BIOMASS BURNING MEASUREMENTS IN EARLINET

The biomass burning events are analyzed using the EARLINET-ACTRIS atmospheric profiling of aerosols using lidars. The period of 2008-2017 was chosen to analyze all the events assigned in the EARLINET database under Forest Fire category. A number of fourteen stations were considered. The data provided, ranged from complete data sets (backscatter, extinction and particle linear depolarization ratio) to single profiles (backscatter coefficient). A thorough quality control was performed. Smoke layers geometry was evaluated and the mean properties within each layer were computed. The Hysplit backward-trajectory technique and the FIRMS fire database were used to double check the source of each layer. Discussions were made under the following scenarios: fire events seen by two stations,


INTRODUCTION
Biomass burning (BB) represents an important source of atmospheric particles (aerosols) which influences the radiative transfer in the atmosphere with in general, a negative effect over the globe [1]. At the ground level or within PBL (Planetary Boundary Layer), strong BB episodes can cause large reduction of the visibility over various regions [2], which can affect the traffic and more importantly can cause serious health issues for humans through the low air quality [3][4]. The monitoring of the wild forest fires events occurring each year in Europe are reported since 2000 in the annual reports from European Commission [5]. Lidars can provide smoke measurements being able to deliver the boundaries of the smoke layers and the optical properties therein. Both the layer geometry and optical properties can be used for validation of the transport model output [6]. The optical properties can be also used in closure studies to model the microphysical properties of the aerosol [7]. Regarding the impact of BB aerosol on weather forecast, the study by Zhang et al. [8] suggests that the BB' effect is seen for AOD (aerosol EPJ Web Conferences 237, 05005 (2020) ILRC 29 https://doi.org/10.1051/epjconf/202023705005 optical depth) at 550nm larger than 1. Within this context, the use of ACTRIS-EARLINET database will bring more insights about the BB aerosol on both spatial and temporal evolution. Selection of results shown, focuses on particular events and one example of geographical clusters.

The EARLINET database
The EARLINET database consists currently of 31 lidar stations covering most of Europe and one location outside Europe. A review of the EARLINET network is given by Pappalardo et al. [9]. There are several categories of submitted data file, one of them being Forest Fire (FF), which are built based on voluntary basis, as opposed to Climatology and Calipso categories. For the present study, we chose the 2008-2017 interval, as most of the data are submitted after 2008, while Hysplit model [10] is available from late 2007. We considered only the data with emission wavelength at 355, 532 and 1064 nm. The initial data set contained ~3700 files at 20 stations.

Data quality control
Several criteria were implemented for data quality control (QC). Firstly, the EARLINET tools on QC were used to eliminate the b (backscatter) and e (extinction) files which did not pass the checks on technical characteristics or the optical properties values [11]. Secondly, additional data checks were performed in-house. After preliminary QC, ~ 2300 files from 14 stations were considered.

Evaluation Procedure
x Firstly, an automatic algorithm was developed to estimate the boundaries of the aerosol pollution layers visible in the lidar profiles. Secondly, a manual check was performed over the automatic boundaries' retrievals. When the automatic retrievals were not accurate, the boundaries were corrected manually.
x Mean values of optical properties in the layers were computed x NOAA Hysplit backtrajectory and fires location from NASA FIRMS [12] were used to select the pollution layers of BB origin, whereby quite often, several layers with different sources were recorded at once. Only the fires on a 100 km radius along backtrajectory and within r 1h of the airmass pass were selected. The backtrajectories for layers without fires found in this range were not considered in analyses.
x profiles with layers of BB origin are selected x The selected events were discussed based on the intensive parameters (lidar ratio LR, backscatter Ångström exponent BAE, extinction Ångström exponent EAE, particles linear depolarization ratio PDR).
x The measurements were quantified as "pure" (with single fire contributing) or "mixed" (more fires contributing) Please note that all the criteria involved in various steps of the procedure will be discussed in detail during conference. Thus, the smoke from the fire occurring at 00:00 on 26 th (48.171N, 30.622E) arrived in "th" at 13:36 on 29 th at 1260m altitude and in "at" at 08:40 on 31 st at 1742m altitude. The intensive parameters (Fig. 2) were: BAE@532/1064 = 0.52 ("th") and BAE@355/532 = 1.69, BAE@532/1064 = 1.32 ("at"). Smoke was considered "pure" (i.e. with one source) for both stations. Figures 3-4 depict backtrajectories and the fires' location for "th" and "at". As better seen on bottom plots, for each station there was only one fire contributing to the measurement. The smoke was observed after 85h30m in "th" and 128h40m in "at" from the time of occurrence. The common variable was BAE@532/1064. The value in "th" was smaller than the value in "at". The difference can be explained by the slightly different path and travel time.

Long range transport from N America
We identified a number of 24 BB events with long range transport (LRT) from N America. The strongest event was recorded by three stations (Belsk -be, Cabauw-ca and Warsaw -wa) during 8-10 July 2013 with the main fires occurring between 30 June to 7 July. Some of those measurements were discussed in literature (e.g. [13][14]). This event will be discussed from a larger perspective during the conference. The statistics of the intensive parameters of LRT from

Geographical clusters
The EARLINET stations considered in BB study were split into four geographical clusters: SE Europe (Athens-at, Bucharest-bu, Potenza-po, Sofia-sf and Thessaloniki-th), SW Europe (Barcelona-ba, Evora-ev, Granada-gr), NE Europe (Belsk-be, Minsk-mi and Warsaw-wa), Central Europe (Cabauw-ca, Leipzig-le, Observatory Hohenpeissenberg-oh). As it will be shown during the conference, each of those clusters measures predominantly smoke from fires occurring in a specific region. As an example, SE Europe cluster is shown in Fig. 5. The stations from this region record BB events with the major source in SE Europe. Each cluster will be discussed in connection with the main meteorological features of each region.

CONCLUSIONS
The Forest Fire category of the EARLINET -ACTRIS database allows to perform a variety of studies on biomass burning aerosols. One can exploit fires' smoke as i) recorded by several stations, ii) long range transport (e.g. N America or W Asia), on the basis of iii) statistics on geographical clusters, and iv) the cases with complicated meteorology when one station can measure mixed BB aerosols, from different sources.