Determination of Longwave Response of Shortwave Solar

Radiometers to Correct for their Thermal Offset Errors

1National Renewable Energy Laboratory Golden CO2The Eppley Laboratory Newport RI3Pacific Northwest Laboratory Richland WA4National Oceanic and Atmospheric Administration, Boulder CO

Based on an article that will be printed in the Oct. issue (Vol. 22, No. 10)in

Journal of Atmospheric and Oceanic Technology (JTECH)by

I. Reda1, J. Hickey2, D. Myers1, T. Stoffel1, S. Wilcox1, C. Long3,E. G. Dutton4, D. Nelson4, and J. J. Michalsky4

Outline

Introduction– Pyranometer Thermal Offset evidence– Thermopile Detector Configurations

• How it is determined and show results for 14 pyranometers

• Explain how to correct for the thermal offset errors

• Explain effect of thermal offset error on two calibration methods and show correction results

• Conclusions and acknowledgements

Eppley Model PSP & 8-48 Pyranometers

Broadband Shortwave: Clear Sky

Night ≤ 0?

Really?

Shaded PSP Zero-Offsets

Nighttime Offsets

Additional Evidence…

ARM Program scientists (e.g., Dr. Robert Cess)

found several examples of clear-sky diffuse

measurements that were less than predicted

for pure atmospheric Rayleigh scattering:

Expected Minimum Diffuse= 100 Wm-2

Measured Diffuse = 90 Wm-2

Still More Evidence…

Comparison of calibration results:

Manufacturer’s pyranometer responsivity (Rs) typically

1.5% - 2% HIGHER than

Outdoor calibration results at NREL.

Rs(indoor) > Rs(outdoor)

Heat Budget of Generic Pyranometer with Single Black Detector

SW tg(1-) + FgdAggTg4 - FdgAddTd

4 - (Td - Ts)k ± Con = 0 xx

SW = shortwave solar t = transmission of glass dome(s) = SW reflectanceF= config factor (area=A) T = temperature (°K) = emissivityk = heat conduction coeff. Con = convection &unwanted conduction g = glass dome

d = detector surface V = pyranometer output voltage (V) s = thermopile heat sink

a b c d e

SW

The Key Concept

Gordon Conference June 14-19, 1998

Solar Radiation & Climate

John Hickey describes concept of characterizing a Model PSP pyranometer in a Blackbody

Pyrgeometer Calibration System

currently under development for ARM…

ARM Radiometers: LongwaveCalibration Traceability

ProductionPrecision

Infrared RadiometerModel PIR

ProductionBlackbodyChamber

EPLABR&D, Production

Reference Standards

SIRS/SKYRAD/GNDRADPyrgeometers

OutdoorCharacterizations

Transfer StandardsModel

PIR & CG4

NREL/ARMPyrgeometer

BlackbodyCalibration System

NRELMetrology LaboratoryReference Standards

National Institute of Standards & TechnologyPrimary References

Electrical, Temperature, MassDimensional, etc.

Calibrated 14 pyranometers in NREL-Pyrgeometer Blackbody Calibration System:

EPLAB PSP & 8-48 (BW)Kipp & Zonen CM22SpectroSun SR-75

to calculate their Indoor NET-IR Responsivity (RSnet)

The Approach

To calculate the thermal offset error of pyranometerswhen calibrated or deployed outdoors

Error: - 0% to 2% error in Responsivity (Rs) from outdoor calibrations - 0 W/m2 to 20 W/m2 error in measured solar irradiance outdoors

Then used RSnet to correct for the thermal offset error…

Simplified Diagram for NREL Blackbodywith a Pyranometer

RTD

Circulating oil(-40°C to 50°C)

Aluminum Plate withCirculating Silicon oil

(-40°C to 50°C)

Pyranometer

Pyranometer Dome

RH = 0% to10%

Blackbody Hemisphere(Blackened on the inside)

RH = 0% to10% BB Completion Hemisphere

(Gold plated)

Thermistor

Blackbody Temperature Plateaus

Blackbody Temperature

( °C)

Case Temperature

( °C)TBB - TCase

( °C)

