Sunlight sensors

Abstract

A measurement is taken using a sunshine sensor comprising at least two light-sensitive detectors 13a-13g and a masking element having a pattern of opaque areas 17 so that at any time at least one dete

Description

This invention relates to methods of correcting measurements from sunlight sensors and of taking measurements with sunlight sensors, and to sunlight sensors. For many years, “hours of sunshine” has been part of the range of meteorological information recorded at monitoring stations throughout the world. Historically, the most widely used meteorological recorder was the Campbell-Stokes recorder which uses a spherical lens to focus the sun's rays to a point to burn a track on a strip of recording paper. Use of a Campbell-Stokes recorder is labour-intensive because the recording paper must be replaced manually each day. Such measurements have become increasingly important with the rise of solar power, where it is important to know the insolation at a location before, for example, deciding whether it is economically advantageous to install a solar photovoltaic installation at that location. More recently, it has become advantageous to know the split between the two components of the total radiation incident on a surface, being the Direct and Diffuse parts. This information is necessary for understanding the energy available on tilted surfaces such as solar panels, or on systems which focus the direct solar beam such as concentrating solar power systems. The Direct part is the radiation which comes directly from the sun, and which is therefore highly directional, and the direction of which varies throughout the day. The Diffuse part is the radiation which is scattered by clouds or other atmospheric components, and therefore has a more uniform distribution across the sky. Weather recording has on the whole become more automated, with most measurements being made by automatic instruments which can, if need be, operate entirely unattended and send their data to a remote monitoring station. As a result of this, there is a disincentive to equip a weather station with a Campbell-Stokes recorder since doing so would negate many of the benefits achieved through automation of other measuring instruments. This has meant that there is a growing gap in the climate record for sunshine information as existing Campbell-Stokes recorders are discarded and not replaced. This is occurring at a time where accurate climatic knowledge is vital to enable monitoring of climate change, and during a period of growing interest in solar power, and optimisation of crop growth in agricultural sciences, all of which applications benefit greatly from sunshine data. There have been attempts to provide an alternative type of sunshine sensor. These sensors have, for the most part, operated by comparing the outputs of two detectors, one of which is shaded from direct sunshine and the other of which is in open sun. Shading of one of the detectors is usually done with a shade ring, which is set to obscure the track of the sun across the sky as seen by the shaded detector. Correct positioning of the shade ring varies seasonally as the solar declination varies by over 46° from June 21st to December 21st and is dependent upon many variables including the latitude and longitude of the sensor's location and proper alignment with true north. Moreover, improper positioning of the shade ring can give results which are incorrect, but which might not appear to be obviously wrong so that errors could go unnoticed. As a result, this type of instrument is not suited to unattended operation. Alternatively, a sun-tracking sensor can be used that follows the sun around the sky, or a rotating shadowband sensor can be used in which a “shadowband” casts a shadow over a pyranometer periodically (e.g. once per minute) from which can be determined the global (unshadowed) intensity, the diffuse (shadowed) intensity and the direct intensity (the difference between the global and diffuse intensities). The most accurate modern instruments work by using an automated solar tracker, which points continuously to the sun’s position, enabling measurement of the Direct component with a pyrheliometer, and using a small shading disk or ball to shade a pyranometer from the Direct solar beam, while allowing measurement of the Diffuse radiation from the rest of the sky. Such instruments are costly to purchase and maintain. In European Patent EP 1 012 633 B, there was described a simpler sunshine sensor, where at least two (typically seven) sunlight sensors are provided with a masking element which ensure that for any solar declination and any diurnal solar position at least one detector can be exposed to direct sunlight through a translucent area and at least one detector is shaded from direct sunlight by an opaque area wherein the masking element is such that in use the two detectors receive substantially equal amounts of diffuse sunlight. Whilst this has the advantage that it does not require as accurate set up as the alternative sensor described above with tracking shade disks, it has been found to be less accurate than such sensors - especially in situations where light clouds or high atmospheric aerosol concentrations create a high circumsolar brightness near to the solar disk. We are aware of attempts such as those described in the article “Recalibration of SPN1 pyranometers against pyrheliometer and its relevance for the evaluation of concentrating solar process heat plants” by Jana Mbllenkamp et al (Solar Energy, Volume 197, February 