Weather conditions directly influence agricultural yields. Hail, disease and drought can have devastating effects on crops. However meteorological-related risks can be reduced through better timing of harvests, application of pesticides or through use of irrigation systems. A clear picture of current and future weather conditions, along with appropriate farm actions, can increase the likelihood of improved yields.

Climate change also influences crop suitability in certain regions where livestock can be negatively affected by migrating diseases and available food. To complicate matters the agricultural sector is also trying to become more sustainable and environmentally friendly in an attempt to meet greenhouse gas emission targets.

This session intends to examine our increasing knowledge of agricultural meteorology, while also attempting to identify opportunities in our changing environment.

We invite presentations related but not limited to:
• Agrometeorological modelling (e.g. modelling agrometeorological related diseases, frost protection warning methods, drought indices etc.)
• Impact of weather and climate extremes on agriculture
• Methods of measurements and observations (e.g. ground based equipment, remote sensing products, citizen science, Big Data etc.)
• Decision support systems & the representation of uncertainty
• Interactions/feedback of farmers and other end users
• Use of future climate projections on agrometeorological models

Convener: Branislava Lalic | Co-conveners: Josef Eitzinger, Sándor Szalai
Lightning talks
| Thu, 09 Sep, 11:00–12:30 (CEST)

Lightning talks: Thu, 09 Sep

Sen Lu

Soil thermal conductivity (λ) is an important physical property in land surface parameterization. The soil thermal conductivity (λ) and matric suction of soil water (h, the negative of matric potential) relationship has been widely used in land surface models for estimating soil temperature and heat flux following the McCumber and Pielke (1981, MP81) λ-h model. However, few datasets are available for evaluating the accuracy and feasibility of the MP81 λ-h model under various soil and moisture conditions. In this study, we developed a new λ-h model and compared its performance with that of the MP81 model using measurements on 18 soils with a wide range of textures, water contents and bulk densities. The heat pulse technique was used to measure λ, and the suction table, micro-tensiometers, pressure plate device, and the dew point potentiometer were applied to obtain soil water retention curves at the appropriate suction ranges. In the range of pF (the common logarithm of h in cm)≤3, the λ-h relationships were highly nonlinear and varied strongly with soil texture and bulk density. In the dry range (i.e., pF > 3), there existed a universal λ-h relationship for all soil textures and bulk densities, and an exponential function was established to describe the relationship. Independent evaluations using λ-h data on five intact soil samples showed that the new model produced accurate λ data from pF values with root mean square errors (RMSE) with the range of 0.03–0.18Wm−1 K−1. While, large errors (RMSEs within 0.17–0.36Wm−1 K−1) were observed with λ estimates from the MP81 model. 

How to cite: Lu, S.: A generalized model between thermal conductivity and matric suction of soils, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-40, https://doi.org/10.5194/ems2021-40, 2021.

Ana Firanj Sremac, Branislava Lalic, Josef Eitzinger, and Stefan Schneider

Tree phenology dynamic is the direct response of the plant to seasonal environmental conditions. Therefore, most trees have been strongly impacted by the increasing frequency of extreme weather events caused by global warming. In the past decade, fruit production in Europe has suffered a catastrophic failure of the yield because of the extreme and adverse weather events occurring (mostly frost) during winter dormancy and tree flowering. Therefore, forecasting the phenological development on the seasonal time scale can help in the organization of the prevention measures in the upcoming production season.

The Central European apple orchards' phenology dynamics are analyzed using the first bloom data and meteorological measurements collected during the COMBIRISK project. Projection to Latent Structures regression analysis (PLSR) from the ChillR package (https://cran.r-project.org/web/packages/chillR/index.html) is used to analyze two dormancy stages in the phonological development: endodormancy (chilling period) and ecodormancy (forcing period) in order to determine the changes in the phenological development patterns. Flowering is modeled considering chilling and forcing plant requirements and focusing on the temperature stresses impact through the stress factor calibration.

ECMWFs seasonal forecasts (SEAS5) are statistically downscaled by the Austrian national weather service (ZAMG) to a 1 km horizontal grid. These seasonal forecasts are used as input meteorological data for the apple phenology dynamic model in the AGROFORECAST project. Obtained results are compared with the observed timing of flowering to test the efficacy of available seasonal forecast for this application. This study is supported by the Ministry of education, science and technological development of the Republic of Serbia (agreement 451-03-9/2021-14/200117). 

How to cite: Firanj Sremac, A., Lalic, B., Eitzinger, J., and Schneider, S.: Modeling phenology of the Central European apple: The seasonal forecast application perspective, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-123, https://doi.org/10.5194/ems2021-123, 2021.

Branislava Lalić and David Fitzjarrald

The winter into spring and the summer into autumn transition seasons can last several weeks. Leaf emergence in midlatitude climates decreases the ratio of sensible (H) and latent heat (LE) fluxes - the Bowen ratio (B).  Because there are many more surface climate stations than flux-measuring sites, researchers seek to link the state variables at standard climate station heights to the leaf development.  Schwartz (1996) found out that, during the midlatitude onset of spring, the DTR trend rapidly increases for several weeks and then levels off.  Adopting an alternate approach, Fitzjarrald et al. (2001) linked changes in H and LE to B to the state variable daily change tendencies. This approach is based on assuming that the surface climate alters as a small fraction of the surface fluxes converge on average into the lower atmosphere.

Schwartz’ approach has the advantage of not requiring information from directly-measured fluxes, but station’s representativeness during the daytime (Tmax) greatly exceeds the area that the Tmin would describe. What’s more, daytime cloudiness depreses Tmax but nocturnal cloudiness enhances Tmin. The Fitzjarrald et al. approach requires long-term day-to-day averages to determine the times of the year when the surface state variables identify the consequences of leaf emergence.

            Here we seek to refine methods to relate plant characteristics to surface climate state, with emphasis on the spring transition at Harvard Forest (HF, MA, USA). At HF, J. O’Keefe kept a careful log of significant phenological events (Klosterman et al., 2018). The transition to the ‘growing season’ begins with bud break (mid-April), ending with nearly fully leafed crowns ("95%") in most species by mid-May.

We revisited the HF data and found that DTR, from the start of spring transition until the end of autumn, changes along with daily sensible heat flux changes, particularly during the period from sunrise until the daily maximum air temperature occurs. Since the seasonal course of daily temperature Td  (°C) follows the latent heat flux trend, we normalized the DTR (DTR/Td) and found that DTR/Td ≈ 1 approximately at budbreak and again at "95%”. When the DTR next approaches Td, the autumn transition is beginning. We use the METAR data to identify cloudy/clear periods and assess the sensitivity of DTR to this effect.

We examined the utility of using DTR/Td ratio as an indicator of spring and autumn transition, exploring  temperature measurements and phenological observations from the HF and PIS network (Lalic et al., 2020). Preliminary results indicate that this approach can identify significant effects of leaf state on local surface climate without the need for averaging over a decade or longer.

Fitzjarrald et al., 2001, 1175/1520-0442(2001)014<0598:CCOLPI>2.0.CO;2.

Klosterman et al., 2018, 1007/s00484-018-1564-9

Lalic et al., 2020, 1007/978-3-030-37421-1.

Schwartz, 1996, 1175/1520-0442(1996)009<0803:ETSDID>2.0.CO;2

How to cite: Lalić, B. and Fitzjarrald, D.: Bowen ratio and daily temperature range thresholds: Are they signals of transient seasons?, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-203, https://doi.org/10.5194/ems2021-203, 2021.

Rudrani Gajraj and Josef Eitzinger

Projections of crop-pest dynamics under climate change (CC) impact negatively throughout the 21st century, suggesting increased pest pressure over central Europe. Variability in environmental conditions significantly affects the spread, abundance, and management of maize pests, such as Western corn rootworm (WCR), European corn borer (ECB) and wireworms, as indicated from the past monitoring data and quantitative pest models. Regional climate change and landscape factors are modifying the development of these thermophile insects and their impact on crops. Several conceptual modelling works related to pest population dynamics, phenology models, spread and risk mapping have been proposed. The challenges become complex for WCR and wireworms due to their soil-dwelling nature, eventually bringing significant site-specific spatial uncertainties to simulate the soil-crop-pest ecosystem. Most of the approaches have been quantitatively standardised, leaving out qualitative factors prioritizing farmer’s behaviour towards pest control measures. In our study we investigate statistical-empirical and process based models and integrated modelling frameworks for future simulation of soil-crop-pest phenology under the CC impact. We identified gaps in the improvements of pest modelling approaches and site-specific agrometeorological indices for four agro-climatic zones. In conclusion we propose a mechanistic physiological based demographic model, including improved population models under defined landscape extent and agrometeorological drivers. For example, temperature, photoperiod, precipitation changes, and topography are the significant drivers for ECB, whereas in addition, WCR and wireworms include soil type, crop rotation, and farmer’s management options. We derive crop damage factors from the quantitative surveys influencing maize grain yield, root damage and pest control measures, including weather conditions, insecticide distribution, pest pressures, and synchrony between pest and crop growth cycle. Agro-climatic zones of central Europe will experience northward shifts in temperature zones, growing season length, by further accelerating climate change. Hence, it is exposed to more vulnerabilities and demands improved alert services and control measures. The improved pest modelling approach integrated with crop models under CC impact provides a reasonable basis for advanced crop protection strategies with reduced pesticide use in maize production. It is used to formulate a regional risk assessment for specific pests to inform an integrated approach of cultivation, biological and chemical measures.

