Reflection–Polarization Characteristics of Greenhouses Studied by Drone-Polarimetry Focusing on Polarized Light Pollution of Glass Surfaces

Abstract

Drone-based imaging polarimetry is a valuable new tool for the remote sensing of the polarization characteristics of the Earth’s surface. After briefly reviewing two earlier drone-polarimetric studies, we present here the results of our drone-polarimetric campaigns, in which we measured the reflection–polarization patterns of greenhouses. From the measured patterns of the degree and angle of linear polarization of reflected light, we calculated the measure (plp) of polarized light pollution of glass surfaces. The knowledge of polarized light pollution is important for aquatic insect ecology, since polarotactic aquatic insects are the endangered victims of artificial horizontally polarized light sources. We found that the so-called Palm House of a botanical garden has only a low polarized light pollution, 3.6% ≤ plp ≤ 13.7%, while the greenhouses with tilted roofs are strongly polarized-light-polluting, with 24.8% ≤ plp ≤ 40.4%. Similarly, other tilted-roofed greenhouses contain very high polarized light pollution, plp ≤ 76.7%. Under overcast skies, the polarization patterns and plp values of greenhouses practically only depend on the direction of view relative to the glass surfaces, as the rotationally invariant diffuse cloud light is the only light source. However, under cloudless skies, the polarization patterns of glass surfaces significantly depend on the azimuth direction of view and its angle relative to the solar meridian because, in this case, sunlight is the dominant light source, rather than the sky. In the case of a given direction of view, those glass surfaces are the strongest polarized-light-polluting, from which sunlight and/or skylight is reflected at or near Brewster’s angle in a nearly vertical plane, i.e., with directions of polarization close to horizontal. Therefore, the plp value is usually greatest when the sun shines directly or from behind. The plp value of greenhouses is always the smallest in the green spectral range due to the green plants under the glass.

1. Introduction

Aquatic insects recognize water surfaces by the horizontal polarization of water-reflected light [1,2,3], and therefore they are attracted to all natural or manmade sources of such light [4]. This phenomenon is the consequence of polarized light pollution being a special kind of ecological photopollution [5,6]. Polarized light pollution (PLP) refers to the adverse effects of horizontally polarized light with high degree of linear polarization reflected from smooth (shiny) and dark (especially black) artificial surfaces on polarotactic water-seeking aquatic insects [7]. Typical PLP sources are photovoltaic solar panels [8] and glass surfaces of buildings [9,10,11], for instance. The knowledge of PLP is important for aquatic insect ecology, since polarotactic aquatic insects are the most endangered victims of artificial horizontally polarized light sources. This is because horizontally polarizing artificial surfaces deceive these insects, which land and lay eggs on them [12]. Since the laid eggs perish due to dehydration [13], this phenomenon endangers the local aquatic insect population.

The polarization characteristics of light can effectively be studied with imaging polarimetry [14]. The reflection–polarization patterns of terrestrial objects can aerially be measured from balloons [15,16,17] and satellites [18,19,20]. Recently, drone-based imaging polarimetry was used to investigate certain aspects of the polarization of ground-reflected light [21,22]. In the Hungarian lake Balaton, dark water patches occur from every autumn to spring because of the inflow of black organic material into the bright lake water. Using drone-polarimetry, Száz et al. [21] found that these dark lake patches reflect light with very high degrees 60% ≤ d ≤ 80% of horizontal polarization at Brewster’s angle (=arc tan n from the normal vector of the reflecting surface, where n is refractive index of the surface material), while the bright lake water is only weakly (d < 20%) horizontally polarizing. There was a large contrast both in the intensity and polarization degree between dark lake patches and bright lake water, while there was no contrast in the polarization direction. The ecological consequence of these polarization characteristics is that these dark lake patches attract water-seeking polarotactic insects, which lay eggs more frequently in them than in the brighter lake water [23]. Thus, the abundance of breeding flying aquatic insects increases where dark lake patches form.

Using drone-polarimetry above a solar panel farm, Takács et al. [22] measured the reflection–polarization characteristics of fixed-tilt photovoltaic panels from the viewpoint of flying polarotactic aquatic insects. They found that the temporal changes of polarized light pollution were complementary for two orthogonal viewing directions relative to the panel rows. The numerical magnitude (plp) of polarized light pollution of solar panels viewed parallel to the panel rows was the highest (plp = 49–58% after sunrise, and plp = 35–48% prior to sunset) at low solar elevations, after sunrise and at or prior to sunset, when many aquatic insect species fly and seek water bodies. The polarized light pollution of solar panels viewed perpendicular to the panel rows was the highest (plp = 29–35%) at the largest solar elevations, near noon, when also many flying aquatic insects actively seek water. Thus, solar panel farms near wetlands can be dangerous for these insects.

A further source of polarized light pollution are greenhouses, the glass surfaces of which can reflect horizontally polarized light with high degrees of linear polarization, which attract flying water-seeking polarotactic aquatic insects from the vicinity. These insects can lure insectivorous bats to the glass surfaces of greenhouses [24,25], and these bats can collide with the glass panes [26,27]. We present here the results of our drone-polarimetric campaigns in which we measured the reflection–polarization patterns of some greenhouses. From the measured patterns of the degree and angle of linear polarization of glass-reflected light, we calculated the quantity (plp) of polarized light pollution of the studied greenhouses.

2. Materials and Methods

Using drone-polarimetry in two field campaigns, we measured the reflection–polarization characteristics of vertical and differently tilted glass surfaces of various greenhouses. These glass surfaces were smooth (shiny), colorless, transparent and did not have an anti-reflective coating. The first field measurement took place on 3 November 2022 in the ELTE Botanical Garden of the Eötvös Loránd University (Budapest, 47°29′02″N, 19°05′08″E) with the permission and consent of László Orlóci, director of the Botanical Garden. This measurement was carried out in the morning, under an overcast sky. The second measurement was conducted on 22 April 2023, at a greenhouse in Vácrátót (eastern Hungary, 47°42′39″N, 19°14′07″E), with the permission of the owner, Lajos Héder. At that time, the sky was cloudless, so the measured polarization patterns of the greenhouse depended significantly on the azimuth angle of the drone polarimeter and the solar meridian.

Our drone-polarimetric technique and evaluation of the measured polarization pictures were the same as described in detail elsewhere [21,22]. Here, we mention only the fact that the polarization camera was mounted on the bottom of the drone, and its optical axis pointed toward Brewster’s angle θ_Brewster = arctan (n = 1.33) = 53°, measured from the vertical, where n = 1.33 is the refractive index of water. A 53° orientation was chosen, since water-seeking polarotactic aquatic insects detect water predominantly by means of the horizontally polarized light with high degree of linear polarization coming mainly from this direction [1,2,12,28]. In the ventral half of the compound eyes of the backswimmer water bug Notonecta glauca, for example, there is a polarization-sensitive special zone, with the field of view oriented toward Brewster’s angle in flight to perceive the horizontally polarized water-reflected light with maximal degree of polarization [1,2]. The 53° orientation of our polarimeter’s optical axis corresponds to this anatomy of the Notonecta’s eyes. Note, however, that 53° from the vertical coincides with Brewster’s angle only for horizontal reflecting surfaces, such as the water surface, for example. For tilted glass surfaces (with a 1.5 refractive index) of greenhouses, Brewster’s angle is 56.3° relative to the surface’s normal. Thus, our polarimeter’s optical axis was usually not oriented towards the maximally polarized light Brewster-reflected from tilted glass surfaces. However, this was not a problem because, due to the relatively large field of view of 86° of our polarization camera, the orientation modelled the polarization-sensitive eyes of flying polarotactic aquatic insects. For further details, see Figures 1A and 5 of Takács et al. [22], depicting the reflection–polarization characteristics of tilted solar panels, measured with the same drone-polarimetry as used in the present work.