2020, Pages 344-358) which describes the calibration of the sensors described above against more accurate instruments, but even with these corrections, further accuracy improvements are desirable. According to a first aspect of the invention there is provided a method of correcting a measurement of the intensity of sunlight at a location and at a time in accordance with claim 1. This method allows for the correction of a measurement of sunlight intensity to be made; by applying the combination of all three correction functions based on the normalised values, the inventor has found that the accuracy of the measurement can be increased, in particular by correcting for the field of view of the sensor and for the spectral response of the sunlight sensors and in particular which in turn leads to inaccuracy in the measurements of diffuse intensity due to scattering in the atmosphere in differing atmospheric conditions (e.g. differing cloud conditions). It will be appreciated that the normalised value of the initial direct intensity is equal to the normalised value of the initial direct normal intensity relative to the top of the atmosphere solar irradiance normal to the solar beam. Typically, the scaling factor used to determine the initial diffuse intensity will be 2±0.5. Preferably, it will be larger than 2, so as to allow for noise in the measurement signals and non-uniform spectral response in the detectors. Typically, it will be 2.3256±0.25, 0.1, or 0.05. The method may comprise the step of applying a cap to the initial diffuse intensity, the cap being the initial global intensity. This corrects for small variations between the responses of the sensor, as the diffuse intensity cannot be larger than the global intensity. The scaling factor used to determine the initial direct normal irradiance may comprise 5 the cosine of the solar zenith angle. The normalised values of each of the initial diffuse intensity, the initial global intensity and the initial direct intensity, may be determined by dividing each intensity by the top of atmosphere irradiance horizontal component, which may be determined as a top of atmosphere irradiance multiplied by the cosine of the solar zenith angle. The skilled reader will appreciate that the determination the solar zenith angle is well known. It can be determined using a look up table based such as are based on any appropriate almanac (e.g. The Astronomical Almanacs published by either the UK Hydrographic Office or the US Naval Observatory) or calculated using a suitable algorithm such as that proposed by Reda, Ibrahim &Andreas, Afshin (2004) Solar position algorithm for solar radiation application. Solar Energy. 76. 577-589. The top of atmosphere irradiance can be determined using a lookup table for position and time or using a suitable algorithm such as that proposed in the Reda et al paper cited above. The correction functions can be applied after calibrating the sunlight sensor against a more accurate instrument, such as a secondary standard pyranometer, typically over a period of two to four weeks. The correction functions may be stored in a lookup table, and may be provided as follows: GCorr = Global * f(Kt) DCOrr = Diffuse * f(KDNl) * f(Kdf) where GCOrr is the corrected global intensity, Global is the initial global intensity, Kt is the normalised global intensity, Dcorr is the corrected diffuse intensity, Kdni is the normalised direct intensity, Kdf is the normalised diffuse intensity, and the three correction functions f may be as given below: Kx f(Kt) f(Kdf) f(KDNI) 0 1.21 1.3 1 0.05 1.11 1.126 1.060445 0.1 1.08 1.06 1.078749 0.15 1.05 1.042 1.092896 0.2 1.03 1.034 1.108647 0.25 1.01 1.018 1.124859 Kx f(Kt) f(Kdf) f(KDNI) 0.3 1 1.001 1.138952 0.35 1 0.989 1.154734 0.4 1 0.971 1.169591 0.45 1 0.958 1.176471 0.5 1 0.943 1.176471 0.55 1 0.937 1.168224 0.6 1 0.919 1.152074 0.65 1 0.91 1.111111 0.7 1 0.91 1.064963 0.75 1 0.91 1.030928 0.8 1 0.91 1.020408 0.85 1 0.91 1.010101 0.9 1 0.91 1 0.95 1 0.91 1 1 1 0.91 1 As such, the correction function dependent on the normalised initial global intensity may start positive at zero intensity and decrease to zero before remaining zero from a limit point up to a maximum value of the normalised initial global intensity of 1. The 5 correction function dependent on the normalised initial diffuse intensity may start at a positive value and decrease past zero to a lower bound from a limit point and stay at that lower bound up to a maximum value of the normalised initial global intensity of 1. The correction function dependent on the normalised direct intensity may start at zero at zero normalised direct intensity, increase to a peak with increasing normalised 10 direct intensity before reducing back to a minimum value of 0 at a limit point before remaining zero up to a maximum value of the normalised direct intensity of 1. The method may comprise applying a cap to the corrected diffuse intensity, so that it cannot exceed the corrected global intensity. 