DIRECT, LONGWAVE RADIATIVE FORCING OF MINERAL DUST: IMPROVEMENT OF ITS ESTIMATION BY MEANS OF TOOLS RECENTLY DEVELOPED BY THE EARLINET

The objective of this work is to investigate how the knowledge of the vertically-resolved fine and coarse mode aerosol optical depth modifies the longwave radiative forcing. Since relatively little the EARLINET (European Aerosol Research Lidar Network) community has developed codes that combine sun-photometer and lidar data to retrieve a set of parameters vertically-resolved related to the size distribution (fine and coarse mode extinction coefficients, fine and coarse mode volumetric concentrations, etc.). We concentrate on the case of mineral dust whose size distribution is often dominated by the coarse mode. This work demonstrates that the knowledge of the vertically-resolved fine and coarse mode aerosol optical depth modifies the LW RF as compared to the classical approach with a unique profile of total aerosol optical depth. The results show that when the coarse mode predominates the classical approach underestimates the dust longwave radiative forcing by 10 to 20 % at the surface. The effect at the top of the atmosphere is not systematic because of the predominance of fine particles near the top of the dust layer.


INTRODUCTION
Atmospheric aerosols have a remarkable effect on the Earth-atmosphere radiative budget.Indeed, aerosols and their interactions with clouds contribute to the largest uncertainties in the estimation of the Earth's changing energy budget [1].Nowadays many radiative transfer models have been developed to locally estimate the aerosol direct radiative forcing (RF) [2] [3].In the longwave (LW) spectral range, the aerosol radiative properties are usually estimated theoretically with a Mie code.The parameter that contains the absorption and scattering quantities, the extinction coefficient, is normalized to the extinction coefficient in the shortwave (SW) spectral range, most of the time in the visible spectral range [4] [5], or to the number concentration [6].As measurements of the extinction coefficient or of its integral, the aerosol optical depth (AOD), are available in the shortwave spectral range, the equivalent extinction coefficient or optical depth in the LW spectral range can be deduced thanks to that normalized, theoretical extinction coefficient.
In this work we use codes recently developed by the EARLINET community to separate the contributions of the small and the large particles.It is known that for a constant value of extinction in the visible spectral range, the theoretical extinction value in the LW spectral range (calculated with a Mie code) will vary with particle size.It is then expected that considering small and large particles separately will give extinction values in the LW different than considering the total bimodal distribution as a whole.
This work investigates how the knowledge of the vertically-resolved fine and coarse mode AOD modifies the LW RF usually estimated with a single profile of total AOD.

METHODOLOGY
Irradiances in the longwave spectral region are calculated using the GAME code [7].The generic parameters used as input in the Mie code (particles' shape, refractive index, size distribution and density) and in GAME (atmospheric pressure, temperature and relative humidity, and surface albedo) are described in [5] and [8].
To estimate the aerosol RF in the LW spectral range the profile of the LW AOD is calculated by multiplying the SW AOD profile to the normalized LW extinction coefficient (previously calculated with a Mie code).The single scattering albedo and the asymmetry factor are intensive parameters and are estimated in the LW with a Mie code.
In the following, the altitude dependency has been omitted for the sake of clarity.
In the classical approach, a bimodal size distribution (e.g.measured by a sun-photometer) is considered to calculate the normalized LW extinction coefficient,

RESULTS
The code used to estimate separately the coarse and fine mode fractions of the extinction coefficient and volumetric concentration is LIRIC (LIdar-Radiometer Inversion Code) [9].LIRIC calculates the fine and coarse particle concentration profiles, using the backscattered lidar signals at three wavelengths (355, 532 and 1064 nm) and the column averaged aerosol microphysical properties retrieved from a sun photometer.More information about its performances can be found in [10] and [11].
We use a total of four lidar and AERONET sunphotometer measurements in Barcelona with different loads and vertical distributions of mineral dust.Table 1 gives some description of the four cases in terms of AOD and size distribution, as well as the results of our radiative forcing calculations with both the MSP and the classical methods.More details about the mineral dust vertical distribution of those four cases can be found in [8].

At the surface
 and V C represent the column- averaged volumetric radius, standard deviation and volumetric concentration.The suffixes c and f stand for coarse and fine, respectively.

CONCLUSIONS
The calculation of mineral dust, and aerosols dominated by large particles in general, direct radiative forcing in the longwave spectral range can be improved when both profiles of coarse and fine mode extinction coefficients can be retrieved separately.Our work demonstrates that the classical method underestimates the LW RF at the surface by 10 to 20 %.As the predominance of the large particles decreases with increasing height, this subestimation is not systematic at the TOA.However the strong predominance of large particles at the surface has also an effect on the forcing in the upper layers.
In presence of large particles, the improved estimation of LW RF suggests that the total (SW+LW) forcing is reduced (since both forcings are usually of opposite sign), and that aerosol impact on regional climate near dust sources could be lower than the actual estimations.
ranges from +3.1 to +11.0 Wm -2 .All values of class F  are 10 to 20 % lower than MSP F  .The largest difference, 20 %, is not AOD , (e.g.measured by a lidar) is considered for the SW AOD profile.The profile of the LW AOD is calculated as:

Table 1 .
LW RF calculated by the MSP and the classical methods,