The annual IGARSS event took place recently in Brussels, though not really in Brussels of course because of COVID-related restrictions. Instead, it was virtual, as you might expect, but as a virtual offering, it was magnificent. An excellent, interactive web interface enabled easy navigation around a technical programme comprising 1500 oral presentations and 1000 multimedia presentations, plus other useful features such as an exhibition hall and a job fair.
As a big conference, with thousands of participants, coverage of the remote sensing discipline was broad overall, including developments in remote sensing instruments, image processing analysis and application to geospatial problems. And, as you would expect, work was presented from all over the world. However, there was a noticeable deficit of African research, despite the urgent need to exploit geospatial tools for tackling environmental and social problems throughout Africa.
Pleasingly, therefore, four Edge Hill researchers – Professor Paul Aplin and PhD students Alex Amoakoh, Kwame Awuah and Anthony Cizek – went some way to redress this imbalance, each presenting an oral presentation on some aspect of vegetation monitoring in sub-Saharan Africa.
Alex presented his work on mapping tropical peatland in Ghana, published recently in Sensors, one of the top remote sensing journals. In particular, Alex created the first comprehensive land cover classification of Greater Amanzule peatland using contemporary remote sensing data and approaches. This is a vital form of spatial inventory – Africa’s peatlands are fragile and valuable ecosystems, and are wholly understudied and often completely unmapped.
Anthony, Kwame and Paul all presented in a series of invited sessions on savannah monitoring. Anthony delivered his research towards Miombo woodland mapping in Zimbabwe, showcasing his technically sophisticated methodology for identifying the time of green-up in woody environments. Green-up is a key phenological characteristic for Miombo woodland, potentially offering a means to distinguish this important and threatened ecosystem from other vegetation cover.
Kwame delivered the latest part of his doctoral research on mapping grazing lawns in southern African savannahs. Specifically, he exploited new Planet data – very high resolution satellite imagery of the tropics provided free of charge to tackle the global deforestation crisis – and fused this with medium resolution Sentinel imagery to distinguish short grass patches from other vegetation classes. These grazing lawns are a very significant source of forage for megaherbivores such as rhinoceros and also form protection against wildfire since there is little fuel to burn.
Paul presented work published recently in Remote Sensing – the biggest remote sensing journal, with an impact factor of 4.8 – on mapping plant functional types using state-of-the-art machine learning classification approaches and spatially and spectrally detailed WorldView-2 satellite imagery. Significantly, this new work sets a high bar for separating spectrally similar vegetation classes in what are very heterogeneous – and difficult to map – savannah environments.
One slightly disappointing aspect of the conference – in addition to the paucity of African work – was the high number of no-shows in some sessions, perhaps partly a result of the online format. Every problem creates an opportunity though, so in the case of the final savannah session, where Paul was giving the closing paper, instead of an awkward silence, Paul and Kwame volunteered to host their Great African Dung Quiz. As a result of their grazing lawn fieldwork, the research team is now proficient in finding traces of animal presence, including, of course, their dung! The quiz – delivered in a recent Edge Hill Festival of Ideas session – proved to be the highlight of the session, creating more animation in the audience than any of the technical papers. A great end to an excellent conference.
If you would like to be part of a Geography and Geology department that has research led teaching at its core, you can view more about our courses here: