Retrieval of aerosol properties from zenith sky radiance measurements

.

The standard instrument of AERONET is CE318 photometer (Cimel Electronique SAS), which records medium Sun-Earth distance; therefore, it has been corrected from Sun-Earth distance for each day of the 157 year.

171
The ZSR has been simulated at 440, 500, 675 and 870 nm with the GRASP forward module using 172 all the mentioned input data whenever it was available. These simulations have been used for calibration 173 purposes as can be observed in Section 3, but also for the sensitivity analysis with synthetic data of Section  The present study aims to retrieve aerosol properties with GRASP using as input the calibrated 178 ZSR from the ZEN-R52 at four effective wavelengths. The versatility of GRASP allows different 179 approaches to model aerosols to maximize the possibilities of the different retrieval schemes. Due to the 180 reduced amount of information produced by the ZEN-R52, the approach called 'models' has been chosen A methodology for the ZEN-R52 calibration is proposed in this Section. This methodology is a field 208 campaign which does not require laboratory measurements except for the dark signal characterization, and 209 it is based on four steps: dark signal correction, quality data filtering, temperature correction, and a final 210 comparison against simulated values to convert the output signal from ADU into radiance units (Wm -2 nm -211 1 sr -1 ). With this purpose ZSR simulations have been performed for the whole dataset of ZEN-R52 212 measurements, using the GRASP forward module fed with the closest AERONET information (Section 213 2.2.1) whenever it was available within ±5 minutes from the ZEN-R52 measurement; considering in good 214 approximation, and as checked later, that aerosol conditions will not change significantly within 5 minutes.

215
Only those AERONET retrievals with a sky error lower than 5% have been used to ensure the quality of 216 the simulations, obtaining a total of 4725 data pairs.

219
For the dark signal (DS) evaluation, the instrument was fully covered with a black piece and 220 introduced into a thermal chamber in the GOA-UVa facilities. The instrument was subjected to a 221 temperature variation in the range from -10 to 50 ᵒC in darkness conditions. The dark signal registered by 222 each channel at each temperature is shown in Figure 1. It shows a constant behaviour for 440 and 500 nm 223 filters. Contrary, for the other wavelengths a stepped exponential behaviour can be seen. In order to 224 characterize this behaviour, the logarithm of the ZEN dark signal has been fitted to a three-degree 225 polynomial. This fitting is after rounded up to the unit to obtain a stepped fitting. The modelled dark signal 226 is also represented in Figure 1 by the black lines. This modelling has been used to subtract the corresponding 227 dark signal value to the raw signal, obtaining dark signal corrected ZSR (ZSRDSC). The DS signal has been 228 characterized in the laboratory in this work in order to cover a high range of temperatures, but it could be 229 calculated from the night-time measurements (dark sky) when a thermal chamber is no available. for each channel between ZSRDSC and ZSRSIM but with some outliers regarding the linear trend (see left 237 panels a, c, e and g). These outliers present higher ZSRDSC values than expected and they could be caused 238 by the presence of clouds in the zenith, instrument malfunction and others.

239
To identify and reject the cloud-contaminated or wrong measurements, several parameters have 240 been considered in this subsection: SZA, ZEN error, temperature, and the time interval between the 241 inversion used to simulate the ZSRSIM and the corresponding ZSRDSC. Some thresholds have been identified 242 after the visual analysis of these parameters in the scatter plot. For the SZA, the signal of ZEN instrument 243 is higher than the expected for SZA values below 30ᵒ, which could be explained by some stray sun light 244 coming to the sensors due to the high elevation of the Sun. Then, ZSRDSC values recorded under SZA below 245 30ᵒ have been discarded, but also the values with SZA above 80ᵒ due to the low signal registered for this  Figure S3). If the measurement of any channel has a ZEN error above this threshold, 251 then the measurements of the four channels are rejected.

252
No clear dependence of the outliers on the rest of the parameters has been observed. The results

279
The obtained calibration coefficients and the ones obtained by intercomparison with a calibrated 280 integrating sphere at IARC facilities are shown in Table 1. Table 1 also presents the relative differences 281 between both calibration coefficients using the coefficients from IARC as reference; the uncertainty 282 involved in the latter calibration method procedure is estimated to be 5% by Walker et al. (1991) These 283 differences are 1.39%, -6.54%, -6.72% and -5.89% for 440, 500, 675 and 870 nm, respectively. The

284
proposed calibration method uses the standard ASTM-E490 solar spectrum for transforming the unitless 285 output from GRASP, as indicated in Equation 1. This fact can increase the relative differences between the 286 two calibration methods. However, the calibration method employed here allows to use the same 287 normalization factor that lately will be applied to the ZSRZEN measurements that will be introduced in the 288 GRASP inversion module, avoiding to introduce a systematic error due to the normalization required by 289 the algorithm. It means that this calibration method is better suited when using the ZSRZEN values as input 290 for GRASP to retrieve aerosol properties, since there is no need for extraterrestrial spectrum normalization.

291
From now on ZSRZEN will stand for the calibrated zenith sky radiances measured by the ZEN-R52 satisfying 292 the stablished quality controls (30ᵒ < SZA < 80ᵒ; ZEN error < 4%).   The CE318 sun-sky photometer allows to perform measurements in the 'cloud mode' scenario. It

301
is carried out when the direct sun measurement indicates an obscured sun, and therefore the aerosol retrieval  screened CE318 sky radiances measured in the PPL geometry, has been labelled as ZSRPPL.

333
The PPL dataset is not directly available in the AERONET webpage; then, it has been extracted  In order to analyse the capabilities of the proposed inversion strategy to invert ZSRZEN measurements 367 with GRASP, a detailed sensitivity analysis is carried out in this section using synthetic data.

368
As mentioned in Section 2.2.2, the chosen method to obtain aerosols properties, considers five aerosol

442
The reason for the observed overestimation could be in the limitations of the GRASP-ZEN method 443 based on the 'models' approach, which only allows to retrieve aerosol properties within the properties of of fine and coarse modes. Therefore, we will focus on the retrieval of AOD at 440, 500, 675 and 870 nm 468 and VCF, VCC and VCT.

470
Once the ZSRZEN measurements have been obtained from ZEN-R52 by the calibration method

480
The smoothness check is done by the analysis of the AOD variation at 500 nm: for each two subsequent 481 values if the variation is higher than 0.01/min the retrieval with larger AOD at 500 nm in the pair is removed.

482
After the smoothness, the stand-alone check is applied: all single retrievals remaining which are more than

531
The VCT, VCC and VCC values from both datasets are shown in Figure 15 for the week from 16 to 532 22 June 2020 (same days than Figure 12), showing again a similar behaviour for the two datasets. Figure   533 15 also reveals that the GRASP-ZEN values are noisier and overestimates the AERONET values, especially 534 for the fine mode.

535
In order to perform a quantitative analysis for the correlation between VCT, VCC and VCC from

574
The proposed methodology for the calibration of ZEN-R52, using simulated ZSR values has been 575 contrasted, showing discrepancies lower than 6% respect to the calibration coefficients obtained against an 576 integrating sphere. This proposed methodology incorporates the advantage that it includes the 577 normalization used by GRASP and therefore there is not any need to use extraterrestrial spectra to normalize 578 the data when they are used as input in GRASP.

579
A new inversion strategy, called GRASP-ZEN, has been proposed to retrieve aerosol properties with

586
The GRASP-ZEN method has been applied to ZSR measurements recorded with a ZEN-R52