1 Introduction

During and after the application of pesticides, parts of the substances may enter the atmosphere, allowing it to be transported over long distances (Gath et al. 1993). These possible atmospheric entries result from spray drift, dust drift, volatilisation and emission of plant protection products (FOCUS 2008; Biocca et al. 2011). Consequently, pesticide residues can be found in soil and plants as well as on the surface of soils and plants in various distances to agricultural areas (Hofmann et al. 2018; Hofmann et al. 2019; Linhart et al. 2019; Kruse-Plaß et al. 2021). For more than 30 years, it is well known that pesticides and their metabolites can enter the atmosphere and may be subject to aerial transport over shorter and longer distances (Gath et al. 1992, 1993). Findings of prosulfocarb and pendimethalin residues in non-treated crops (Seiler et al. 2007) led to a decision of the German council of agriculture ministers to further examine the areal transport of pesticides for longer distances (Agrarministerkonferenz 2015). Starting a German-wide pesticide monitoring to investigate aerial substance concentrations at different distances from treated fields, and measuring the deposition of the compounds on soil and plants for pesticides was considered the best option to collect robust information.

Therefore, a feasibility study initiated by the Federal Office of Consumer Protection and Food Safety (BVL) was carried out in order to analyse if a representative German-wide monitoring could evaluate the findings of pesticides in non-target areas. (Bolz and Kubiak 2020). The next step was a Germany-wide analysis of suitable monitoring areas, which is described here. The aim of this study was to select monitoring areas and sites which cover the conditions of different types land of use and climate conditions in Germany, by using a geographical information system (GIS) and different digital geo data layers such as different meteorological parameters, land use and the pesticide use intensity. Based on this, a map of Germany with its average climate conditions as well as its agricultural treatment picture was created. Since the aim of this study was to find suitable monitoring sites for a national air monitoring, the goal was to find the most efficient approach for a monitoring, meaning to integrate this monitoring into an already existing measuring network. Therefore, the locations of the German agro-meteorological network were used.

2 Material and methods

Various measurement techniques (particularly active air sampler) will be necessary for a valid data situation in order to record the entire range of exposure. A structured air monitoring with specific measuring devices will be necessary in order to be able to make statements about the extent of the transport of plant protection products and to be able to take appropriate measures. In active air samplers, air is pumped through a filter and then through a solid adsorbent. Pesticides in the particle phase are removed from the air by the filter. With active air samplers, the air concentration can be determined and thus the temporal and spatial changes in the concentrations of pesticides can be tracked. Bulk collectors record both wet deposition by rain (which washes substances dissolved in the gas phase and particle-bound from the air) and dry deposition. With active biomonitoring, plants that have been brought forward would be placed at various measuring points. Kale plants, for example, are suitable for such biomonitoring. Due to the growth form of the kale, the leaves can be well bathed in water. They also have a large surface due to their curling and the mostly lipophilic substances can accumulate well due to the pronounced wax layer of the leaves. The results obtained with kale would correlate well with accumulations in other plants, which means that kale is well suited as a representative plant for biomonitoring. Therefore, the measuring sites for a structured air monitoring would include active air samplers, bulk samplers as well as measuring sites for an active biomonitoring. Furthermore, the meteorological stations were chosen at different distances (0–100, 100–1000 m, and > 1 km) to the treated fields.

All geographical analyses were based on:

  1. 1.

    stations of the agrometeorological monitoring network and existing air monitoring stations in individual Federal States (Bavaria, Brandenburg and North Rhine-Westphalia),

  2. 2.

    selected climate data of the German Weather Service (DWD),

  3. 3.

    land use data provided by the Thünen Institute in combination with data of plant protection product (PPP) treatment intensities originating from the “Panel of Plant Protection Applications” (PAPA) and from the "Network of Comparative Farms Crop Protection Network" of the Julius-Kühn Institute (JKI), as well as

  4. 4.

    geodata of the official German Topographical Cartographic Information System (ATKIS).

The basic data were stored in an object-relational PostgreSQL database with the extension PostGIS storage. PostgreSQL with PostGIS forms a geodatabase, which allows to store geographic objects and to integrate them into other GIS. Moreover, multiple spatial functions, operators and analyses were enabled with Post-GIS in combination with functionalities of the programming language PYTHON 3 to implement all data analysis.

2.1 Climate areas

For the determination of the prevailing climatic conditions during the vegetation period (March–August), data starting from the early 1990s of the German weather service were used. In detail, air temperatures, wind velocities, precipitation and evapotranspiration were available as Germany-wide raster datasets with a spatial pixel-resolution of 1 × 1 km2. Using the clustering algorithm “k-Means” with the extension k-Means++ (Sergei and David 2006), the country was grouped into 9 climate regions with different climate conditions. Based on the “elbow” method, 9 clusters were found to be optimal for this meteorological data analysis. The specific distance determination to the nearest agricultural area for each weather station, as well as the fraction quantification for the agricultural area, as part of the total area in the relevant upstream 45° sector with a radius of 5000 m, was based on the information provided by ATKIS landscape model, which is a polygon vector dataset with the scale of 1: 25,000. The distance was calculated using the minimum Euclidean distance between the XY point coordinate of each station and the nearest ATKIS agricultural area in a 45° sector in the mean main wind direction. The percentage of area per station corresponds to the percentage of ATKIS agricultural area of this sector.

