = 22.57 ºC (a) T Psy = 22.68 ºC (a) RH Vais = 77.88% (A) RH Psy = 78.16% (A)

10 5

RH (%)


Page 6 of 11

T (ºC)

236

30 20 10 0

13:0 0 18:0 0 23:0 0 04:0 0 09:0 0 14:0 0 10:0 0 15:0 0 20:0 0 01:0 0 06:0 0 11:0 0 16:0 0 21:0 0 02:0 0 07:0 0 12:0 0 17:0 0 22:0 0 03:0 0 08:0 0 13:0 0

0

Time of day

As can be seen in Figs. 6 and 7, specifically with respect to the HOBO data loggers, there were different amplitudes of temperature and relative humidity under the different types of shelters, the greatest amplitudes being for the Hsun and Hcup combinations. This demonstrates the inefficiency of the white plastic cup as a shelter. The use of an inappropriate shelter provides a greater thermal gradient, creating atmospheric destabilization at the time of measurement, thereby skewing the measurement. This is because the incident solar radiation, as well as that reflected by clouds, the ground and surrounding objects, 40 38

JD84

JD85

JD88

JD89

JD90

JD91

36 34 32 30 28 26 24 22 20 18 16 14 12

T Hsun T Hcup T Hgill T Vais T Psy

10 13:0 0 18:0 0 23:0 0 04:0 0 09:0 0 14:0 0 10:0 0 15:0 0 20:0 0 01:0 0 06:0 0 11:0 0 16:0 0 21:0 0 02:0 0 07:0 0 12:0 0 17:0 0 22:0 0 03:0 0 08:0 0 13:0 0

T (ºC)

Fig. 6 Measurements of air temperature (T, °C) obtained with the device-condition combinations Hgill (HOBO data logger shielded within a gill-type shelter (multi-plate prototype plastic); Hsun (HOBO data logger in full sunlight); Hcup (HOBO data logger shielded within a white plastic cup with windows for air circulation); Psy (copperconstantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight); and Vais (temperature/ relative humidity probe under a commercial, multi-plate radiation shield). JD, Julian day

alters the measurement of temperature by the sensor, because it absorbs some of this energy, which could result in an inaccurate temperature reading. Therefore, to ensure that the sensor measures the true temperature, it must be protected against these effects with an efficient, ventilated shelter that allows the free circulation of air around the sensor (Lin et al. 2004) and minimizes these environmental effects (Van der Meulen and Brandsma 2008). In terms of the measurements of temperature and relative humidity, there were no significant differences between the Hcup and Hsun combinations, although

Time of day

Environ Monit Assess (2015) 187:236 100

JD84

JD88

JD89

JD90

JD91

90 80 70 60

RH (%)

Fig. 7 Measurements of relative humidity (RH, %) obtained with the device-condition combinations Hgill (HOBO data logger shielded within a gill-type shelter (multi-plate prototype plastic); Hsun (HOBO data logger in full sunlight); Hcup (HOBO data logger shielded within a white plastic cup with windows for air circulation); Psy (copperconstantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight); and Vais (temperature/ relative humidity probe under a commercial, multi-plate radiation shield). JD, Julian day

Page 7 of 11 236

JD85

50 40 30

RH Hsun RH Hcup RH Hgill RH Vais RH Psy

20 10

13:0 0 18:0 0 23:0 0 04:0 0 09:0 0 14:0 0 10:0 0 15:0 0 20:0 0 01:0 0 06:0 0 11:0 0 16:0 0 21:0 0 02:0 0 07:0 0 12:0 0 17:0 0 22:0 0 03:0 0 08:0 0 13:0 0

0

Time of day

those combinations did differ significantly from the Hgill, Vais, and Psy combinations (Table 2). The Vais combination presented the lowest values of temperature, corroborating the findings of Oliveira (2007). When protected from direct sunlight and rain by an appropriate multi-plate radiation shield (i.e., when there is thermal inertia), the Vaisala probe is accurate to within 0.4 °C for temperature and to within 2–3 % for relative humidity (CSI 1990). We found that there was a significant (2 °C) difference between the temperatures measured with the Psy and Vais combinations and those measured with the Hsun and Hcup combinations. This is due to inefficiency and

lack of shelter, respectively, because the accuracy of the measurement of temperature, in this case, depends much more on the effect of solar radiation than on the characteristics of the devices or electromagnetic interference (Marin et al. 2001). According to Mendes (2008), the difference in temperature might be related to the exposure of the sensor, in particular as regards the characteristics of the shelter, given that an ineffective shelter can generate a temperature difference of up to 3 °C (WMO 2008). Significant differences were also found by Oliveira (2007), who used copper-constantan thermocouple psychrometers housed in four different types of shelters and observed a temperature difference of approximately 3 °C

