Lidar backscattering

Lidar backscattering analysis (continued)

4.3 Hygroscopic factor

To better evaluate the changes in aerosol optical properties with increasing RH, a scatterplot of aerosol backscattering versus relative humidity (humidogram) is shown in Figure 18. The data suggests a general increase of aerosol backscatter with relative humidity, but there is quite a large scatter in the data. Consequently, it is impossible to determine a relationship between backscatter and RH with any statistical significance.

Fig. 18. Scatterplot of aerosol backscatter versus relative humidity from the ARM/SGP Raman lidar on April 3^rd 1998, for data collected from 1800 to 2400 UTC.

To evaluate if this high variability in the results are related to observed noise in the Raman lidar derived RH, another method is used to create RH profiles below cloud base. In-situ water vapor mixing ratio, temperature and pressure measurements from sensors located at 60m on the ARM/SGP instrumented tower are used to build RH profiles in the boundary layer. Assuming a well-mixed boundary layer, water vapor mixing ratio is assumed constant with height while temperature is assumed to decrease at the dry adiabatic lapse rate. The pressure profile is determined using the hydrostatic relation. Then the saturation water vapor pressure is determined from the temperature profile using a relationship proposed by Buck (1981)

, (5)

where T is the temperature in ^oC. The saturation water vapor mixing ratio can then be determined with

, (6)

where p is the atmospheric pressure and is equal to 0.622. The relative humidity is then determined with

, (7)

where is the water vapor mixing ratio tower measurement. Thus RH profiles are determined using 10-minute averaged tower measurements corresponding to times where lidar backscatter estimates are available. Tower water vapor mixing ratio observations were corrected by subtracting 0.4 g/kg in order to obtain a better agreement between heights at which the constructed RH reach 100% and cloud base heights determined with the lidar. This is an acceptable correction as its magnitude is small enough to be considered within the acceptable level of calibration error of the humidity sensor. Examples of the estimated RH profiles are shown in Figure 19. It can be seen that these profiles are much smoother than those derived using the lidar, and the variation of RH with height agrees very well with the soundings. The derived RH profiles show higher relative humidities than the soundings by about 5%. But RH values of the soundings seem to have a dry bias as saturation is not reached near cloud base. These differences are well within the possible calibration errors of the instruments.

A new aerosol backscattering versus RH humidogram is produced using RH profiles determined from surface (tower) data (Figure 20). The resulting relationship is now much more compatible with results obtained by Wulfmeyer and Feingold (2000) in a marine boundary layer. A clearly defined increase in aerosol backscattering for RH larger than 85% is now observed. Largest increases in backscatter occur for RH larger than 95%. These results are in good qualitative agreement with theory (section 3) and the results of Wulfmeyer and Feingold (2000).

Fig. 19. Profiles of relative humidity derived from tower measurements and assuming a well-mixed boundary layer, compared to available balloon soundings. ARM/SGP site on April 3rd 1998.

Fig. 20. Scatterplot of aerosol backscatter from the ARM/SGP Raman lidar versus relative humidity derived from tower measurements. April 3rd 1998, data collected from 1800 to 2400 UTC.

Even though the expected behavior of aerosol backscattering is more clearly observed in Figure 20, significant scatter of the data is again present. To minimize this scatter, the time window during which data are considered is reduced. When data from 2100 UTC to 2400 UTC are considered (instead of 1800 to 2400 UTC), the resulting humidogram is shown in Figure 21. An even clearer relationship between aerosol backscatter and relative humidity emerges, illustrating the important increase in aerosol backscatter for RH close to saturation.

Fig. 21. Scatterplot of aerosol backscatter from the ARM/SGP Raman lidar versus relative humidity derived from tower measurements. April 3^rd 1998, data collected from 2100 to 2400 UTC.

To gain a more qualitative assessment of the hygroscopic effect on aerosol backscatter, the normalized backscattering can be calculated by taking the ratio of aerosol backscatter over the mean backscatter observed for a reference relative humidity ( ).
Here, is taken for a relative humidity of 70%. From the data, we determined that = 0.0005 (km-sr)^-1. The backscatter data from Figure 21 is normalized with this value and the resulting normalized humidogram is shown in Figure 22. It is seen that aerosol backscatter increases by a factor larger than 3 for RH>95%, from backscatter values at 70%.

Fig. 22. Scatterplot of normalized aerosol backscatter from the ARM/SGP Raman lidar versus relative humidity derived from tower measurements. A best fit to the data is also shown. April 3^rd 1998, data collected from 2100 to 2400 UTC.

As in Im et al. (2001), the data is fitted with a relationship of the form

, (8)

The best fit to the data (in a least square sense), yields values of the regression coefficients of a=0.43 and b=0.72, with an R²equal to 0.85 indicating a good fit to the data. The resulting regression is shown in Figure 21 as the solid red line. Im et al. (2001) studied the hygroscopic growth factor over the East coast of the United States using humidified nephelometer data. They used a coefficient of a=1 in their fit to the data. There results were that the b coefficient seemed to be insensitive to the type of air mass affecting the region. They obtained b=0.38 for polluted continental, continental and maritime air masses. The difference between our results and theirs may be due to the fact that they used a reference relative humidity of 30%, compared to 70% use in our study. We could not reproduce the same analysis, as the lowest value of RH present in our dataset is about 63%. The relationship obtained by Im et al., adjusted so that the normalized backscatter is close to 1.0 for RH=70% but keeping the original exponent, is also shown on Figure 22 (solid green line). Is is clear that our data suggest a much “steeper” relationship between aerosol backscatter and RH as compared to Im et al. This difference may possibly be explained by the fact that the chemical composition of aerosols present in the eastern part of the United States is most likely different that those over the central part of the US. It is a well-known fact that aerosols in the eastern portion of the continent are composed of sulfates. Through a modeling study, Wulfmeyer and Feingold (2000) found that aerosols with larger mass fractions of soluble material exhibit a steeper increase in backscatter compared to partially soluble particles. This leads us to conclude that aerosols over the ARM-SGP CART site are most likely composed of material with a higher degree of solubility than what was observed by Im et al. (2001) over the East Coast of the US. Thus, the results obtained here can be used to gain more insights on the types of aerosols present over the CART site. As in Wulfmeyer and Feingold (2000), models of aerosol growth and changes of refractive index could be used to determine more precisely which types of aerosols could produce the observed change in lidar backscatter as a function of relative humidity. This type of analysis is left for future efforts.