The reflection–polarization characteristics were measured in the blue (450 nm), green (550 nm) and red (650 nm) spectral ranges, to which our CMOS sensor was sensitive. In this work, we present only the polarization patterns of greenhouses measured in the green part of the spectrum, because they were very similar in the red and blue spectral ranges. The sun-to-drone angle of the optical axis of the polarization camera measured from the solar meridian is marked by β. Furthermore, d is the degree (%) of linear polarization, and α is the angle of polarization measured clockwise from the vertical. The definition of both polarization variables is available in [14], for instance. According to our test of the polarization camera and the evaluation of its polarization images, the net uncertainties Δd and Δα of the measured degree of linear polarization d and the angle of polarization α of light are Δd ≈ ±1%, Δα ≈ ±1° for d = 100%, and Δd ≈ ±3% and Δα ≈ ±3° for d = 15% (=threshold of polarization sensitivity of aquatic insects [28]). The evaluation process of polarization images and further details of our drone-polarimetry were described by Száz et al. [21].

We use the quantity plp = N_water/N_glass as the quantitative measure of the polarized light pollution of greenhouse glass surfaces, where N_water is the number of glass pixels detected as water by a polarotactic insect, and N_glass is the number of the whole glass area. These numbers were determined as described in [22]. Here, we mention only that in every picture, these pixels were determined based on the following conditions: d > d* = 10% and |90° − α| < α* = 25° [4]. Such pixels represent what a hypothetical polarotactic aquatic insect, possessing polarization sensitivity thresholds d* and α*, could mistakenly sense as water. These pixels are marked in blue color in Figure 1E, Figure 2E,J,O and Figure 3E,J,O,T (and in the last row of Supplementary Figures S1–S13). Finally, the number N_water of blue pixels (without any over- or underexposed pixels) of the areas detected as water was also counted, and the quotient plp = N_water/N_panel was calculated.

3. Results

3.1. Polarization Patterns of Greenhouses in the ELTE Botanical Garden

Figure 1 shows the five triangular tilted glass panels (1., 2., …, 5.) of the central pyramid of the Palm House in the ELTE Botanical Garden, together with both (6. and 7.) glass-covered side naves. The pattern of the polarization degree d is almost mirror-symmetric to the pyramid’s vertical axis. The glass roofs of the right (6.) and left (7.) side naves (Figure 1C) reflect less polarized light from the side closer to the pyramid, while d increases farther away from the pyramid. The light originating from the 1. and 5. glass triangles of the pyramid is largely unpolarized, because the light is reflected from them at a very flat angle. At the same time, glass triangles 2. and 4. reflect light with a high d due to reflection close to Brewster’s angle. Glass triangle 3., perpendicular to the optical axis of the drone camera, reflects light at almost 90°, thus being practically unpolarized (d ≈ 0%). The polarization angle of the glass surfaces varies depending on their inclination: glass panes 2., 4., 6. and 7. reflect light with a moderate d and almost vertical polarization (Figure 1D). Therefore, none of the conditions of polarized light pollution are met for the majority of glass surfaces in Figure 1E. However, the top part of glass triangle 3., close to the pyramid’s apex, reflects light with a nearly horizontal polarization, which is why it appears as a surface detected as water (i.e., polarized-light-polluting) in Figure 1E.

The reflection–polarization patterns of the glass roof of the Palm House in the ELTE Botanical Garden were practically independent of the wavelength. One of the reasons for this is that the sky was completely cloudy; thus, the glass surfaces were illuminated by white cloud light. Another reason is that the glass surfaces are also colorless. Supplementary Figures S1–S8 display further photographs and polarization patterns of the same Palm House in the ELTE Botanical Garden, measured by drone-polarimetry from different heights and viewing directions.