15 The corrected direct normal irradiance (DNI) may be determined by subtracting a constant term from the difference between the corrected diffuse intensity and the corrected global intensity. The constant term may correct for erroneously indicated DNI in overcast conditions. Furthermore, a lower bound, typically zero, can be applied to the corrected direct normal irradiance so as to prevent it falling into negative figures. Typically, the constant term would be 5 Wm'2 ± 5, 2, 1, 0.5, 0.2 or 0.1 Wm'2. The corrected direct normal irradiance may be calculated by applying a further correction function to correct for the response of the sensor when the sun is near the horizon. As such, the correction function may be determined by dividing by a correction function, which is 1 at solar zenith angles up to a limit and decreases from 1 at increasing solar zenith angles above the limit. However, this will depend on the particular sunlight sensors chosen, and it is possible that, instead, above the limit the correction function increases above 1, or increases to a peak above 1 and then drops below 1. In one embodiment, typical values of the correction function are as follows (with the solar zenith angle SZA in degrees and f(SZA) being the correction function): SZA f(SZA) SZA f(SZA) SZA f(SZA) 0 1 30 1 60 1 3 1 33 1 63 1 6 1 36 1 66 1 9 1 39 1 69 0.99 12 1 42 1 72 0.98 15 1 45 1 75 0.97 18 1 48 1 78 0.94 21 1 51 1 81 0.9 24 1 54 1 84 0.86 27 I 57 1 87 0.8 A final global intensity can then be calculated by adding the corrected direct normal irradiance, multiplied by the cosine of the solar zenith angle, to the corrected diffuse intensity. A system as described as above has been found to mirror the results of the gold standard pyranometer more accurately than the prior art instrument described in European Patent 1 012 633. The method may be applied to results that have previously been captured using the sensor at a previous time; as such, this method may improve the accuracy of previously-captured results. However, this method is equally suitable for processing results as they are captured. As such, according to a second aspect of the invention, there is provided a method of measuring the intensity of sunlight, the method comprising taking at least one measurement of sunlight intensity using sunlight sensor comprising at least two light sensitive detectors and a masking element, where the masking element has pattern of translucent and opaque areas which are disposed to ensure that at any time at least one detector can be exposed to direct sunlight through a translucent area and at least one detector is shaded from direct sunlight by an opaque area and each detector has an output indicative of the intensity of sunlight incident thereon, the method further comprising correcting each measurement using the method of the first aspect of the invention. The method may comprise determining the position of the sunlight sensor, typically from a map or from a GPS sensor. As such, the sunlight sensor may comprise a receiver for a global positioning network (such as the Global Positioning System (GPS), the Global Navigation Satellite System (GLONASS) or the Galileo system or others) and the method may comprise determining the location using the receiver. The receiver may also conveniently determine the time accurately, which can be used as the time in the method. According to a third aspect of the invention, there is provided a sunlight sensor, comprising at least two light sensitive detectors and a masking element, where the masking element has pattern of translucent and opaque areas which are disposed to ensure that at any time at least one detector can be exposed to direct sunlight through a translucent area and at least one detector is shaded from direct sunlight by an opaque area and each detector has an output indicative of the intensity of sunlight incident thereon, the sunlight sensor also comprising a processor arranged to carry out the method of the first aspect of the invention. The sensor may comprise a receiver for a global positioning network (such as the Global Positioning System (GPS), the Global Navigation Satellite System (GLONASS) or the Galileo system), with the processor being arranged to determine the location using the receiver. The receiver may also conveniently determine the time accurately, which can be used as the time by the processor. There now follows description of embodiments of the present invention, described with reference to the accompanying drawings, in which: Figure 1 is a plan view of a sunshine sensor embodying the invention; Figure 2 is a side elevation of the detector of Fig. 1; Figure 3 is an end elevation of the detector of Fig. 1; Figure 4 is an isometric view of the detector of Figure 1; Figure 5 is a sectional view of the detector of Figure 1, showing the effect of variation in solar declination; Figures 6 to 9 are representations of alternative patterns of masking elements suitable for use in various embodiments of the invention; Figure 10 shows the processing circuitry of the sensor of Figure 1; Figures 11 and 12 show a flowchart showing the method carried out by the detector of Figure 1; Figure 13 shows an example set of data captured by a detector in accordance with Figure 1; Figure 14 shows the correction functions applied using the normalised intensities; Figure 15 shows the correction functions used to apply the “cosine” correction; and Figures 16 to 21 shows graphs of data processed using a method in accordance with a second embodiment of the invention. As shown in the drawings, a sunshine sensor 10 being a preferred embodiment of the invention comprises includes a cover 11 formed as a hemispherical dome carried on a flat base plate 12. An area of the base plate 12 beneath the cover 11, carries seven light-sensitive detectors, which are in this embodiment, miniature thermopiles 13 which serve as light detectors. Processing circuitry 14 is disposed in a housing 16 below the base plate 12 (See Fig. 5). The cover 11 comprises a hemispherical dome of transparent glass. Immediately inside the glass dome, there is provided a metal mask having