How to cite: Gajraj, R. and Eitzinger, J.: Review on modelling tools for maize pests risks in agro-climatic zones of central Europe., EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-263, https://doi.org/10.5194/ems2021-263, 2021.

Thibault Moulin and Pierluigi Calanca

European permanent grasslands not only represent a backbone for dairy and meet production, but also are hotspots of biodiversity, providing important ecosystem services to society. Understanding how climate variability and change affect the botanical composition of permanent grasslands is therefore essential for informing adaptation and helping farmers targeting sustainable development goals. It is also a key requirement for gauging climate change effects on forage quality, an aspect often overlooked in impact assessments. In this contribution, we present results of a modelling effort to understand short- and long-term changes in grassland biodiversity in response to climatic variations. We use DynaGraM, a recently developed process-based model for simulating community dynamics in multi-species managed grasslands. Earlier we demonstrated that DynaGraM is capable of representing the composition of permanent grasslands in the French Jura Mountains inferred from floristic relevés. In these earlier investigations, we also showed that the model predicts highest, resp. lowest vegetation diversity for extensive grazing, resp. extensive mowing. We further found that the time scales of responses to external perturbations largely dependent on management, with shorter time scales (of the order of 5 to 10 years) under grazing than under mowing (of the order of 50 years).

Here we apply the model to examine how increasing summer aridity affects the species composition of pastures in the same geographic area. To drive the model, we use a set of climate change scenarios obtained from the CMIP5 repository, which we downscaled with the help of the LARS-WG stochastic weather generator. The results underline that management intensity modulates the impact of summer drought on both yield as well as botanical diversity, with largest changes over time in the latter under extensive grazing. Apart from presenting the results in more detail, we also discuss their practical implications and opportunities to extend in future the scope of this work.

How to cite: Moulin, T. and Calanca, P.: Agricultural ecosystem services under climate change – Modelling grassland biodiversity, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-278, https://doi.org/10.5194/ems2021-278, 2021.

Padraig Flattery, Klara Finkele, Mladen Cucak, Steven Kildea, Rowan Fealy, and Paul Downes


Potato late blight caused by oomycete Phytophthora infestans (Mont) De Bary, is arguably the most important disease of potato in terms of economic losses and environmental and economic impact of the disease management in Ireland and globally. The pathogen spreads rapidly in humid weather in the foliage of potatoes and tomatoes, causing the foliage to decay and the infection of the potato/tomato. Due to the pathogen’s dependence on weather conditions, a number of forecasting methods have been developed to reduce the input of pesticides and fungicides required to control the disease. Blight is of particular significance to Ireland, following the decimation of the potato crop in the 1850s which, among other colonial factors, contributed to widespread famine, emigration and death. Until recently, blight in Ireland has been forecast using the Irish Rules. The model was developed in the 1950s and calibrated based on the sparse meteorological station coverage at the time and lower risk perception by growers.


Since then, the understanding of the pathogen’s development and its relationship to meteorological situation has advanced. This has led to the development of a modified open-source Irish Rules model written in R which facilitates improved forecasting and seasonal re-evaluation of the Irish Rules (Cucak et al., 2019). The new method reduces the threshold for relative humidity from 90% to 88% and the initial (sporulation) period from 12 hours to 10 hours, the analysis also showed thresholds for blight epidemics could be changed from 10°C to 12°C. Though risk estimation has increased compared to the previous rules, estimated chemical usage is still lower compared to standard grower’s practice. The new methodology is now referred to as the New Irish Rules.


