Project highlights

  • Weather patterns
  • Crop yield variability prediction
  • Machine learning


Projected increases in the frequency, intensity and duration of extremes such as heat waves and extreme precipitation can have devastating impacts on crop production. Predicting how these events are likely to affect future crop yields is of major importance and can help governments and businesses to better respond to food production shocks and food price spikes.

Anomalous weather conditions cause crop yield variations; previous research has highlighted the importance of growing season temperature and precipitation in explaining crop yield variability. For example, significant crop production losses across the EU are mainly due to drought and/or heavy precipitation caused by large-scale weather systems. Large-scale weather systems can affect crop yield in many countries across the world simultaneously. For example, in 2007 and 2012, eastern European countries were affected by severe drought conditions, while western countries experienced problems generated by heavy rainfall, which both influenced crop yields.

Weather patterns (Figure 1), sometimes referred to as weather regimes or types, are large-scale atmospheric circulation patterns that drive specific temperature and precipitation patterns. Weather patterns have recently been used to model hydrological and meteorological extremes such as floods and heat waves. Weather patterns have also higher predictability compared to small scale events, such as precipitation and or temperature extremes, making them particularly useful.

Though previous research i) highlighted the role of climate and weather deriving crop yield variability and ii) applied the weather types approach to enhance the predictability of extremes, no study has yet investigated the association between weather types, their frequency and persistence and crop yields or assessed the extent to which weather types can be used to enhance year-to-year crop yield prediction.

The main hypothesis of this PhD project is that weather types can be used to explain spatial and temporal climate-related crop variability and can provide longer crop yield prediction lead times than for example using specific indicators such as precipitation or temperature.

The study will be applied to Europe and North America given the availability of the crop yield data, with a focus on the UK.


Cranfield University


  • Climate and Environmental Sustainability


Project investigator

  • Dr Abdou Khouakhi, Cranfield University


  • Dr Toby Waine, Cranfield University

How to apply


The aim of this PhD project is to develop an understanding of the spatial and temporal variations in crop yield and weather patterns across different parts of Europe and North America; examine their changes in the past and future; and also study the value addition of using weather patterns for seasonal crop yield prediction. The student will: 1) Use gridded reanalysis data such as geopotential height, sea level pressure, precipitation and temperature, to  develop and train machine learning algorithms to define relevant weather patterns across Europe and north America with a focus on the UK. Analyses can be performed for different seasons and using various reanalysis products. 2) Using weather patterns defined for each region, the student will then observe changes in the frequency and persistence of WPs with crop yield variations. Other crop production indices based on remote sensing data may also be considered.The analysis will be conducted at different spatial (regional) and temporal (seasonal and annual) scales, and for different crop yields. 3) Finally, using key global subseasonal to seasonal climate models, the candidate may evaluate the skill of crop prediction associated with WPs.

Training and skills

The student will have access to relevant MSc modules taught at Cranfield (e.g. Applied Remote Sensing, Geographical Information Systems, Big data analytics, Environmental Resource Survey, Modelling Environmental Processes). Training will be provided in Python and R for data science, statistical analysis and modelling by the supervisory team. There will be opportunities to attend various other trainings, short courses and seminars through Cranfield’s Doctoral Training Centre, encouraging effective and vibrant research.

Possible timeline

Year 1

Literature review, data gathering and development algorithms to define weather patterns over the UK, Europe, and North America at spatial and temporal scales relevant to crop yield variability predictions. Training to further enhance data science programming and machine learning skills in Python, R and Google earth engine.

Year 2

Research the association between weather types and crop yield at different spatial and temporal scales. Examine the year to year variability of crop yield with weather patterns defined in 1. Use remote sensing to construct indices relevant to crop yields. Prepare first journal article, and attend relevant conferences such as the EGU.

Year 3

Use global climate models to forecast weather patterns and crop yield at the seasonal scale and beyond, and across regions within the study area. Discuss and exchange with end-users and practitioners, final analyses, publish in peer-reviewed journals, present research to national and international crop forecasting communities. Write up the thesis.

Further reading

The following papers are provided as a starting point only, and applicants are encouraged to seek broader literature on the topic.

Journal articles:

Ray, D. K., Gerber, J. S., Macdonald, G. K., & West, P. C. (2015). Climate variation explains a third of global crop yield variability. Nature Communications, 6, 1–9.

Basso, B., Cammarano, D., & Carfagna, E. (2013). Review of Crop Yield Forecasting Methods and Early Warning Systems. The First Meeting of the Scientific Advisory Committee of the Global Strategy to Improve Agricultural and Rural Statistics, 1–56.

Richardson, D., Fowler, H. J., Kilsby, C. G., & Neal, R. (2018). A new precipitation and drought climatology based on weather patterns. International Journal of Climatology, 38(2), 630–648.

Chattopadhyay, A., Hassanzadeh, P., & Pasha, S. (2020). Predicting clustered weather patterns: A test case for applications of convolutional neural networks to spatio-temporal climate data. Scientific Reports, 10(1), 1–13.

Neal, R., Fereday, D., Crocker, R., & Comer, R. E. (2016). A flexible approach to defining weather patterns and their application in weather forecasting over Europe. Meteorological Applications, 23(3), 389–400.

Lamb HH. 1972. British Isles weather types and a register of the daily sequence of circulation patterns 1861–1971. Geophys. Mem. 116: 85.

Web page with an author:

D Ray, D. (2019) Climate change is affecting crop yields and reducing global food supplies Available at:

Web page—author as an organisation:

UK Met Office ‘UK dry periods and weather patterns’. Available at:


The project is largely a desk-based study. Low risk of COVID-19 is anticipated on the delivery of the project. However, COVID-19 might limit the physical interaction of the student with other student and researchers. COVID might also affect physical attendance to conferences and meetings.