shading areas 17 which necessarily are opaque (where the metal is present) and empty areas 18 which are transparent (where there is no metal). The pattern is symmetrical with respect to a first diametrical axis 20, such that one half of the cover 11 is a mirror image of the other half. With respect to a second axis 21, extending normally the first axis, the pattern is symmetrical in shape, but is reversed in the sense that a point on one side of the second axis which is in an opaque area corresponds with a point on the opposite side of the axis which is in a translucent area. The miniature thermopiles 13 are arranged such that a first of them 13g is central below the cover 11 at the intersection of the first and second axes 20,21. The other six miniature thermopiles 13a.. 13f are located at the vertices of a regular hexagonal locus centred upon the first miniature thermopile 13g. The effect of variation in the altitude of the sun, due to diurnal motion, or seasonal change in declination is illustrated in Fig. 5. In a lower altitude of the sun, rays 22 impinge upon the detector 13a, leaving other detectors such as 13g, or 13d, in the shade cast by an opaque area 17. In a higher altitude, e.g. at midsummer noon, at least detector 13g, is illuminated by direct solar radiation by rays 23, whilst detectors 13a, and 13d, are in shade. At any time, the detector or detectors 13 illuminated by direct sunlight provide a measure of the intensity of direct sunlight whilst the shaded detectors 13 measure the intensity of diffuse sunlight which impinges upon them from all directions other than from the sun directly. This can be seen in Figure 13 of the accompanying drawings, which shows a sensor with seven detectors whose outputs are shown at TP1 to TP7. As the day progresses, it can be seen that the individual sensors peak at a maximum and recede to a minimum that are shared between the detectors. The patterns shown in Figures 6 to 9 in an equiangular projection are all produced to work with a layout of seven detectors as described above, with the separation between the centres of the detectors being, respectively, 0.14, 0.16, 0.19 and 0.26 times the diameter of the dome. Experimentation will reveal families of patterns which will work. Figure 10 of the accompanying drawing shows the processing circuitry. This comprises a processor 50 which takes as an input the seven outputs 51 (potentially via an analogue to digital converter, not shown) of the detectors 13. The processing circuitry also comprises a global positioning system receiver 52, which provides outputs 54 to processor recording the position of the sensor and the time. The processor 50 is arranged, as will be described below, to provide outputs 55 at which corrected values of the global, diffuse and direct normal intensities will be output. The calculation follows the flow chart shown on Figures 11 and 12 of the accompanying drawings. In step 100, the seven detector measurements are analysed to find the maximum and minimum readings MAX and MIN. In step 102, an initial value for the global illumination is determined as the sum of MAX and MIN. An initial value for diffuse illumination is determined as 2.04 times MIN. A check is made that this initial value of diffuse illumination is not more than the initial global illumination (as it cannot exceed the global illumination); if it is, then the initial diffuse illumination is set to the initial global illumination. This step corrects for small variations between the response of each of the sensors. In step 104, the initial direct illumination is determined as the difference between the initial global and diffuse illuminations multiplied by 0.99. The initial diffuse illumination value is further increased by a factor of 1.14 (so a total factor of 2.3256 is in effect applied to MIN), and the initial global illumination can then be recalculated as the initial direct illumination plus the initial diffuse illumination. These corrections correct for non-uniform spectral response in the detector elements 13. The processor also makes use of the position and time measurements from the GPS receiver 52. At step 108, this is used to determine the solar zenith angle (SZA) and the intensity of the solar irradiation at the top of the atmosphere (ToA). Both figures can be determined using a lookup table with data from a suitable astronomical almanac (such as those published by the UK Hydrographic Office or the US Naval Observatory) or calculated using a suitable algorithm such as that proposed by Reda, Ibrahim &Andreas, Afshin (2004) Solar position algorithm for solar radiation application. Solar Energy. 76. 577-589. At step 110, the processor determines the cosine of the SZA (hereafter p) and the component of the ToA in a horizontal plane (TOA x p). At step 106, these figures can be used on the captured measurements. An initial Direct Normal irradiance (DNI) can be calculated as the difference between the initial global and diffuse intensities, divided by p. At step 112, the figures are used to normalise the captured measurements. Three values are determined: Kt = Global / (ToA * p) Kdf = Diffuse / (ToA * p) Kdni = (Global - Diffuse) / (ToA * p) As such, these represent normalised values of the initial global, diffuse and direct intensities respectively, and will vary between 0 and 1 (as it would be unusual to receive more sunlight at the sensor than was incident at the top of the atmosphere). At step 114, these normalised values are used to apply correction functions to the measured data. The correction functions