Models for monitoring and forecasting

Integrated bark beetle management consists of three pillars: prevention (reducing infestation risk), monitoring (surveying the situation), and reactive measures (mitigating infestation impacts) (FVA, 2024). Monitoring, of e.g. swarming activity, beetle development, and infestation risk, provides the necessary up-to-date information for often time-critical reactive measures for sanitation like detection, cutting, and removal of infested wood. Until now, pheromone traps and brood observation trees at selected monitoring sites have provided indications of the current threat level. However, these data have several shortcomings compared to models, which limit their application:

  1. They are only available at specific locations and thus provide insufficient information for the entire forest area.
  2. Trap data reflect swarming activity but not the actual infestation risk.
  3. Weekly observations only provide retrospective information – valuable time gets lost, and forecasting is not possible.
  4. The data are not integrated into a digital, comprehensive information system that forest owners can directly access.

For these reasons, models will play an increasingly important role in monitoring and forecasting within future bark beetle management. The goal is to support forestry practitioners with better information through PHENIPS-Clim and IpsRisk, thereby optimizing resource efficiency in infestation detection and sanitation. As a consequence, effective management means smaller areas of damage and less damaged wood.

The creation of such models is based on detailed knowledge of the bark beetle biology and the factors that influence the risk of infestation. These insights have been gathered over recent years through various research projects at the FVA or compiled from literature, and are incorporated into the models in the form of mathematical functions (Fig. 1). An example of this is the temperature-driven development of bark beetles (HOFMANN ET AL, 2024).

On the other hand, comprehensive input data are required to drive the models with sufficiently high temporal and spatial resolution, such as daily weather data from the Deutscher Wetterdienst (DWD, German Weather Service) as well as data on site conditions, forest stands, and previous infestations. Based on these components, model outputs are calculated on a grid basis and made available to forestry practitioners in the form of maps or site-specific diagrams that are updated daily, including a 7-day forecast.

In addition to providing short-term decision-making support, these models also enable the calculation of long-term trends resulting from increasing climate change - for example, regarding the activity period and number of beetle generations, or the drought stress that favors beetle infestation in spruce. These trends ultimately serve as essential planning tools for forestry practitioners to assess the medium- to long-term suitability and vulnerability of spruce in the future forests.

 

Scheme of the daily workflow for the individual raster cells.

Fig.  1: Scheme of the daily workflow for the individual raster cells.

The phenology model PHENIPS-Clim

This model provides raster-based (1 km x 1 km) information on the current developmental stage of the individual bark beetle generations during the bark beetle season, as well as on the number of generations established to hibernate at the end of the season. 

Such a phenology model only requires input data on weather conditions (temperature, global radiation) and daylength to calculate beetle-related parameters such as the start of swarming and infestation in spring, the development from egg, larval and pupal stages to the mature juvenile beetle, or the end of brood establishment in late summer/autumn. The model thus provides information about beetle development within a bark beetle population at a specific location (Fig. 2), but not – and this is an important distinction – about the infestation risk there, which depends on additional factors.

PHENIPS-Clim graphics exemplary for three points in time for Baden-Württemberg (maps above) as well as a single raster cell (approx. 960 m above sea level; marked with an asterisk on the map) over time in 2024; dotted areas on the map indicate the establishment of a new generation in the next 7 days; at the example site a maximum of two generations have developed; the most important information is labeled (F1 / F2 / F3 = 1st / 2nd / 3rd generation).
PHENIPS-Clim graphics exemplary for three points in time for Baden-Württemberg (maps above) as well as a single raster cell (approx. 960 m above sea level; marked with an asterisk on the map) over time in 2024; dotted areas on the map indicate the establishment of a new generation in the next 7 days; at the example site a maximum of two generations have developed; the most important information is labeled (F1 / F2 / F3 = 1st / 2nd / 3rd generation).

Fig. 2: PHENIPS-Clim graphics exemplary for three points in time for Baden-Württemberg (maps above) as well as a single raster cell (approx. 960 m above sea level; marked with an asterisk on the map) over time in 2024; dotted areas on the map indicate the establishment of a new generation in the next 7 days; at the example site a maximum of two generations have developed; the most important information is labeled (F1 / F2 / F3 = 1st / 2nd / 3rd generation).

PHENIPS-Clim was developed at the FVA based on an existing model (PHENIPS: BAIER ET AL., 2007) and new research findings. Adjusted parameters ensure improved model quality, especially under the changing climate conditions in Central Europe. The model thus sets new standards for application in forestry practice and research. For instance, early swarming periods in spring (as observed in 2024) can be well predicted for the first time, as well as late brood establishments due to warm weather in autumn (as in 2018). Furthermore, PHENIPS-Clim takes into account fluctuating day temperatures for calculating development speed, as well as frost-related mortality of the pre-imaginal developmental stages (egg, larva, pupa).

What time-critical management-relevant information can be derived from PHENIPS-Clim for forestry practice?

