Applying Astronomy Capability to Map an Invasive Weed: Leveraging Satellite Surveys to Inform "Famine Weed" Policy in Pakistan

It is projected that food demand will more than double by 2050 due to climate changes. The new generation of Earth observation data, including the Sentinel missions, offer the potential to identify and counteract crop yield declines. Astronomers are experts in analysing large multiband imaging surveys by making sense of the plethora of information using smaller targeted observations at high resolution and/or with spectroscopic information. This is exactly the expertise needed to utilise these new Earth observations to address global food security challenges. In this proposal, for the first time, we build crossover capabilities by focussing on a critical threat to two of the four most important crops in Pakistan (>75% of total output). Beyond the specific outputs of this project, this multi-disciplinary collaboration between astronomy, Earth observations and food security will also deliver transformative open-source tools. They will have secondary benefits to the UK in addressing global food security, and could be applied to challenges such as blackgrass in wheat. Furthermore they showcase the huge potential of STFC capabilities to contribute to tackling global challenges.

Food security in Pakistan is reliant on its ability to produce wheat and rice. A new invasive species of weed, the "Famine Weed" (Parthenium hysterophorus), has been identified as a critical threat to agriculture and human prosperity in the region. Parthenium affects crop yields by as much as 90%, and has serious impacts on human and livestock health, and is therefore this project is of great importance to Pakistan. The University of Manchester and the Centre for Agriculture and Bioscience International (CABI) have joined forces to define a key challenge area for STFC research: the use of astronomy image analysis for mapping Parthenium across Pakistan. With their Islamabad office, CABI have over sixty years of on-the-ground experience in Pakistan and a proven track record of undertaking agricultural research projects in this region. Combined with the world-leading expertise in astronomy image analysis and optical instrumentation engineering at Manchester, we propose a four-stage project to provide detailed mapping of Parthenium in wheat and rice agriculture for the major arable regions of Pakistan.

Initially, we will aggregate and preprocess multispectral satellite imagery (WP1) and implement a campaign to collect high quality ground reference data (WP2) which is currently lacking. These will be combined to produce a computing framework for the detection of Parthenium in crops across the country. This will deliver static maps for the current extent of the weed (WP3) as well as a mechanism for future monitoring to be adopted by policy makers (WP4). We also place a strong emphasis on local academic capacity building through employment of local researchers, delivery of technical workshops, data hack challenge events and a researcher exchange program between Manchester and universities in Punjab (WP4).

Impact is achieved via three interlinked outcomes:
1) Policy makers become aware of and able to use satellite data for decision-making in agriculture, livestock, and human health;
2) Improved methods allow for automated mapping and monitoring of the distribution of Parthenium in rice and wheat fields; and
3) Pakistani researchers develop strengthened capacities in their fields that contribute to their ongoing work.

This project seeks to improve food security, livelihoods, and human and environmental health in Pakistan by applying UK astronomy expertise to the monitoring and mapping of Parthenium so that it can be more effectively managed by national stakeholders. We will generate impact by focusing our activities around three outcomes:

Outcome 1: Policy makers become aware of and able to use satellite data for decision-making in agriculture, livestock, and human health
We will engage with key decision-makers to raise awareness educate them about new data-driven approaches for Parthenium weed monitoring. Two high-level policy workshops will allow us to engage with a cross-sectoral group of policy makers. Authorities can then use the maps generated for targeted weed control and the measurement of their effectiveness. We will work in collaboration with the Pakistan Agricultural Research Council to produce policy briefs.

Key measures: Attendance at workshops, workshop evaluation forms, joint policy briefs published, usage analytics /download stats, system user survey

Outcome 2: Improved methods allow for automated mapping and monitoring the distribution of Parthenium in rice and wheat fields
Quantifying the large-scale distribution of invasive weeds is paramount for the sort of ecological modelling that will allow authorities to target their efforts. This requires one to obtain systematic monitoring in remote areas. We will develop a new framework to distill low-level satellite imaging into more user-friendly data products. Their reliability will undergo extensive testing and calibration by verifying their accuracy against a comprehensive set of field observations, which will pave the way for their adoption. The methods that will be developed in this project have a much broader scope of application than Parthenium, with direct use for connected agricultural problems faced by the UK and other countries as well as possible translations into novel use of remote sensing for society.

Key measures: Publications in refereed journals, self-evaluation of map reliability through 'double-blind' data challenge, new detections of Parthenium confirmed by ground measurements

Outcome 3: Pakistani researchers and technicians develop strengthened capacities in their fields that contribute to their ongoing work
Academic and non-academic researchers in Pakistan will play a pivotal role in the data collection and analysis. The involvement of a postdoctoral research assistant will allow us both to build capacity for the remote sensing and to build research partnerships with Pakistani universities. Exchange visits between UK and Pakistani researchers will strengthen the ability of everyone involved to apply astronomy expertise towards addressing the development challenges faced by Pakistan.

Activities such as 'hack-a-thons' will cross-connect new science communities. These events offer opportunities to foster a STEM culture, but also enable scientists from 'traditional' STFC backgrounds to discover new career paths such as data science in humanities.

We will train 60 individuals to perform high-quality field work. Most will be undergraduate students recruited for the task, who will gain useful survey experience to complement their studies.

Key measures: Outputs (publications, presentations) produced by PDRAs, feedback evaluations from exchange visits, numbers of attendees at hackathon, hackathon evaluation forms, numbers of field assistants trained

Impact beyond this project
Parthenium is a global problem, however, and the potential impact of this work goes well beyond Pakistan. We will explore possibilities to extend this work in other parts of the world. We may also identify opportunities for applying similar methods to other invasive weeds with major food security impacts.

Tools and data products generated will also be made available on an open source platform, for use beyond the end of the project and by other stakeholders.