will be stored in lookup tables and have been calculated by calibrating sensors against gold standard pyranometers for a lengthy period of time. The calculations are of the form: GCorr = Global * f(Kt) 5 DCOrr = Diffuse * f(KDNi) * f(Kdf) where GCOrr is a corrected global intensity, Global is the initial global intensity, Kt is the normalised global intensity, Dcorr is a corrected diffuse intensity, Kdni is the normalised direct intensity, Kdf is the normalised diffuse intensity, and the three 10 functions f may be as given below: Kx f(Kt) f(Kdf) f(KDNI) 0 1.21 1.3 1 0.05 1.11 1.126 1.060445 0.1 1.08 1.06 1.078749 0.15 1.05 1.042 1.092896 0.2 1.03 1.034 1.108647 0.25 1.01 1.018 1.124859 0.3 1 1.001 1.138952 0.35 1 0.989 1.154734 0.4 1 0.971 1.169591 0.45 1 0.958 1.176471 0.5 1 0.943 1.176471 0.55 1 0.937 1.168224 0.6 1 0.919 1.152074 0.65 1 0.91 1.111111 0.7 1 0.91 1.064963 0.75 1 0.91 1.030928 0.8 1 0.91 1.020408 0.85 1 0.91 1.010101 0.9 1 0.91 1 0.95 1 0.91 1 1 1 0.91 1 The correction functions f are also shown in Figure 14 of the accompanying drawings. A check is made at this stage that the corrected diffuse intensity does not exceed the corrected global intensity; if that does occur, then the corrected diffuse intensity is capped at the corrected global intensity. At step 116, a corrected direct normal irradiance is determined as the difference between the corrected global and diffuse intensities, less a further constant term of 5 Wm'2. This corrects for an otherwise overestimation of DNI in diffuse sky conditions. A lower bound of zero is placed on the corrected direct normal irradiance. At step 120 (Figure 12) a “cosine correction” is applied to the corrected direct normal irradiance; this is of the form of a factor by which the corrected direct normal irradiance is divided, and corrects for errors when the sun is near the horizon. The function is 1 until the SZA nears 90, and then drops slightly, as shown in the following table and as shown in Figure 15 of the accompanying drawings: SZA f(SZA) SZA f(SZA) SZA f(SZA) 0 1 30 1 60 1 3 1 33 1 63 1 6 1 36 1 66 1 9 1 39 1 69 0.99 12 1 42 1 72 0.98 15 1 45 1 75 0.97 18 1 48 1 78 0.94 21 1 51 1 81 0.9 24 1 54 1 84 0.86 27 1 57 1 87 0.8 A final value of the Global intensity can then be reconstructed by adding the product of the cosine corrected direct normal irradiance with p to the corrected diffuse intensity at step 122. The (final) corrected values can then be output at the outputs 55 of processor 50. The outputs represent a more accurate measurement of the various intensity values than if the corrections had not been made, in particular by correcting for the field of view of the sensor and for the spectral response of the sunlight sensors and in particular which in turn leads to inaccuracy in the measurements of diffuse intensity due to scattering in the atmosphere in differing atmospheric conditions (e.g. differing cloud conditions). Whilst we have described the above embodiment with respect to a system where the processor 50 is contained within the sunlight sensor, that is not obligatory. The processor 50 could be in a suitably programmed computer elsewhere and receive data and process it on the fly. Alternatively, the processor could be that of a suitably programmed computer, and the data may be historic, thus allowing the improvements in accuracy to be applied to data that has previously been collected, meaning that old data sets can be re-processed to get more accurate intensity measurements. The effect of the processing, whether it be as the data is captured or for historic data, has positive effects on the accuracy of the measured data, as can be seen in Figures 16 to 21 of the accompanying drawings. In these drawings, the graphs show the average slope of the data (shown as a box) over one minute (Figures 16 to 18) or one hour (Figures 19 to 21) periods of the Global (Figures 16 and 19), Diffuse (Figures 17 and 20) and Direct (Figures 18 and 21) intensities, and the root mean square variation about the line of best fit (shown as error bars). The data is shown for the raw data (“original”) captured from sensors in series of locations, and averaged over all locations (“combined”). The same data is shown after the data has been calibrated against a known instrument for the global measurement (“global cal”) and after the corrections described above have been applied (“final”). Finally, the values obtained from a reference instrument at each location are shown (“ref’). The error bars on the “ref’ data set represent a likely limit on how accurate the measurements using this method could become. It can therefore be seen that the calibration of the global measurement (i.e. “global cal”) removes much of the variation in the global intensity. The correction of this method (“final”) act to tighten the errors bars. However, the effect of the above correction method is more apparent in the diffuse measurement, where the “final” set of data shows a correction of a below-zero offset in the “original” data, as well as a tightening of the error bars. The “final” data is 5 much closer to the “ref’ data than the “original” data. The direct intensity shows a similar theme, where the “final” data corrects for a positive offset in the data as compared to the “original” data, ending up with a result closer to the “ref’ data and with smaller error bars than the “original” data, 10 approaching those of the “ref’ set. As such, it can be seen that the data as corrected by the method described herein is generally more accurate than before the correction is applied.