-35 -5 -30

-20 -5 -15

-20 10 -30

-5 10 -15

10 25 -15

Simulate the possible sky and ambient temperatures at NREL

BW- 21096

-150

0

-150 0

BW- 33253

-150

0

-150 0

BW- 33237

-150

0

-150 0

CM22- 000019

-150

0

-150 0

PSP-14612F3

-350

0

-150 0

PSP- 25825F3

-350

0

-150 0

PSP- 28403F3

-350

0

-150 0

PSP- 31146F3

-350

0

-150 0

PSP- 31147F3

-350

0

-150 0

PSP- 31148F3

-350

0

-150 0

PSP- 31155F3

-350

0

-150 0

PSP- 31156F3

-350

0

-150 0

PSP- 31157F3

-350

0

-150 0

SS- 73- 93

-350

0

-150 0

Plots of the thermopile output versus the net longwave radiation for 14 pyranometers

WNET (W/m2)T

hermopile O

utput (uV

)

The Net IR (WNET)= [(Tblackbody)4 - (Tcase)4]WNET is Negative → TP output is Negative

Black Body Calibration Results for Pyranometers in April, 2004

Model-Serial Number RSbb [uV/(W/m2)] R2

BW-21096 0.363 0.9109

BW-33253 0.8894 0.8834

BW-33273 0.7734 0.9872

CM22-000019 0.8872 0.972

PSP-14612F3 2.6591 0.9975

PSP-25825F3 2.5076 0.9951

** PSP-28403F3 2.3824 0.9992

PSP-31146F3 2.1232 0.9946

PSP-31147F3 2.1278 0.9988

PSP-31148F3 2.1707 0.9992

PSP-31155F3 2.2002 0.9966

PSP-31156F3 2.188 0.9974

PSP-31157F3 2.2046 0.998

SS-73-93 1.8151 0.9962

** has case thermistor

Shortwave/NET-IR Equivalence

where,

V = thermopile output voltage during BB-Cal (uV) RSbb = Longwave (blackbody) responsivity (uV/W/m-2) RSmfr = Shortwave (manufacturer) responsivity (uV/W/m-2)

E = V / RSmfr

V / RSbb

= RSbb

RSmfr

Model-Serial Number RSbb (uV/W/m^2) R^2

BW-21096 0.363 0.9109

BW-33253 0.8894 0.8834

BW-33273 0.7734 0.9872

CM22-000019 0.8872 0.972

PSP-14612F3 2.6591 0.9975

PSP-25825F3 2.5076 0.9951

** PSP-28403F3 2.3824 0.9992

PSP-31146F3 2.1232 0.9946

PSP-31147F3 2.1278 0.9988

PSP-31148F3 2.1707 0.9992

PSP-31155F3 2.2002 0.9966

PSP-31156F3 2.188 0.9974

PSP-31157F3 2.2046 0.998

SS-73-93 1.8151 0.9962

SW/Net Equiv. (E)

RSnet =

RSbb * ERSmfr

0.033 0.012 8.97

0.097 0.087 8.22

0.079 0.061 12.05

0.096 0.085 9.3

0.267 0.710 9.93

0.256 0.642 8.69

0.265 0.630 8.37

0.262 0.557 9.75

0.258 0.548 8.13

0.263 0.572 8.47

0.261 0.574 8.56

0.262 0.573 9.8

0.258 0.568 8.24

0.209 0.379 9.13

Blackbody Calibration Results for Pyranometers in April, 2004

Thermal Offset Voltage

RsSW-Corr = VTP - ∆Vcorr / (Cavity * Cos(Z) + Diffuse)

where,

∆ Vcorr = WNET * RSbb * E = WNET * RSNET

WNET is measured using a collocated pyrgeometerWNET = TPpyrg * K1

K1 = pyrgeometer thermopile sensitivity

Two pyranometer calibration methods: Summation and Shade/Unshade

8.3

8.4

8.5

8.6

8.7

8.8

16:00:00 17:12:00 18:24:00 19:36:00 20:48:00 22:00:00

UTC

Res

pons

ivity

(uV

/W/m

2 )

28403F3 from sh/unsh

28403F3 from summ

Calibrated 28403F3 on 3/30/2004

RSsumm ~ 1.9%↓

Thermal Offset Effect on Calibration Methods

Summation (Component Sum) is effected by Net IR:

RS = U

B * Cos z + D

+ ∆u

Where:- RS = Responsivity [uV/(W/m2)]- U = Unshaded thermopile output (uV)- B = Beam irradiance, measured by a cavity (W/m2)- z = Zenith angle (°)- D = Diffuse irradiance (W/m2).