Table 2 Temperature and relative humidity values obtained with the various device-condition combinations Variable

Device-condition combination Hsun Mean±CI SD

Hcup Mean±CI SD

Hgill Mean±CI SD

Vais Mean±CI SD

Psy Mean±CI SD

Temperature (°C)

24.7±0.5a,ba (5.6)

24.1±0.5b (4.9)

22.7±0.3c (3.1)

22.6±0.3c,d (2.8)

22.7±0.3c,d (3.2)

Relative humidity (%)

68.5±2.1Aa (23.1)

67.5±1.7A (18.0)

75.8±1.2B (12.9)

77.9±1.1C (12.1)

78.2±1.1C (11.7)

Hsun HOBO data logger in full sunlight, Hcup HOBO data logger shielded within a white plastic cup with windows for air circulation, Hgill HOBO data logger shielded within a gill-type shelter (multi-plate prototype plastic), Vais temperature/relative humidity probe under a commercial, multi-plate radiation shield, Psy copper-constantan thermocouple psychrometer exposed to natural ventilation and protected from sunlight a

Values followed by the same letter on the same row did not differ significantly on the t test at a significance level of 5 %

236


Page 8 of 11

between a sensor without no isolation or ventilation and an insulated sensor in a condition of forced ventilation. Using 25 HOBO data loggers, Mauder et al. (2008) found that temperature readings were up to 1.2 °C lower when a naturally ventilated multi-plate shelter was used. In the present study, the relative humidity obtained with the Hsun and Hcup combinations differed from that obtained with the Hgill combination by approximately 10 %, demonstrating the need for sheltered measurements of relative humidity (Table 2). Another noteworthy finding was that, because appropriately sheltered sensors are in environments that are more stable, the standard deviations also trended lower for measurements obtained from sheltered devices, indicating that the measurements of temperature and relative humidity were more accurate when obtained under such conditions. One must be aware of the type of shelter used because, according to Oliveira (2007), shelters whose construction differs from that of the multi-plate shelter used in the Vais combination can heat up and transfer some of that heat to the sensor, resulting in readings that are higher than the actual values, indicating that such shelters offer no protection against diurnal or nocturnal radiation. In addition, shelters that are not open on all sides limit natural ventilation and take longer to cool after nightfall. Because the shelter used in the Vais combination has overlapping plates, it provides better protection against solar radiation without limiting natural ventilation, putting it in compliance with the recommendations of the WMO (2008). As can be seen in Figs. 6 and 7, the equipment is subjected to extremes

of radiative heating (during the day) and radiative cooling (at night); hence, the separate analyses of diurnal and nocturnal data. Because the copper-constantan thermocouple psychrometer is considered the standard for the measurement of temperature and relative humidity (WMO 2008), we compared the diurnal and nocturnal data obtained with the Psy combination and used statistical indicators to quantify the significance of the differences. Tables 3 and 4 show the statistical indicators for the comparisons among the measurements of temperature and relative humidity obtained with the various devicecondition combinations. As can be seen, the Vais combination showed the best concordance (coefficient of determination and index of agreement), lowest mean absolute error and lowest maximum error, followed by the Hgill combination, indicating that those two combinations showed the greatest similarity with the Psy combination. This shows that the precision and accuracy of measurements obtained with an instrument depends not only on the sensitivity of the instrument but also on its appropriate exposure. Due to the effect of solar radiation, the mean absolute error and maximum error were larger for the diurnal data (Table 3). However, they were within the allowable 0.3 °C margin for temperature at a working range of −40 to 40 °C and the 1–3 % margin for relative humidity at a working range of 0 to 100 %, as described by Santana et al. (2008). In many studies, such as those conducted by Soylu and Çömlekçioğlu (2009) and Loheide and Gorelick

Table 3 Correlations of the temperature and relative humidity measurements obtained with the various device-condition combinations in relation to those obtained with the copper-constantan

thermocouple psychrometer exposed to natural ventilation and protected from sunlight (Psy). Diurnal data. Botucatu, Brazil

Combination

Mean

R2

d

T_Hsun

28.4

0.462

0.991

T_Hcup

27.2

0.611

0.995

T_Hgill

24.5

0.838

T_Vais

24.0

RH_Hsun

MAE

ME

y=bx

Estimate (%)