The model helps to:

  • start infestation surveys in spring on time by predicting the beginning of swarming and infestation
  • coordinate the implementation of measures for sanitation and rendering infested wood harmless during the bark beetle season by indicating the current maximum stage of development (e.g. debarking before transition to the juvenile beetle stage, removal before the next generation begins to emerge)
  • guide infestation surveys in autumn by predicting late brood establishments (e.g., searching for bore dust, frequency of surveys)

The dynamic infestation risk model IpsRisk

Models for estimating bark beetle infestation risk already exist, but they mostly focus on identifying infestation factors (e.g., OVERBECK & SCHMIDT, 2012) or serve to assess the basic (static) infestation predisposition (e.g., NETHERER & NOPP-MAYR, 2005). More recent developments, however, such as in the Czech Republic (PIRTSKHALAVA-KARPOVA ET AL., 2024) and Austria (BOKU / BFW, 2024), increasingly aim at a dynamization of the risk assessment in order to be able to flexibly guide bark beetle management during the season. With IpsRisk, forestry practitioners now have access to a spatially and temporally high-resolution dynamic risk model for the first time. It considers risk factors of varying dynamics for both bark beetles and spruce trees (Fig. 3): Based on daily weather data (DWD), beetle phenology (PHENIPS-Clim) and swarming activity are calculated, while spruce drought stress is determined using site data as well. The factor of previous infestation is based on annual infestation reports, and spruce predisposition is derived from stand data.

The input data for IpsRisk include relevant bark beetle and spruce-related factors for each 250 m raster cell at different temporal resolutions. The model output includes the basic risks "Previous infestation" (beetles) and "Susceptible spruce trees" (stand; outlined in red), as well as the dynamic risk consisting of three components: beetle "Phenology" and "Swarming activity", as well as "Drought stress" of spruce trees (outlined in purple).

Fig. 3: The input data for IpsRisk include relevant bark beetle and spruce-related factors for each 250 m raster cell at different temporal resolutions. The model output includes the basic risks "Previous infestation" (beetles) and "Susceptible spruce trees" (stand; outlined in red), as well as the dynamic risk consisting of three components: beetle "Phenology" and "Swarming activity", as well as "Drought stress" of spruce trees (outlined in purple).

IpsRisk was parameterized using multi-year datasets from across Baden-Württemberg and validated using independent infestation data. The model output provides users three risk components in a 250 m x 250 m grid (Fig. 3):

  • beetle-related basic risk (previous infestation),
  • stand-related basic risk (susceptible spruce trees), and a
  • dynamic risk, consisting of beetle phenology and swarming activity, as well as drought stress of spruce trees.

The two basic risks (= predisposition) are calculated once at the beginning of the season and remain constant throughout. They are displayed as separate map layers as the generally most important factors promoting infestation (Fig. 4).

The basic risk output by IpsRisk at the beginning of the season, "Previous infestation" (left; previously infested areas in purple) and "Susceptible spruce trees" (right), shown as examples for Baden-Württemberg (top) and a zoomed-in section south of Freiburg (bottom); the spatially endangered stands are recognizable.

Fig. 4: The basic risk output by IpsRisk at the beginning of the season, "Previous infestation" (left; previously infested areas in purple) and "Susceptible spruce trees" (right), shown as examples for Baden-Württemberg (top) and a zoomed-in section south of Freiburg (bottom); the spatially endangered stands are recognizable.

The representation of dynamic risk is color-coded in 4 graduated risk classes: none - low - medium - high risk (Fig. 5). This risk is recalculated daily and changes accordingly during the season. To avoid a potentially daily changing risk display, the representation always corresponds to the highest risk at this location during the last 7 days. This ensures practicality in application. The dynamic risk is available via two layers, which can be usefully displayed in combination: the current risk (4 color levels) and the 7-day forecast (dotted when an increase is predicted). In addition to the map display, it is also possible to display a localized diagram that visualizes the temporal progression (Fig. 6). However, this is currently only possible via the FVA-Bark beetle-Portal (updated weekly there for representative monitoring sites).

The dynamic infestation risk calculated by IpsRisk, shown as an example for Baden-Württemberg (top) and a zoomed-in section south of Freiburg (bottom) on June 10, 2024 (left) and June 24, 2024 (right). The prioritized stands for each point in time and the increase in risk over the two weeks are clearly visible. The dotted area in the bottom-right image indicates a predicted increase in risk within the next 7 days; for this, the layers for current and forecasted risk were overlaid. White areas represent a low proportion (<10%) of susceptible spruce. The asterisk marks the raster cell (approx. 960 m above sea level; see Fig. 2) for which the diagram is shown as an example in Fig. 6.

Fig. 5: The dynamic infestation risk calculated by IpsRisk, shown as an example for Baden-Württemberg (top) and a zoomed-in section south of Freiburg (bottom) on June 10, 2024 (left) and June 24, 2024 (right). The prioritized stands for each point in time and the increase in risk over the two weeks are clearly visible. The dotted area in the bottom-right image indicates a predicted increase in risk within the next 7 days; for this, the layers for current and forecasted risk were overlaid. White areas represent a low proportion (<10%) of susceptible spruce. The asterisk marks the raster cell (approx. 960 m above sea level; see Fig. 2) for which the diagram is shown as an example in Fig. 6.