- ∆u = unshaded thermal offset error signal (uV).Negative for clear sky conditions.

Thermal Offset Effect on Calibration Methods

• Shade/Unshade is NOT effected by Net IR:

RS = U

B * Cos z

+ (∆u)

Where:- S = Shaded thermopile output (uV)

- S[ + (∆s)]

Assuming the Net IR is stable during the shade/unshade period,then ∆u = ∆s.

= U

B * Cos z

- S

- ∆s = shaded thermal offset error signal (uV).

Why Not Use the Shade/Unshade?

We calibrate up to 100 pyranometers at each calibration session, from sunrise to sundown

The shade/unshade will require more trackers with shading disks which will increase cost, labor,

maintenance, space, etc...Compared to

The component sum, one tracker for the diffuse reference, and all pyranometers are installed on

horizontal tables.

Correcting Summation Results for Thermal Offsets Brings Agreement with

Shade/Unshade Method

8.3

8.4

8.5

8.6

8.7

8.8

16:00:00 17:12:00 18:24:00 19:36:00 20:48:00 22:00:00

UTC

Re

spo

nsi

vity

(u

V/W

/m2)

RS-summ-corr.

28403F3 fromsh/unsh

28403F3 from summ

Summation Calibration with Correction on 4/17/2004

7

7.5

8

8.5

9

25 30 35 40 45 50 55 60 65 70 75 80 85 90

Z

RS 28403F3

28403F3corr

Results: Shade/Unshade (SU), Uncorrected Component Sum (UCS), and

Corrected Component Sum (CCS)

Model-S/Nz

( ° )Net-IR(W/m22)

RSSUSU RSUCSUCS RSCCSCCS%Error

RSSUSU - RSUCSUCS %Error

RSSUSU - RSCCSCCS

B&W-21096 36.1 -177.9 11.500 11.552 11.554 -0.45 -0.47

B&W-33253 36.1 -177.9 8.736 8.758 8.777 -0.25 -0.46

B&W-33273 36.1 -177.9 9.379 9.381 9.394 -0.01 -0.16

% RMSE for the B&W sample 0.09 0.15

CM22-000019 31.2 -168.1 9.350 9.311 9.326 0.42 0.25

PSP-28403F3 36.1 -177.9 8.722 8.589 8.725 1.52 -0.04

PSP-31146F3 31.2 -168.1 7.975 7.861 7.966 1.43 0.11

PSP-31147F3 36.1 -177.9 7.920 7.807 7.927 1.42 -0.09

PSP-31148F3 31.2 -168.1 7.971 7.860 7.970 1.39 0.01

PSP-31155F3 31.2 -168.1 8.203 8.095 8.205 1.31 -0.03

PSP-31156F3 31.2 -168.1 8.102 7.956 8.066 1.81 0.45

PSP-31157F3 31.2 -168.1 8.169 8.060 8.171 1.33 -0.03

% RMSE for the PSPs sample 1.47 0.18

Conclusions & Recommendations- We developed a method to characterize/correct most of the thermal

offset error of all-black thermopile pyranometers.- The method resulted in reducing the difference between the

component sum and shade/unshade method from 1.47% to 0.18% for model PSP, and 0.42% to 0.25% for CM22. The 8-48 difference increased from 0.09% to 0.15%, but this is negligible.

- Small differences between corrected component and shade/unshade responsivities could be attributed to calculating the NET-IR as the difference between the sky and pyranometer case radiation, rather than the difference between the sky and detector surface radiation.

! needs further research !

! More accurate WNET will result in a minor thermal offset correction for the shade/unshade method !

- Routine field measurements must be corrected for coincident thermal offset errors (actual WNET) regardless of the calibration method.

Acknowledgements

We thank Peter Gotseff for his patience and expertise in the blackbody and outdoor calibrations, and Bev Kay for her continuous administrative support.

Research funds for this work were provided by the Atmospheric Radiation Measurement (ARM) Program, NREL-National Center for Photovoltaics, and NREL-Metrology Laboratory.