Figure 2 shows the measured polarization patterns of two other greenhouses in the ELTE Botanical Garden. Glass surfaces 2. and 3. in Figure 2A–E reflect light with the highest degree of linear polarization d, because they reflect light closest to Brewster’s angle. Glass surfaces 1., 2. and 3. in their entirety and the middle of glass pane 4. are horizontally polarized, since they reflect the skylight coming from above, and thus the reflection plane is vertical. Due to their sufficiently high degree of polarization and horizontal polarization direction, certain parts of glass surfaces 1., 2., 3. and 4. are polarized-light-polluting. Although the left half of glass surface 5. in Figure 2A–E has a relatively high d, it is not polarized-light-polluting, because the reflected light is nearly vertically polarized.

In Figure 2F–J, the highest d is reflected by glass surfaces 1. and 3. Polygonal tower 5. (similar to the glass tower of the Palm House in Figure 1) also reflects light with a high d. Glass surfaces 1., 2., 3. in their entirety and the top center of glass roof 4. reflect horizontally polarized light, while the rest of glass surface 4. reflects oblique or vertical polarization. The polarization direction of the glass triangles of tower 5. depends on the angle of incidence of light from them to the drone polarimeter. Apart from glass surface 4., some parts of glass surfaces 1., 2., 3. and 5. are polarized-light-polluting. Furthermore, some of the water basins 6. also reflect strongly and horizontally polarized light, as a result of which they are detected as water surfaces in Figure 2J.

Figure 2K–O shows the same greenhouses as Figure 2F–J, but from a viewing direction turned horizontally by 90°. Glass surfaces 3., 4. and 7. were covered with a transparent white plastic sheet. The highest d is reflected by the entirety of glass surfaces 2. and 5., as well as by the upper contact strip of glass plates 3. and 4.. The polarization direction of light reflected from glass surfaces 2. and 5. is nearly horizontal, while the distribution of polarization direction reflected from glass surfaces 3. and 4. is very inhomogeneous. Glass surfaces 2. and 5. in their entirety are polarized-light-polluting due to their nearly horizontal polarization direction and high polarization degree. Among the partially vegetated pools 9.–11., pools 9. and 10. reflect strongly and horizontally polarized light, so they appear partly as water surfaces in Figure 2O.

3.2. Polarization Patterns of Greenhouses and a Water Pool in Vácrátót

Figure 3 (see also Supplementary Figures S9–S13) shows the reflection–polarization patterns of the examined greenhouses in Vácrátót (a larger greenhouse with white tarpaulin stripes and a smaller greenhouse next to it, without such stripes) and the areas detected as water by polarotactic aquatic insects. A significant difference, compared to the drone-polarimetric measurement in the ELTE Botanical Garden, is that the sky was cloudless in Vácrátót, and because of this, the angle between the solar meridian and the azimuth direction of the optical axis of the drone polarimeter greatly influenced the polarization characteristics of the glass surfaces. On the one hand, the polarization characteristics of cloudy (overcast) and cloudless skies differ considerably. The polarization degree of light from an overcast sky is significantly lower than that from a clear sky due to the depolarizing effect of multiple scattering on cloud particles (water droplets or ice crystals) [10]. On the other hand, the reflection–polarization properties of glass surfaces also depend strongly on the angle of incidence of light relative to the surface’s normal, and this incident angle changes with the angle between the meridian of the sun (being the strongest light source in the celestial hemisphere) and the azimuth direction of the optical axis of the polarimeter [22].

The larger greenhouse in Vácrátót had retractable white tarpaulin sheets that could be used to control the amount of light entering the greenhouse. During the measurement, these white sheets were pulled together into thin strips, which reduced the polarization degree d of the glass roofs in the strips due to their whiteness (Figure 3C). The reason for this reduction is Umov’s law [29]: darker surfaces reflect light with lower degrees of linear polarization d. The d of the glass surfaces was much higher where these white tarpaulin stripes were not present. This banded structure of the d-pattern also appears on all other scenes measured from different viewing directions (Figure 3H,M,R, Supplementary Figures S9–S13).

