{"id":3077,"date":"2026-10-26T12:57:36","date_gmt":"2026-10-26T12:57:36","guid":{"rendered":"https:\/\/sites.exeter.ac.uk\/ted-feldpausch\/?p=3077"},"modified":"2026-10-26T13:03:56","modified_gmt":"2026-10-26T13:03:56","slug":"phd-studentship-can-low-cost-sensors-and-models-show-which-amazon-forests-are-most-at-risk-of-burning","status":"publish","type":"post","link":"https:\/\/sites.exeter.ac.uk\/ted-feldpausch\/2026\/10\/26\/phd-studentship-can-low-cost-sensors-and-models-show-which-amazon-forests-are-most-at-risk-of-burning\/","title":{"rendered":"PhD studentship: can low-cost sensors and models show which Amazon forests are most at risk of burning?"},"content":{"rendered":"\n

Fire is changing the microclimate beneath Amazon forest canopies, and we cannot yet predict where or how. This fully supervised PhD will build and test the sensors, models and machine learning needed to forecast which forests are warming, drying and becoming flammable.<\/strong><\/p>\n\n\n\n

Wildfire, degradation, land-use change and climate extremes open the forest canopy and alter the microclimate beneath it. Forests get warmer and drier, more trees die, and the forest becomes easier to burn. Predicting these changes needs fine-scale measurements and models, yet both remain poorly tested in the tropics. Low-cost sensors drift, fail and vary with design and placement. Physics-based microclimate models have rarely been checked against dense vertical profiles across burned and degraded forest. And scaling point measurements up to landscapes adds uncertainty that is seldom quantified.<\/p>\n\n\n\n

The project<\/h2>\n\n\n\n

The doctoral researcher will develop and test an integrated sensing and modelling framework for predicting fine-scale climate in complex, data-sparse environments, using Amazon forests as a demanding test case. It has four objectives:<\/p>\n\n\n\n

    \n
  1. Sensors:<\/strong> quantify how sensor design, placement, drift and data gaps affect microclimate measurements, by testing established dataloggers against custom-built and industry-partner prototype devices.<\/li>\n
  2. Models:<\/strong> test physics-based models of energy, heat and water exchange against measured vertical profiles, from soil to canopy, across intact, burned and degraded forest.<\/li>\n
  3. Machine learning:<\/strong> combine in-situ sensor data with airborne and spaceborne Lidar (e.g.\u00a0GEDI<\/a>) and optical satellite data, using machine learning to calibrate low-cost sensors and correct systematic model errors.<\/li>\n
  4. Upscaling:<\/strong> propagate uncertainty from the sensor to landscape-scale predictions, and forecast where warmer, drier conditions put trees under heat and drought stress and raise fire risk.<\/li>\n<\/ol>\n\n\n\n

    What you will build on<\/h2>\n\n\n\n

    The project draws on the large environmental sensor network and datasets of the Amazon PyroCarbon Project<\/a>, which records soil temperature, soil moisture and near-surface air temperature across forests with contrasting fire and land-use histories. These are complemented by open remote sensing and land-use data, and by the open-source microclimate models developed by Prof Ilya Maclean\u2019s group at the University of Exeter<\/a>, already validated across tropical forest sites worldwide. An industry sensor partner will support testing of prototype devices, offering a possible route to a placement.<\/p>\n\n\n

    \n
    \"Two
    The PVC radiation shield used across the PyroCarbon sensor network. Photo: \u00a9 Ted Feldpausch.<\/figcaption><\/figure>\n<\/div>\n\n\n

    The supervisory team is Prof Ted Feldpausch (lead), Prof Ilya Maclean and Dr Oscar Kennedy-Blundell.<\/p>\n\n\n\n

    Who we are looking for<\/h2>\n\n\n\n

    You do not need to be a forest ecologist. We welcome applicants with a background in physics, engineering, computer science, mathematics, environmental science, geography, ecology or a related discipline. Strong quantitative skills and programming experience (e.g.\u00a0R, Python or MATLAB) are essential. Experience with environmental sensors or electronics, numerical or statistical modelling, machine learning, remote sensing or GIS would be an advantage, but training will be provided.<\/p>\n\n\n\n

    The doctoral researcher will take a leading role in shaping the project and can emphasise sensor development, modelling or data science. Training covers sensor design and calibration, physics-based and statistical modelling, machine learning, geospatial analysis, R and Python including high-performance computing, and reproducible research practice. Fieldwork in Brazil is possible, subject to research permits. Alternatively, the project can be completed as a desk-based study using existing data and ongoing data collection by partners across Amazonia. The methods developed will transfer well beyond the tropics, to environmental monitoring in agriculture, cities, conservation and wildfire management.<\/p>\n\n\n\n

    How to apply<\/h2>\n\n\n\n

    Applications close at midday on Tuesday 8 December 2026<\/strong>, for entry in September 2027.<\/p>\n\n\n\n

    This is a fully funded studentship, funded by the Engineering and Physical Sciences Research Council (EPSRC) Doctoral Landscape Award at the University of Exeter. For eligible applicants it covers UK tuition fees, a tax-free stipend of \u00a321,805 per year for up to 3.5 years, and a \u00a35,000 research training support grant. Please check the residence eligibility criteria<\/a> before applying. General interviews are likely between 15 and 24 February 2027, and references are due by 8 January 2027.<\/p>\n\n\n\n