2.2 Intensity of pesticide use

To determine areas with comparable pesticide treatment intensities, 6 treatment classes (TI) were used based on data from 2017 to 2019 collected by the Julius–Kühn Institute and the Thünen-Institute. The classes were delimited by the number of pesticide applications per year:

  • class 0 (< 2.5 applications),

  • class 1 (2.5–5 applications),

  • class 2 (5–10 applications),

  • class 3 (10–15 applications),

  • class 4 (15–20 applications),

  • class 5 (> 20 applications).

The classes were defined by experts. Class limits were set with the aim to reduce the data amount while still be able to distinguish areas which are dominated by general crop classes like cereals, vegetables, grapevine, hops and orchards. In order to evaluate the suitability of the monitoring site, for each available measuring station the mean TI value based on a 45° sector with a radius of 5000 m in main wind direction (time period 1995–2010) was calculated and classified into the predefined 6 TI classes. The TI crop type grids correspond to the maps of the agricultural landscape in Germany and were developed in cooperation with the Humboldt University of Berlin (HU) and the Leibniz-Centre for Agricultural Landscape Research (ZALF), based on satellite data of the Copernicus program. The JKI treatment intensity data include information from the Panel of Plant Protection Applications (PAPA) on the actual application of chemical PPPs in agriculture during the years 2017–2019 or data from the "Comparison farms crop protection network" (JKI 2017). The PAPA data are compiled for crop-specific networks of survey farms in detail in anonymized form for a selection of 9 crops (winter wheat, winter barley, winter rye, corn, potatoes, sugar beets, desert apple, hops and grapevine). The data from the "Network of Comparative Farms Crop Protection Network" correspond to the annual collection of data on the application of crop protection agents in main crops, and other information relevant for plant protection on representative farms. Within the framework of this farm network, treatment indices (TI) were determined for individual vegetable varieties (carrots, cabbage, asparagus, onions).

3 Results and discussion

The GIS analysis showed some areas with high application intensity corresponding to the orchards, hops and vineyard areas in Germany. The underlying pesticide treatment index shows the spatial distribution of the 6 treatment index classes defined by expert judgement (Table 1; Fig. 1).

Table 1 Pesticide treatment intensity of agricultural products in Germany (2017–2019)
Fig. 1
figure 1

Spatial distribution of six treatment intensity classes within agricultural areas in Germany (left), and the9 identified climate clusters (right)

Based on information from the German weather service,9 different climatic areas in Germany could be identified using monthly air temperature, precipitation, wind speed, and real evapotranspiration over grass and sandy loam (Fig. 1). These analyses led to an overlay of the GIS pesticide treatment layer and the calculated climate cluster (CC) layer and resulted in the Germany-wide map with 54 possible combinations of treatment and climate. Figure 1 only separates the distribution of TI-classes and climate clusters, since inside the map of all the occurring TI/CC-combinations it is visually hard to identify all combinations. Fifteen combinations had an area share > 1%, 14 of these combinations had a TI class of 1 or 2. Both TI classes occur in almost all climate clusters and cover ~ 90% of the agricultural area in Germany. TI classes representing high treatment intensities (3, 4 and 5) cover ~ 5% of the area. The remaining ~ 5% area had a TI class of 0 (i.e., low intensity) and was found in nearly all climate clusters. Combinations not found are particularly explained by climatic cluster 4. This corresponds to the mountain climate of the high alpine areas of the Swabian-Bavarian Pre-Alps, in which agricultural areas with a TI value are present only in very low quantity.

For the selection of TI/CC combinations for the air monitoring of pesticides, TI/CC combinations that were particularly representative for Germany in terms of area fraction and additionally, different treatment intensities were considered. Moreover, the 3 distance classes < 100 m, 100–1000 m and > 1000 m between the agricultural areas and sample locations were covered in the subsequent selection of suitable measurement sites within these TI/CC combinations (Fig. 2). This finally, led to a selection of 5 TI/CC combinations, in which 6 stations (2 per distance class) were proposed for a German-wide air monitoring. Based on this, the monitoring concept comprised 30 stations, which was considered a well manageable scope in practice. The stations identified and proposed for sample collection were selected based on the GIS analysis (Fig. 2).

Fig. 2
figure 2

Sampling sites considering different distances to treated fields and covering the relevant combinations of pesticide treatment index and climatic conditions

4 Conclusions

The amount of pesticide applications (treatment index) depends on the crop as well as on the climatic conditions. The GIS analysis enabled the finding of monitoring clusters which are representative for different pesticide applications in Germany. In combination with the network of the agro-meteorological weather stations, 30 collection locations were identified, reflecting all relevant scenarios for aerial pesticide transport in Germany. For a representative monitoring, 9 representative climatic areas and 6 representative agricultural areas (identified via treatment index classes) in Germany were identified. At least 5 of these representative combinations of climate classes and treatment index classes should be used for a German wide monitoring.

Furthermore, for the monitoring sites, 3 different distance classes ("close range" < 100 m, "medium range" 100 − 1000 m and "long range" > 1000 m) should be covered. With the described design, this air monitoring would only allow a correlation with the results for medium range transport. However, since the correlation with long range transport would be the most difficult, because the interference of many parameters would have to be taken into account, it was decided to go forward with the described design as the most practical approach. Furthermore, 2 stations per distance class to the next agricultural area in the main wind direction were proposed for the 5 representative combinations of climate areas and index classes, which means that at least 30 monitoring sites should be used for a German wide air monitoring of pesticides. After the results are available, the evaluation of the data takes on a special role. Premature conclusions should be avoided. If conspicuous clusters are found, the findings should be clarified in further steps and other sources of input should be considered.