In Figure 3D, the prevailing polarization angle α of the glass surfaces around −45° (coded with yellow-blue colors) is perpendicular to the oblique sunlight coming from the top right corner of the scene. However, the α of darker glass surfaces with a higher polarization degree d is usually closer to horizontal, while that of lighter glass surfaces with a lower d deviates more significantly from horizontal. Both d and α are slightly color dependent. Because of the blue skylight, the d values were usually slightly higher in the blue spectral range than in the red and green. In Figure 3E, the polarized light pollution (i.e., the proportion of area detected as water) is low because, in the case of this viewing direction, only a small glass surface reflects light with nearly horizontal polarization and a sufficiently high d. In Figure 3E, the blue areas detected as water are also striped due to the previously mentioned white tarpaulin stripes.

The scene in Figure 3F–J is viewed from the same direction as Figure 3A–E, but with the difference that the water pool is also visible on the left side of the former. The sun shone again from the top right direction; therefore, the polarization characteristics and polarized light pollution of the greenhouses are very similar to those in Figure 3A–E.

In Figure 3K–O, the sun was shining with β = 20° right from the polarimeter’s optical axis, so its reflections appeared on some tilted glass roofs. The water surface of the pool that stores firefighting water, visible on the left, reflected light with a relatively high d and horizontal polarization. Since the water in the pool had a slightly greenish hue due to green algae, the d of the water-reflected light was the highest in the blue spectral range and the lowest in the green range. Viewed from this direction, many glass surfaces have a very high d (Figure 3M) due to reflection close to Brewster’s angle. Since the sunlight was reflected in an almost vertical plane, the dominant polarization direction is horizontal (Figure 3N). Thanks to this, a large proportion of the glass surface is polarized-light-polluting (Figure 3O).

In Figure 3P–T, the sun was shining with β = 5° left from the polarimeter’s optical axis; therefore, the polarization patterns of greenhouses, water pool and areas detected as water are very similar to those in Figure 3K–O: a high d, nearly horizontal polarization and considerable polarized light pollution. On the left side of Figure 3R, it is clear that the d of the smaller greenhouse is much higher than that of the large greenhouse on the right side. The reason for this is that there were no white tarpaulin stripes in the small greenhouse, which reduced the polarization degree of the large greenhouse. Supplementary Figures S9–S13 represent further photographs and polarization patterns of the same greenhouses in Vácrátót, measured by drone-polarimetry from different viewing directions.

3.3. Numerical Values of the Polarized Light Pollution of Greenhouses

Table 1. Numerical values of the polarized light pollution, plp (%) of greenhouses in the ELTE Botanical Garden measured in the red (650 nm), green (550 nm) and blue (450 nm) spectral ranges. The plp value is the smallest in the green range.

Figure Number	Red (%)	Green (%)	Blue (%)
Figure 1 (Palm House)	3.6	2.3	3.9
Figure 2A–E (tilted-roofed greenhouse)	25.7	24.8	27.4
Figure 2F–J (tilted-roofed greenhouse)	33.1	31.3	34.3
Figure 2K–O (tilted-roofed greenhouse)	39.7	38.4	40.4
Figure S1 (Palm House)	7.1	4.4	8.2
Figure S2 (Palm House)	3.9	1.7	3.9
Figure S3 (Palm House)	4.9	2.9	5.1
Figure S4 (Palm House)	7.1	3.7	6.6
Figure S5 (Palm House)	6.9	4.4	6.1
Figure S6 (Palm House)	13.7	8.8	12.0
Figure S7 (Palm House)	7.6	4.2	6.4
Figure S8 (Palm House)	8.3	5.1	7.0

and

Table 2. Numerical values of the polarized light pollution, plp (%) of the greenhouses in Vácrátót, measured in the red (650 nm), green (550 nm) and blue (450 nm) spectral ranges. The plp is the largest in the blue range.