4.08

11.69

y=1.164x

↓ 16.4

2.92

8.82

y=1.118x

↓ 11.8

0.999

0.31

3.18

y=0.984x

↑ 1.6

0.886

0.999

0.16

1.95

y=1.003x

↓ 0.3

53.0

0.646

0.961

19.17

49.01

y=0.746x

↑ 25.4

RH_Hcup

55.8

0.757

0.977

16.34

38.71

y=0.781x

↑ 21.9

RH_Hgill

69.2

0.994

0.999

3.02

4.18

y=0.960x

↑ 4.0

RH_Vais

71.8

0.994

0.999

0.35

3.36

y=0.995x

↑ 0.5

R2 coefficient of determination, d index of agreement, MAE maximum absolute error, ME mean absolute error, T temperature, RH relative humidity, Hsun HOBO data logger in full sunlight, Hcup HOBO data logger shielded within a white plastic cup with windows for air circulation, Hgill HOBO data logger shielded within a gill-type shelter (multi-plate prototype plastic), Vais temperature/relative humidity probe under a commercial, multi-plate radiation shield


Page 9 of 11 236

Table 4 Correlations of the temperature and relative humidity measurements obtained with the various device-condition combinations in relation to those obtained with the copper-constantan

thermocouple psychrometer exposed to natural ventilation and protected from sunlight (Psy). Nocturnal data. Botucatu, Brazil

Combination

Mean

R2

d

MAE

ME

y=bx

Estimate (%)

T_Hsun

19.5

0.991

0.999

0.88

1.47

y=0.957x

↑ 4.3

T_Hcup

19.6

0.986

0.999

0.73

1.33

y=0.964x

↑ 3.6

T_Hgill

20.2

0.991

0.999

0.18

0.56

y=1.009x

↓ 0.9

T_Vais

20.5

0.998

0.999

0.12

0.38

y=0.994x

↑ 0.6

RH_Hsun

90.7

0.959

0.999

3.98

7.59

y=1.045x

↓ 4.5

RH_Hcup

84.2

0.969

0.999

2.52

6.01

y=0.970x

↑ 3.0

RH_Hgill

85.2

0.972

0.999

1.50

3.50

y=0.983x

↑ 1.7

RH_Vais

86.5

0.989

0.999

0.19

2.33

y=0.998x

↑ 0.2

R2 coefficient of determination, d index of agreement, MAE maximum absolute error, ME mean absolute error, T temperature, RH relative humidity, Hsun HOBO data logger in full sunlight, Hcup HOBO data logger shielded within a white plastic cup with windows for air circulation, Hgill HOBO data logger shielded within a gill-type shelter (multi-plate prototype plastic), Vais temperature/relative humidity probe under a commercial, multi-plate radiation shield

(2006), HOBO data loggers have been used for measuring for temperature and relative humidity without detailing whether or not a shelter was employed, which can invalidate the measurements obtained. The HOBO data logger (HOBO 2002) is a good, affordable alternative for small-scale users (Palmieiri 2009). However, it is not accurate for measurements of temperature and relative humidity; nor is it the standard equipment for comparisons with other sensors. One must be careful using this type of equipment. Camerini et al. (2011) used the HOBO data logger as a reference to evaluate other low-cost alternative instruments and found major differences in the relative humidity readings, which were due to inadequate wetting of the muslin covering the wet bulb and not to the influence of air pressure, as the authors declared. Wong et al. (2007) measured temperature and relative humidity using HOBO data loggers, each of which was inserted into a white wooden box with ventilation holes on both sides. After comparing the measurements obtained under these conditions with those obtained through the use a HOBO data logger sheltered according to WMO standards, the authors found the former type of shelter to be inappropriate. In addition, the results obtained by Hien et al. (2007), who used eight HOBO data loggers housed under those same conditions, were certainly compromised. Because HOBO data loggers use internal sensors to measure temperature and relative humidity, they cannot be directly exposed to atmospheric conditions. Therefore, it is not recommended that HOBO data loggers be used in the way in which they are presented, because, without

adequate shelter, they provide measurements of temperature and relative humidity quite different from those obtained with standard sensors, such as the Vaisala probe and copper-constantan thermocouple psychrometer, which are the instruments that are the most widely used in agrometeorological studies. Therefore, care must be taken to use each type of device or sensor appropriately and for the intended application, so as not to incur serious errors or draw erroneous conclusions. Mauder et al. (2008) also observed measurement errors with HOBO data loggers and recommended its use sheltered according to WMO (2008), in order to avoid the erroneous analysis of the obtained data.