The dynamic infestation risk calculated by IpsRisk over time, shown for an example raster cell (see Fig. 5) in 2024. The points in time shown in Fig. 5 are marked here with arrows. The gray-shaded area represents the period before swarming begins (here early April), and after the end of the brood establishment (plus delay); the colors of the risk classes correspond to the map representation in Fig. 5.

Fig. 6: The dynamic infestation risk calculated by IpsRisk over time, shown for an example raster cell (see Fig. 5) in 2024. The points in time shown in Fig. 5 are marked here with arrows. The gray-shaded area represents the period before swarming begins (here early April), and after the end of the brood establishment (plus delay); the colors of the risk classes correspond to the map representation in Fig. 5.

In addition to the individual risk components explained so far, IpsRisk also offers the option to display these components in combination. For example, it is possible to show dynamic risk only for areas with an elevated basic risk (Fig. 7).

Combined display by activating multiple layers: dynamic risk on June 24, 2024 masked to areas with elevated basic risk "Previous infestation" (left) and "Susceptible spruce trees" (right). It is also possible to combine dynamic risk with both basic risks - the highest possible level of focus. The colors of the risk classes correspond to the map representation in Fig. 5. The asterisk in the zoomed-in maps marks the raster cell (approx. 960 m above sea level) for which the diagram is shown as an example in Fig. 6.

Fig. 7: Combined display by activating multiple layers: dynamic risk on June 24, 2024 masked to areas with elevated basic risk "Previous infestation" (left) and "Susceptible spruce trees" (right). It is also possible to combine dynamic risk with both basic risks - the highest possible level of focus. The colors of the risk classes correspond to the map representation in Fig. 5. The asterisk in the zoomed-in maps marks the raster cell (approx. 960 m above sea level) for which the diagram is shown as an example in Fig. 6.

What time-critical management-relevant information can be derived from IpsRisk for forestry practice?

The model helps to:

  • identify areas with elevated basic risk, on which infestation surveys should focus when resources are limited
  • promptly identify areas and time periods with elevated dynamic risk, on which infestation surveys should focus when resources are limited
  • use available resources more efficiently overall, detect recent infestations earlier, and ultimately minimize damaged areas

IpsRisk does not indicate infestation but rather the risk of infestation. Therefore, infestation can, in principle, also occur in areas with low indicated risk – it is just less likely. For this reason, it is recommended not to leave areas with low or medium risk completely unchecked. The overall effort required for infestation monitoring is not necessarily reduced by using the IpsRisk decision-support tool but is primarily better guided. While the basic risks can be used particularly for spatial prioritization (or for locating overwintering trees before swarming begins), the dynamic component additionally aids in temporal prioritization during the season, for example, within areas with elevated basic risk. The model can be continuously improved in the future by incorporating additional or more precise input data, such as high-resolution infestation data from future years or currently unavailable data like satellite-detected windthrow. Incorporating feedback from practitioners into potential future developments is also planned.

Where is the information available?

Forest information systems

In Baden-Württemberg, both models, PHENIPS-Clim and IpsRisk, are available from April 2025 as WMS (Web Map Service) in the form of additional layers as part of the public geoportal and established digital forestry information systems (Fig. 8). The integration of the models thus offers forestry practitioners immediate access for quick and precise decisions, i.e., directly via tablet on-site, updated daily, zoomable, and possibly in combination with other layers. Similar developments are possible for Rhineland-Palatinate and Saarland in the future.

The provision of the model output as WMS enables integration, for example, into the public geoportal of Baden-Württemberg (top) as well as various forest-specific applications such as WaldExpert (right)

Fig. 8: The provision of the model output as WMS enables integration, for example, into the public geoportal of Baden-Württemberg (top) as well as various forest-specific applications such as WaldExpert (right) (Image sources: www.geoportal-bw.de; https://waldexpert.waldportal-bw.de).

FVA-Bark beetle-Portal

PHENIPS-Clim has been available for Baden-Württemberg, Rhineland-Palatinate, and Saarland since 2024 through the FVA-Bark beetle-Portal, updated weekly in the form of statewide overview maps and diagrams at representative monitoring sites. With the start of the 2025 season, for Baden-Württemberg IpsRisk is also available in a similar format on the portal (under the "Infestation Risk" button). The model information is thus directly linked to the weekly collected data from trap and brood tree monitoring ("Daten"), as well as a textual interpretation ("Aktuelle Situation") as part of the FVA practical advisory service.

R-package "barrks"

For advanced applications, such as independent simulations, PHENIPS-Clim is available within the R-packages "barrks" (JENTSCHKE, 2025). The "barrks" package also includes alternative development models for the European spruce bark beetle and European six-toothed spruce bark beetle, including a description of the various model functionalities and comparative example simulations. The publication of IpsRisk as an R-package is also planned.