Figure Number	Red (%)	Green (%)	Blue (%)
Figure 3A–E	25.7	23.0	29.3
Figure 3F–J	29.3	27.5	30.6
Figure 3K–O	69.6	68.5	70.0
Figure 3P–T	73.0	71.3	76.7
Figure S9	21.4	24.8	31.2
Figure S10	9.5	9.4	15.9
Figure S11	14.8	19.4	25.9
Figure S12	17.3	16.9	25.9
Figure S13	11.4	9.3	15.9

contain the numerical values of the polarized light pollution plp (%) of the greenhouses, examined by drone-polarimetry in the red (650 nm), green (550 nm) and blue (450 nm) ranges of the spectrum. The plp value was determined as follows: after marking manually the glass surfaces on the drone images with a homogeneous red mask (see Figure 1B, Figure 2B,G,L and Figure 3B,G,L,Q, as well as the subfigure N of Supplementary Figures S1–S13), we subtracted the under- and overexposed pixels from this mask, and counted the number N_glass of the remaining red pixels. In the mask, we also counted the number N_water of blue pixels detected as water in Figure 1E, Figure 2E,J,O and Figure 3E,J,O,T (and in the last row of Supplementary Figures S1–S13). Finally, we calculated the ratio plp = N_water/N_glass.

Table 1 shows the plp values of greenhouses in the ELTE Botanical Garden. The glass surfaces of the Palm House (Figure 1, Supplementary Figures S1–S8) have only a relatively small polarized light pollution of 3.6% ≤ plp ≤ 13.7% in all three (R, G, B) tested color ranges. In the green range, the plp value of the Palm House and the other greenhouses is the smallest. The reason for this is that most of the light is reflected from the green plants under the glass surfaces in the green range of the spectrum, which, passing through the glass, becomes slightly obliquely or vertically polarized after refraction. This green-intensive weakly and non-horizontally polarized light reduces the horizontal polarization of light reflected from the outer glass surface, whereby the resulting polarization degree is the smallest in the green range, which is mostly responsible for the smallest plp value.

The polarized light pollution 24.80% ≤ plp ≤ 40.38% of the greenhouses in Figure 2A–O in the ELTE Botanical Garden is much higher than that of the Palm House. The main reason for this is that the former greenhouses only have tilted roofs, in contrast to the dominant central glass pyramid of the Palm House. This different geometry of the glass surfaces results in different polarization patterns, which causes different plp values.

Table 2 contains the percentage values of the polarized light pollution (plp) of the greenhouses in Vácrátót, measured in the red (650 nm), green (550 nm) and blue (450 nm) spectral ranges. The values of 9.2% ≤ plp ≤ 76.7% in Vácrátót are much higher than those (3.6% ≤ plp ≤ 13.7%) of the Palm House in the ELTE Botanical Garden (Table 1). In Vácrátót, the plp value is the greatest in the blue range, because in this spectral region, the net polarization degree of the sunlight and skylight reflected from the glass surfaces is the highest, due to the blue skylight. It is a physical fact that the net reflection–polarization characteristics of a reflecting surface are determined by the superposition of the following two light components: (1) light reflected from the surface and (2) light returned (back-scattered and refracted) from the material below the surface. The first and the second components are partially linearly polarized perpendicular and parallel to the plane of reflection, respectively. If the intensity of the first/second component is stronger, the net polarization direction is perpendicular/parallel to the plane of reflection. Due to this polarization orthogonality of these two components, they disrupt each other’s influence; that is, they decrease the net polarization degree. Since the intensity of blue skylight is strongest in the blue spectral range because of Rayleigh’s scattering, the first component is strongest in the blue range, which results in the polarization degree being highest in the blue part of the spectrum.