Claims

1. A method of correcting a measurement of the intensity of sunlight at a location and at a time, the measurement having been taken using a sunlight sensor comprising at least two light sensitive detectors and a masking element, where the masking element has pattern of translucent and opaque areas which are disposed to ensure that at any time at least one detector can be exposed to direct sunlight through a translucent area and at least one detector is shaded from direct sunlight by an opaque area and each detector has an output indicative of the intensity of sunlight incident thereon, the method comprising the following steps carried out on at least one processor:determining a maximum intensity indicated by the output signals and a minimum intensity indicated by the output signals;determining an initial global intensity based upon a sum of the maximum and minimum intensities;determining an initial diffuse intensity by scaling the minimum intensity by a scaling factor;determining an initial direct intensity based on the difference between the initial global intensity and the initial diffuse intensity multiplied by a scaling factordetermining a solar zenith angle for the location and time, being the angle which the sun’s rays make with a vertical line at the location and time;determining an initial direct normal irradiance by dividing the initial direct intensity by a divisor dependent on the solar zenith angle;determining a top of atmosphere irradiance horizontal component for the location and time, being a value indicative of the solar irradiance on a horizontal plane at the top of the atmosphere above the location at the time;determining a normalised value of each of the initial diffuse intensity, the initial global intensity and the initial direct intensity;determining a correction function for each of the initial diffuse intensity, the initial global intensity and the initial direct intensity based on the normalised value of each of the initial diffuse intensity, the initial global intensity and the initial direct intensity;applying the correction function to each of the initial diffuse intensity and the initial global intensity to form a corrected diffuse intensity and a corrected global intensity; anddetermining a corrected direct normal irradiance based upon the difference between the corrected diffuse intensity and the corrected global intensity.