Conclusion The HOBO data loggers are inaccurate when measurements are made without shelter or under shelters that do not meet the WMO criteria. The gill-type (multi-plate prototype plastic) shelter used in the present study is recommended, because it effectively protected the equipment from the effects of solar radiation.

References Azevedo, T. R. (2001). Derivation anthropogenic climate in the metropolitan region of São Paulo addressed as a function of the weekly rhythm of human activities. Thesis in Physical

236

Page 10 of 11

Geography, Faculdade de Filosofia, Letras e Ciências Humanas, Universidade de São Paulo, São Paulo, Brazil. 474 pp. Brandsma, T., & Van Der Meulen, J.P. (2008). Thermometer screen intercomparison in De Bilt (the Netherlands) – Part II: Description and modeling of mean temperature differences and extremes. International Journal of Climatology, 28, 389–400. Camerini, N. L., Mendes, L. B., Mota, J. K. M., Nascimento, J. W. B., & Furtado, D. A. (2011). Evaluation of alternative agrometeorological instruments for monitoring poultry shed environment. Engenharia na Agricultura, 19, 125–131. Campbell, G. S. (1990). Tanner symposium on biophysical measurements and instrumentation. Theoretical and Applied Climatology, 42, 201–202. CSI Campbell Scientific Inc. (1990) Model HMP45C temperature and relative humidity probe: instruction manual. Revision: 04/99. Logan, Utah: Copyright©, 1990–1999. Cunha, A. R., & Martins, D. (2004). Comparative study among meteorological elements obtained in conventional and automated meteorological stations in Botucatu, SP, Brazil. Revista Brasileria de Agrometeorologia, 12, 103–111. Cunha, A. R., Escobedo, J. F., & Galvani, E. (2001). Evaluation of a thermocouple psychrometer of low cost. Revista Brasileria de Agrometeorologia, 9, 17–22. Danni-Oliveira, I. M. (2002). Thermometer inspection conducts to the empirical studies in geographical climatology. RA’E GA, 6, 75–80. Dunham, J., Chandler, G., Rieman, B., & Martin, D. (2005). Measuring stream temperature with digital data loggers: a user’s guide. USDA Forest Service Gen. Tech. Rep. RMRSGTR-150WWW. 15pp. Fritschen, L. J., & Gay, L. W. (1979). Environmental instrumentation. New York: Springer. 212pp. Hien, W. N., Yok, T. P., & Yu, C. (2007). Study of thermal performance of extensive rooftop greenery systems in the tropical climate. Building and Environment, 42, 25–54. HOBO. (2002). BoxCar Pro 4.3 User’s Guide. USA: Copyright© On Set Computer Corporation, 76pp. Hubbard, K. G., Lin, X., & Walter-Shea, E. A. (2001). The effectiveness of the ASOS, MMTS, Gill, and CRS air temperature radiation shields. Journal of Atmospheric and Oceanic Technology, 18, 851–864. Huwald, H., Higgins, C. W., Boldi, M. O., Bou-Zeid, E., Lehning, M., & Parlange, M. B. (2009). Albedo effect on radiative errors in air temperature measurements. Water Resources Research, 45, 1–13. Lin, X., Hubbard, K. G., & Meyer, G. E. (2001). Airflow characteristics of commonly used temperature radiation shelters. Journal of Atmospheric and Oceanic Technology, 18, 329– 339. Lin, X., Hubbard, K. G., & Meyer, G. E. (2004). Sensor and electronic biases/errors in air temperature measurements in common weather networks. Journal of Atmospheric and Oceanic Technology, 21, 1025–1032. Loheide, S. P., & Gorelick, S. M. (2006). Quantifying streamaquifer interactions through the analysis of remotely sensed thermographic profiles and in situ temperature histories. Environmental Science and Technology, 40, 3336–3341. Marin, F. R., Angelocci, L. R., Coelho Filho, M. A., & Villa Nova, N. A. (2001). Construction and evaluation of an aspirated thermocouple psychrometer. Scientia Agricola, 58, 839–844.