The plp values in Figure 3O,T are particularly large, because almost all glass surfaces reflect light with high and horizontal polarization. This was caused by the fact that the azimuth of the drone polarimeter was close to the solar meridian, and the tilted glass surfaces were favorably oriented relative to the polarimeter. From Table 2, it is also clear that the plp value of a greenhouse strongly depends on the direction from which the observer (drone polarimeter or polarotactic aquatic insect) looks at the greenhouse, relative to the direction of inclination of the glass panes and the solar meridian.

4. Discussion

The polarized light pollution (PLP) can be monitored with imaging polarimetry, the newest technique of which is drone-polarimetry. Drone-based imaging polarimetry is a valuable new tool of remote sensing of the polarization characteristics of the Earth’s surface. In this work, we present the results of our drone-polarimetric campaigns, in which we measured the reflection–polarization characteristics of greenhouses and determined the PLP of their glass surfaces for flying water-seeking polarotactic aquatic insects, which are the most endangered victims of this kind of photopollution [7].

Water-seeking polarotactic insects can land and lay eggs on polarized-light-polluting glass surfaces of greenhouses. These glass panes imitate the water surface if they reflect horizontally polarized light. However, a horizontally polarizing glass pane attracts polarotactic aquatic insects only if its surface is larger than a species-dependent critical value [8]. The reason for this is that if the water surface is smaller than the critical surface, it can, for example, easily dry out and/or run out of oxygen and nutrients, and/or it can easily overheat in summer and/or freeze to the bottom in winter, and/or insect larvae that develop in this water can more easily fall prey to predators. Thus, even though many sub-areas of a larger region of the optical environment reflect light with a sufficiently high polarization degree and a polarization direction close enough to the horizontal, if the surface of these sub-regions is smaller than the critical surface in question, harmful PLP will not occur. This is the case in Figure 1, Figure 2 and Figure 3 (and Supplementary Figures S1–S13), depicting regions that are not glass surfaces, and yet they are blue in their PLP patterns. However, this does not mean that they are also surfaces (in addition to greenhouse glass) that are dangerous to water-seeking polarotactic aquatic insects. These non-glass blue regions are composed of numerous separated tiny areas, which do not merge into continuous, large-enough regions, therefore providing false “water” readings.

The PLP of smooth glass surfaces of greenhouses can be reduced by the following methods [4,7]: (1) If the glass panes are covered with a grid of thin (1–5 mm) white lines/stripes, their PLP is reduced. The denser these white lines/stripes, the lower is the PLP. (2) If the outer surface of glass panes is made rough by sand blasting, then it reflects light diffusely, which considerably decreases the polarization degree d, and thus also the PLP of reflected light.

We emphasize that drone-based imaging polarimetry enables several new types of polarization measurements. With its use, for instance, the reflection–polarization patterns of water surfaces accessible only from the air can be measured. Such areas are smaller bodies of backwaters covered with floating seaweed, which indicate the presence of water from long distances to water-seeking aquatic insects, on the basis of the horizontal polarization of water-reflected light [30]. The color of water bodies and the polarizing ability of their surface can change significantly during the seasonal overgrowth of certain algae and bacteria [31]. Drone-polarimetry is an ideal tool for investigating these periodic changes.

In this work, we used the threshold values d* = 10% and α* = 25° originating from the literature [4]. The values d* and α* depend on the concerned insect species. The change in these values results in a change in the plp values such that the smaller d* is and the larger α* is, the higher the plp value is. If biologists measure the values of d* and α* for many different polarotactic insect species, the species-specific plp values can be determined. These could be used to categorize which species are more endangered by a given polarized-light-polluting source (e.g., greenhouses) than other. However, until now, the values of d* and α* have been measured only for a few aquatic insect species [32,33], because such insect physiological measurements are difficult and time consuming.