2. The method of claim 1, in which the normalised values of each of the initial diffuse intensity, the initial global intensity and the initial direct intensity, are determined by dividing each intensity by a top of atmosphere irradiance horizontal component, determined as a top of atmosphere irradiance multiplied by the cosine of the solar zenith angle.

3. The method of claim 1 or claim 2, in which the correction functions are applied as:Gcorr = Global * f(Kt)Dcorr = Diffuse * f(KDNl) * f(Kdf)where GCOrr is the corrected global intensity, Global is the initial global intensity, Kt is the normalised global intensity, DCOrr is the corrected diffuse intensity, Kdni is the normalised direct intensity, Kat is the normalised diffuse intensity, and f represents the three correction functions4. The method of claim 3, in which at least one following applies:the correction function dependent on the normalised initial global intensity starts positive at zero intensity, and decreases to zero before remaining zero from a limit point up to a maximum value of the normalised initial global intensity of 1;the correction function dependent on the normalised initial diffuse intensity starts at a positive value, and decreases past zero to a lower bound from a limit point and stays at that lower bound up to a maximum value of the normalised initial global intensity of 1; andthe correction function dependent on the normalised direct intensity starts at zero at zero normalised direct intensity, increases to a peak with increasing normalised direct intensity before reducing back to a minimum value of 0 at a limit point before remaining zero up to a maximum value of the normalised direct intensity of 1.

5. The method of any preceding claim, comprising applying a cap to the corrected diffuse intensity, so that it cannot exceed the corrected global intensity.

6. The method of any preceding claim, in which the corrected direct normal irradiance (DNI) is determined by subtracting a constant term from the difference between the corrected diffuse intensity and the corrected global intensity.

7. The method of any preceding claim in which the corrected direct normal irradiance is calculated by applying a correction function to correct for the response of the sensor when the sun is near the horizon.

8. The method of claim 7, in which the application of the correction function to the direct normal irradiance comprises dividing the corrected direct normal irradiance by a correction function, which is 1 at solar zenith angles up to a limit and decreases from 1 at increasing solar zenith angles above the limit.

9. The method of any preceding claim, comprising calculating a final global intensity by adding the corrected direct normal irradiance, multiplied by the cosine of the solar zenith angle, to the corrected diffuse intensity.

10. A method of measuring the intensity of sunlight, the method comprising taking at least one measurement of sunlight intensity using sunlight sensor comprising at least two light sensitive detectors and a masking element, where the masking element has pattern of translucent and opaque areas which are disposed to ensure that at any time at least one detector can be exposed to direct sunlight through a translucent area and at least one detector is shaded from direct sunlight by an opaque area and each detector has an output indicative of the intensity of sunlight incident thereon, the method further comprising correcting each measurement using the method of any preceding claim11. The method of claim 10, comprising determining the position of the sunlight sensor.

12. A sunlight sensor, comprising at least two light sensitive detectors and a masking element, where the masking element has pattern of translucent and opaque areas which are disposed to ensure that at any time at least one detector can be exposed to direct sunlight through a translucent area and at least one detector is shaded from direct sunlight by an opaque area and each detector has an output indicative of the intensityof sunlight incident thereon, the sunlight sensor also comprising a processor arranged to carry out the method of any of claims 1 to 11.

13. The sensor of claim 12, comprising a receiver for a global positioning network, 5 with the processor being arranged to determine the location using the receiver.

Images