Environ Monit Assess (2015) 187:236 Mauder, M., Desjardins, R. L., Gao, Z., & Van Haarlem, R. (2008). Errors of naturally ventilated air temperature measurements in a spatial observation network. Journal of Atmospheric and Oceanic Technology, 25, 2145–2151. Mendes, M. M. T. (2008). 10 Anos de estações automáticas: perspectivas para a caracterização do clima em Portugal. Licenciado em Ciências Geofísicas - Dissertação de Mestrado em Ciências Geofísicas - Mestrado em Ciências Geofísicas (Especialização em Meteorologia). Faculdade de Ciências – Universidade de Lisboa. 91pp. http://repositorio. ul.pt/bitstream/10451/1315/1/19581_ulfc091372_tm_ TeseMestradoMM.pdf Nakamura, R., & Mahrt, L. (2005). Air temperature measurement errors in naturally ventilated radiation shelters. Journal of Atmospheric and Oceanic Technology, 22, 1046–1058. Oliveira, M. J. (2007). Avaliação experimental de interferências causadas pela radiação solar global e pela velocidade dos ventos em medições climatológicas da temperatura do ar. 127 pp. Monografia. Escola de Engenharia de São Carlos, Universidade de São Paulo, São Carlos. http://www.tcc.sc. usp.br/tce/disponíveis/18/180300/tce-12072010-111433/ publico/Oliveira_Marcos_Jose_de.pdf Palmieiri, A. M. (2009). Desenvolvimento de sistema automatizado de baixo custo para coleta e armazenamento de dados e variáveis climáticas: aplicações no ambiente agrícola. 100p. Dissertação (Mestrado em Física do Ambiente Agrícola) - Escola Superior de Agricultura BLuiz de Queiroz^, Universidade de São Paulo, Piracicaba. http:// www.teses.usp.br/teses/disponiveis/11/11131/tde-18022010142019/pt-br.php Santana, M. A. A., Guimarães, P. LO, Thomaz, J. C. Jr, & Arlino, P. R. A. (2008). Rastreabilidade metrológica e os critérios de aceitação para a equipamentação. METEOROLÓGICA/ AMBIENTAL: ENQUALAB-2008 - Congresso da Qualidade em Metrologia Rede Metrológica do Estado de São Paulo – REMESP, São Paulo, Brasil. 6pp. Soylu, M. K., & Çömlekçioğlu, N. (2009). The effects of high temperature on pollen grain characteristics in tomato (Lycopersicon esculentum M.). Journal of Agriculture Faculty HR U, 13, 35–42. Spetalen, A., Lofseik, C., Nordli, P. (2000). A comparison of air temperature radiation screens by fields experiments and Computational Fluid Dynamics (CFD) simulations. In: WMO Technical Conference on Meteorological Instruments and Methods of Observation, Beijing, China, 23–27. Instruments and Observing Methods N.74, WMO, Geneva, 54–57. Sun, B., & Baker, C. B. (2004). A comparative study of ASOS and CRN temperature measurements. In: Eighth Symposium on Integrated Observing and Assimilation Systems for Atmosphere, Oceans and Land Surface. 84th AMS Annual Meeting, Seattle. Van Der Meulen, J. P., & Brandsma, T. (2008). Thermometer screen intercomparison in De Bilt (The Netherlands), Part I: understanding the weather-dependent temperature differences). International Journal of Climatology, 28, 371–387. White, G. M., & Ross, I. R. (1991). Humidity. In Z. A. Henry, G. A. Zoerb, & G. S. Birth (Eds.), Instrumentation and measurement for environmental sciences (pp. 8.01–8.13). Saint Joseph: ASAE. Whiteman, C. D., Hubbe, J. M., & Shaw, W. J. (2000). Evaluation of an inexpensive temperature datalogger for meteorological

Environ Monit Assess (2015) 187:236 applications. Journal of Atmospheric and Oceanic Technology, 17, 77–81. Willmott, C. J., Ackleson, S. G., Davis, R. E., Feddema, J. J., Klink, K. M., Legates, D. R., O’donnell, J., & Rowe, C. M. (1985). Statistics for the evaluation and comparison of models. Journal of Geophysical Research, 90, 8995–9005. WMO. (2008). World Meteorological Organization. Guide to meteorological instruments and methods of observation,

Page 11 of 11 236 seventh edition, 2008. WMO n.8. http://www.wmo.int/ pages/prog/www/IMOP/publications/ CIMO_Guide/ CIMO%20Guide%207th%20Edition,%202008/CIMO_ Guide-7th_Edition-2008.pdf Wong, N. H., Tan, P. Y., & Chen, Y. (2007). Study of thermal performance of extensive rooftop greenery systems in the tropical climate. Building and Environment, 42, 25–54.