This research presents the implementation of this new blight forecasting model in Met Éireann’s operational infrastructure, ensuring methods for forecasting blight are as up-to-date as possible and are using industry-wide best-practice.


Cucak, M., Sparks, A., Moral, R. D. A., Kildea, S., Lambkin, K., & Fealy, R. (2019). Evaluation of the ‘Irish rules’: the potato late blight forecasting model and its operational use in the Republic of Ireland. Agronomy, 9(9), 515.

How to cite: Flattery, P., Finkele, K., Cucak, M., Kildea, S., Fealy, R., and Downes, P.: Implementing the modified potato blight risk forecasting model in the Republic of Ireland – the New Irish Rules, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-283, https://doi.org/10.5194/ems2021-283, 2021.

Updated Operational Implementation of the Canadian Forest Fire Weather Index System in Ireland
Padraig Flattery, Klara Finkele, Paul Downes, Ferdia O'Leary, and Ciaran Nugent
Sabina Thaler, Josef Eitzinger, and Gerhard Kubu

Weather-related risks can affect crop growth and yield potentials directly (e.g. heat, frost, drought) and indirectly (e.g. through biotic factors such as pests). Due to climate change, severe shifts of cropping risks may occur, where farmers need to adapt effectively and in time to increase the resilience of existing cropping systems. For example, since the early 21st century, Europe has experienced a series of exceptionally dry and warmer than usual weather conditions (2003, 2012, 2013, 2015, 2018) which led to severe droughts with devastating impacts in agriculture on crop yields and pasture productivity.

Austria has experienced above-average warming in the period since 1880. While the global average surface temperature has increased by almost 1°C, the warming in Austria during this period was nearly 2°C. Higher temperatures, changing precipitation patterns and more severe and frequent extreme weather events will significantly affect weather-sensitive sectors, especially agriculture. Therefore, the development of sound adaptation and mitigation strategies towards a "climate-intelligent agriculture" is crucial to improve the resilience of agricultural systems to climate change and increased climate variability. Within the project AGROFORECAST a set of weather-related risk indicators and tailored recommendations for optimizing crop management options are developed and tested for various forecast or prediction lead times (short term management: 10 days - 6 months; long term strategic planning: climate scenarios) to better inform farmers of upcoming weather and climate challenges.

Here we present trends of various types of long-term weather-related impacts on Austrian crop production under past (1980-2020) and future periods (2035-2065). For that purpose, agro-climatic risk indicators and crop production indicators are determined in selected case study regions with the help of models. We use for the past period Austrian gridded weather data set (INCA) as well as different regionalized climate scenarios of the Austrian Climate Change Projections ÖKS15. The calculation of the agro-climatic indicators is carried out by the existing AGRICLIM model and the GIS-based ARIS software, which was developed for estimating the impact of adverse weather conditions on crops. The crop growth model AQUACROP is used for analysing soil-crop water balance parameters, crop yields and future crop water demand.

Depending on the climatic region, a more or less clear shift in the various agro-climatic indices can be expected towards 2050, e.g. the number of "heat-stress-days" for winter wheat increases significantly in eastern Austria. Furthermore, a decreasing trend in maize yield is simulated, whereas a mean increase in yield of spring barley and winter wheat can be expected under selected scenarios. Other agro-climatic risk indicators analysed include pest algorithms, risks from frost occurrence, overwintering conditions, climatic crop growing conditions, field workability and others, which can add additional impacts on crop yield variability, not considered by crop models.

How to cite: Thaler, S., Eitzinger, J., and Kubu, G.: Trends of weather-related impacts on Austrian crop production under changing climate, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-287, https://doi.org/10.5194/ems2021-287, 2021.

Peter K. Musyimi, Balázs Székely, and Tamás Weidinger

Reference evapotranspiration (ET0) and real evapotranspiration (ET) are vital components in hydrological processes and climate related studies. Understanding their variability in estimation is equally crucial for agricultural planning processes. The primary goal of this study was to analyze and compare estimates of ET0 and ET from two different climatic regions of Kenya using long term quality controlled synoptic station datasets from 2000-2009 with 3-hour time resolution. Methodology of Linear interpolation was used if the missing measurement periods were equal or smaller than 12 hours. Mean daily course of the meteorological elements combined with the measured variables before and after the data gap was used for longer missing periods.