The flying height of insects could also influence the behavior of insects, because they also have an olfactory sense, which is helpful for them to sense water. However, in the case of the studied greenhouses olfactory influence was out of the question, because the glass panes were odorless. Schwind [1,2,3] showed that flying polarotactic aquatic insects detect water predominantly by the horizontal polarization of water-reflected light, and other cues (e.g., odor, color, light intensity) are secondary or even marginal. The horizontally polarized optical cue is very effective even at remote distances, while olfaction usually functions only at closer ranges.

Finally, we mention that we are developing a new mounting mechanism for the polarization camera. This customized mounting ensures that both the azimuth and tilt angles of the camera can be changed arbitrarily with stability and precision (Figure 4). It is available in two different versions, to meet individual needs and interchangeability. This mounting uses a modular structure, so it can be easily applied to other drone types by changing the baseplate, and the individual elements can be freely combined and matched [34]. For future drone-polarimetric studies, we plan to use this new camera mounting, making it possible to change the azimuth and tilt angles of the camera on board of the flying drone during measurements. This improved approach enhances the effectiveness of our drone-based imaging polarimetry system by ensuring an arbitrary direction of the optical axis of the polarization camera, compared to the fixed Brewster’s angle of 53° of the present polarimeter.

5. Conclusions

Greenhouses always contain more or less polarized-light-polluting glass surfaces, thus luring flying water-seeking polarotactic aquatic insects, independently of the sky cloudiness, solar elevation and direction of view. The visually deceived aquatic insects frequently alight and oviposit on horizontally polarizing glass panes, and the laid egg batches perish due to dehydration. On the basis of the results presented in this work, we conclude the following:

• The magnitude of the plp values of the polarized light pollution (PLP) of glass surfaces of greenhouses ranges between low (~4%) and high (~76.7%) values, depending mainly on the direction of observation, the surface’s tilt angle, solar position and cloud cover.
• Under overcast skies, the polarization patterns and PLP of greenhouses practically only depend on the direction of view relative to the glass surfaces because the rotationally invariant diffuse cloud light is the only light source then. However, under cloudless skies, the polarization patterns of greenhouses significantly depend on the azimuth viewing direction and its angle relative to the solar meridian because, in this case, sunlight is the dominant light source, rather than the sky.
• In the case of a given direction of view, these glass surfaces are the strongest polarized-light-polluting sources, from which sunlight and/or skylight are reflected at and near Brewster’s angle in a nearly vertical plane, i.e., with a polarization direction close to horizontal. Therefore, the PLP is usually greatest when the sun shines directly or from behind.
• If a glass surface only reflects the skylight, then due to the average vertical reflection plane, the polarization direction of glass-reflected light is horizontal or close to it, which favors the PLP.
• If sunlight hits a greenhouse from the side, obliquely, the polarization direction of glass-reflected light is usually perpendicular to the direction of the sun, i.e., vertical or oblique, which does not attract polarotactic aquatic insects, i.e., the glass is not polarized-light-polluting.
• Under clear skies, the PLP of greenhouses is the greatest in the blue range of the spectrum because of the dominant blue color of skylight. The PLP of greenhouses is always the smallest in the green spectral range, due to the green plants under the glass.
• If there are no plants in a greenhouse, only soil, then under cloudy skies, the polarization patterns of glass surfaces are practically independent of wavelength due to the white (colorless) cloud light and the also practically colorless soil.
• If the white tarpaulins that protect against strong sunlight are drawn in a greenhouse, the PLP of the glass surfaces is significantly reduced, mostly due to the drastically reduced polarization degree and, to a small extent, because the white-tarpaulin-reflected light can even result in a vertical or oblique polarization direction when passing through the glass, which is far from the horizontal polarization favorable for polarotactic aquatic insects.
• The PLP of greenhouses can be decreased by making the outer glass surfaces rough (matte) and/or by covering the glass panes with a white grid pattern.