 copyright Eumetsat - Kipp & Zonnen |
Ground stations (GRD) measurements versus HelioClim satellite (SAT) estimations: Important Facts |
A problem of units? of terminology? Consult the Education service. |
Sources of differences
GRD: Punctual measurements / SAT: integrated over the pixel: the values delivered by each system mainly differ due to the sensor and due to the way of collection of the data itself. The GRD measurements are very localized in space, whereas the SAT integrates the radiation over its pixels. Consequence:
Ground station data: strenghts and weaknesses
| Characteristics | Ground Stations - measured radiation values | Satellite - estimated radiation values |
|---|---|---|
| Spectral band of acquisition | - pyranometer (thermal => sensible to all the spectra of the sun): 0 - 3 µm - solar panel: |
Two images, 600 and 800 nm, calibrated to fit the GR spectral range |
| Instant of acquisition | real integration over Δt | punctual in time! 8 µs (micro second, to be checked) are needed to scan each line => almost instantaneous |
| Space | very localized in space | spatial integration on the area covered by the pixel => average radiation on each pixel |
| Horizon | affected by the shadowing effect of the near and far horizon | No, except in clear sky conditions, large orographic structures can partly affect and decrease the radiation value of the pixel |
| Particular case of DNI | angle of acceptance of the pyranometer: tube to collect only the sun, but depend on the solid angle | Error of the model of estimation of the radiation over the horizontal plane form Meteosat images, combined with the error of the model of decomposition of the global radiation into is diffuse and direct component as well as the computation over the tilted plane. |
| sun-tracking system that hosts the irradiance sensors: the astronomical data are insufficient track the sun, because of the atmospheric refraction. This refraction depends on the atmospheric content, and thus evolves with time. As a consequence, the sun position also evolves with time. That is why a sun tracker is essential to correct the uncertainty of the sun position. As an example, if the sky is cloudy, the sun tracker will approximately aim at maximum of the halo of the sun. As soon as the sun gets uncovered, the tracker correct its position to face the sun. But during all these phases of tracking the sensor goes on integrated the radiation values. | ||
| Limit in the sensor course: it is possible that the sensor of the panels cannot reach all the position where is the sun (ex: vertical position) | ||
| Maintenance | Dirt. Not homogeneous on the globe, depending on the frequency, the people of the maintenance. | from ground => leads to Missing Data |
| Breakdowns | ||
| Temporal drift | human error | No |
Philosophy
The Bias: this statistical quantity corresponds to the mean of the difference between two series of data. Most of the time, it is the mean of the estimated values minus the reference values, divided by the mean of the reference values. The obtained value can be divided by 100 in order to express the stats in percents.
| Bias (%) = | mean(V2 - v1) mean(v1) |
* 100 | (1) |
We would like to focus your attention on a point: the bias computed between the HC3 and the ground station data depend on the station, i.e. with the same HC3 data would lead to as many different bias as the number the ground stations. As a consequence, as the mean is a universal quantity, should we still talk about "bias"?
NOTES:
some wrong values due to limited malfunctions of the trackers that host the irradiance sensors
Pyranometer: no need to apply a corrective term bewteen the HelioClim data and the pyranometer measurement since it is a thermopile, i.e. a thermometer. As it is functionning with heat, it is sensitive to all the incident sun spectra, like with HelioSat-2. All the sensitivity of the spectra must be considered.
On the contrary, measurements with photodiodes or PV panels, a corrective term needs to be applied to calibrate the pyranometer. In the particular case of a measurement with a pv panel, several elements need to be taken into account: the network connections, the band spectral of reaction of the panel, the type of jonction (1, 3 => more reaction of the panel...), and finally the type (quality) of panels.
email miroslav, 22-08-11: Dear Miroslav, The pyranometer (thermopile) measures in a wide range spectrum, from 285 to 2800 nm, which is much wider than the solar cell. Our Helioclim database is supposed to give the irradiation values for the same wavelength range as the pyranometer, thus the irradiation values should be comparable. Your measurements are peaking to 1264 W/m2. This is much higher than the maximum GHI value at the top of atmosphere of 1066 W/m2 at your point of interest for this day. The GHI at TOA is computed from the solar radiation constant and the sun position, and this is very precise. It seems that there is an issue with the pyranometer measurements, maybe a non correct calibration. The signal giving the irradiation value is proportional to the pyranometer sensitivity which can vary from 7 to 14 µV/W/m2. Thus, without proper calibration, the same measured voltage can be interpreted as more or less irradiation. As we are not privy to the installation details, all this reasoning is sort of a “guessing game”, based on the irradiation measured, the Helioclim typical irradiation levels and the technical reference manuals of the CMP11 Kipp&Zonen. As a conclusion, we are quite sure that the measured level of radiations is not plausible for this site.
Last update: 2/23/12 5:20 PM