Three weather stations (Mombasa, Voi and Garissa) were sought from lowlands (between 57 m to 579 m above sea level) characterized by savannah tropical, arid and semi-arid conditions (Aw, Bsh and Bwh) while the other three (Nyeri, Eldoret and Embu) were sought from Kenya highlands (>1350m above sea level) with humid conditions (Cfa, Cwa). Reference Evapotranspiration was calculated based on the FAO56 standard methodology on a daily base. One dimension Palmer-type soil model was used for estimating ET using the wilting point, field capacity and soil saturation point for each station at 1 m deep soil layer. Ratio of real and reference evapotranspiration dependent on the soil moisture stress linearly. Application of the site-specific crop coefficients (Kc) were also used. Calculations of ET were made on daily and monthly basis.

Results indicated that the difference between daily and monthly scale calculation of ET estimates was small. This was because of high temperatures, and high global radiation experienced in the tropics where Kenya lies. For instance, in Voi (03 23S, 38 33E), monthly ET0 ranged from 148±11.6 mm in November to 175. ±10.8 mm in March and ET (without specific crop type, ) was only from 8±4.5 mm in September to 105±50.3 mm in January. This was because in February and from June to August the annual precipitation was below 10 mm while in November and in December it was higher than 100 mm. The choice of Voi, was necessitated by its location in Taita-Taveta County an emerging agriculturally productive region. The study also established that there was zero runoff in the area which was necessitated by low amounts of annual precipitation 574±206 mm thereby influencing ET estimates. The study is suitable because it will enable analyses of 3-hour time resolution data set for longer time period up to 2020 and compare it with ERA5 hourly dataset by investigating the uncertainty of calculations.

How to cite: Musyimi, P. K., Székely, B., and Weidinger, T.: Long term reference and real evapotranspiration modelling using one-dimensional Palmer-type soil model for different climatic regions of Kenya , EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-330, https://doi.org/10.5194/ems2021-330, 2021.

Sinclair Chinyoka and Gert-Jan Steeneveld

This study focuses on the assessment of the impact of downscaling seasonal forecasts from the Climate Forecast System version 2 (CFSv2) using the Weather Research and Forecasting (WRF) mesoscale model over Zimbabwe on a spatial resolution of 21km and 7km for Southern Africa and Zimbabwe respectively. We used a 7-day re-initialization simulation strategy for 212 days per season and was repeated for eights seasons between 2010 and 2018. The impact of downscaling global seasonal forecasts was further evaluated in crop forecasting using the WOrld FOod STudies (WOFOST) model. Statistical analysis of the forecasted seasonal rainfall revealed a reduction of the bias from about -2 mm/day from CFSv2 forecasts to about 0.5mm/day from WRF forecasts in most parts of the country. We also found that an improvement in seasonal tercile rainfall prediction from 25%, 50%, and 75% by CFSv2 in three different regions to about 62.5% by WRF in all regions. Substantial improvement was achieved in Standard Precipitation Index-driven seasonal forecasts with two regions with a percent correct of 75% and region 2 with 100% by WRF compared to 62.5% by CFSv2 in all regions. Hence, the characterization of seasonal rainfall in terms of drought forecasts is better than the tercile rainfall prediction system and will be more beneficial to farmers in Zimbabwe. WRF seasonal rain forecasts improved both in magnitude and in forecasting the onset of the growing season. This was indicated by the accumulated absolute maize yield error which factored in a miss of onset of the growing season by each model. WRF outperformed CFSv2 for maize and sorghum yield forecasts in 6, 6, and 8 (out of 8) seasons in Karoi, Masvingo, and Gweru sites respectively. WRF forced crop simulations reduced mean absolute percent error of maize yield by 12.2% and sorghum yield by 9.3 % from CFSv2 forced simulations. Our results also show that maize will be more productive and less risky at Karoi and Masvingo and sorghum at the Gweru site. In our view, there should be no farming of both maize and sorghum at Beitbridge due to the high risk of crop failure unless a proper irrigation system is in place.

How to cite: Chinyoka, S. and Steeneveld, G.-J.: Downscaling global seasonal weather forecasts for crop yield forecasting over Zimbabwe, EMS Annual Meeting 2021, online, 6–10 Sep 2021, EMS2021-402, https://doi.org/10.5194/ems2021-402, 2021.


Supporters & sponsors