Periodic Reporting for period 1 - LucFRes (Land-use change and the resilience of food production systems (LucFRes))
Período documentado: 2022-02-01 hasta 2024-01-31
LucFRes studied how land-use changes reshape the resilience of smallholder farming systems. As land use change constrains the future of food production in the face of climate change and human pressure on natural resources, how agricultural land use change builds or undermines farming systems and their resilience is still poorly understood. The project integrates the analysis of agricultural land-use changes from satellite remote sensing with stakeholders' perceptions of land-use futures in Southwest Nigeria, which it uses as a test case. Current Remote Sensing methods for identifying and predicting crops are challenged in intercropped farming systems because they were developed for large-scale crop monocultures. Hence, these have limited applicability in heterogeneous agricultural systems such as those dominating smallholder agriculture worldwide, especially in developing countries. In LucFRes, a Sentinel-1 and Sentinel-2 image-based framework was developed for mapping multiple crops and intercropping considering two growing cycles. All objectives and planned activities in LucFRes were achieved except the final stakeholder workshop as the project was terminated earlier than planned to allow the Experienced Researcher (ER) to take up an Associate Professorship. The final stakeholder workshop is planned for July 2024. The MSCA research supported the expansion of the ER’s professional network and contributed to enabling the ER to gain a position as an Associate Professor in Geomatics at Karlstad University, Sweden since November 2023.
All deliverables and milestones were achieved in LucFRes except the land use scenario modelling and final stakeholder workshop. The latter is because the project was terminated earlier than planned due to the ER taking up an Associate Professorship. The support received from the Marie Skłodowska-Curie grant was acknowledged in presentations and publications.
1) The first paper under review mapped land use over 23 years and quantified the intensity of LULC change (i.e. annual rates) during two intervals (i.e. 2000 – 2013 and 2013 – 2022) across seven agroecological zones (AEZs). The intensity of LULC change accelerated during the study period in all AEZs (e.g. rainforest, mangrove), except in the semi-arid Sudan and Sahel savannah where speed was higher in 2000 – 2013 than in 2013 – 2022 due to grassland cultivation made possible by large irrigation schemes. The extent of human appropriated natural land cover in Nigeria in the last 23 years was estimated. The main change processes in Nigeria (2000 – 2022) are primarily related to the dominance of human activities as more natural cover was lost than was gained for nature during all time intervals due to cropland expansion and artificialisation, e.g. settlement development .
2) The second paper, currently under review, identifies important farming system classes for producing food in southwest Nigeria (SWN). Combining Sentinel-1 radar and optical Sentinel-2 time-series, a Remote Sensing and Machine Learning framework was developed to map multiple crops and intercropping during two growing cycles in smallholder mixed farming systems. Using deep transfer learning, we found monocropping positively related to field size in the Nigerian lower Guinea Savannah of SWN.
3) I developed content and co-taught a course in the MSc Geography programme at UBERN on Land Systems and Sustainable Land Management I am currently supervising two PhD and one MSc theses.
4) I participated in programs providing personalised coaching for the academic appointment process, which proved invaluable to my securing the Associate Professorship. These are 1) Coaching, Mentoring and Training Programme (COMET), University of Bern, 2) Women in Natural Sciences (WINS) programme at Humboldt Universität zu Berlin, Germany.
5) Information about LucFRes and the results were well disseminated in both scientific forums and to the public. Planned activities were embarked upon early in the project. The project flier was prepared and disseminated (online and printed), and two websites were published. Three seminars were organised at the Land System and Sustainable Land Management unit at UBERN. Five other seminars were held at universities in Germany, Sweden and Nigeria. I also presented results of the Remote Sensing-based land use change intensities and the extent of human appropriated natural land cover in Nigeria since the last 23 years at the International Association for Landscape Ecology conference in Nairobi (July 2023). Aspects relating to identifying and predicting multiple crop types in smallholder intercropping systems during multiple growing cycles were presented in September 2023 at the Tropentag in Berlin, Germany.
6) LucFRes linked up with EU-funded SUSTAINFORESTS (ERC grant) and AQUATIC (MSCA-EF) to organise exhibitions and presentations.
1) The first paper under review mapped land use over 23 years and quantified the intensity of LULC change (i.e. annual rates) during two intervals (i.e. 2000 – 2013 and 2013 – 2022) across seven agroecological zones (AEZs). The intensity of LULC change accelerated during the study period in all AEZs (e.g. rainforest, mangrove), except in the semi-arid Sudan and Sahel savannah where speed was higher in 2000 – 2013 than in 2013 – 2022 due to grassland cultivation made possible by large irrigation schemes. The extent of human appropriated natural land cover in Nigeria in the last 23 years was estimated. The main change processes in Nigeria (2000 – 2022) are primarily related to the dominance of human activities as more natural cover was lost than was gained for nature during all time intervals due to cropland expansion and artificialisation, e.g. settlement development .
2) The second paper, currently under review, identifies important farming system classes for producing food in southwest Nigeria (SWN). Combining Sentinel-1 radar and optical Sentinel-2 time-series, a Remote Sensing and Machine Learning framework was developed to map multiple crops and intercropping during two growing cycles in smallholder mixed farming systems. Using deep transfer learning, we found monocropping positively related to field size in the Nigerian lower Guinea Savannah of SWN.
3) I developed content and co-taught a course in the MSc Geography programme at UBERN on Land Systems and Sustainable Land Management I am currently supervising two PhD and one MSc theses.
4) I participated in programs providing personalised coaching for the academic appointment process, which proved invaluable to my securing the Associate Professorship. These are 1) Coaching, Mentoring and Training Programme (COMET), University of Bern, 2) Women in Natural Sciences (WINS) programme at Humboldt Universität zu Berlin, Germany.
5) Information about LucFRes and the results were well disseminated in both scientific forums and to the public. Planned activities were embarked upon early in the project. The project flier was prepared and disseminated (online and printed), and two websites were published. Three seminars were organised at the Land System and Sustainable Land Management unit at UBERN. Five other seminars were held at universities in Germany, Sweden and Nigeria. I also presented results of the Remote Sensing-based land use change intensities and the extent of human appropriated natural land cover in Nigeria since the last 23 years at the International Association for Landscape Ecology conference in Nairobi (July 2023). Aspects relating to identifying and predicting multiple crop types in smallholder intercropping systems during multiple growing cycles were presented in September 2023 at the Tropentag in Berlin, Germany.
6) LucFRes linked up with EU-funded SUSTAINFORESTS (ERC grant) and AQUATIC (MSCA-EF) to organise exhibitions and presentations.
The national-level assessment of land cover change and change intensities in LucFRes is the most current for Nigeria. The first land-based (FPS) resilience indicators for SWN was co-produced with stakeholders’ input In LucFRes, an S1 and S2-based framework was developed for mapping multiple crops and intercropping considering two growing cycles in smallholder mixed farming systems. This framework can be used in similar smallholder contexts or adapted for mapping mixed farming in other parts of the world, including newly emerging intercropping systems in Europe.
A novelty is harnessing the monthly NDVI time-series to visually identify spatio-temporal phenology stages to actively label different crop types. For example, temporal characteristics of farming activities from farmer interviews and farm visits – field clearing, crop emergence, crop peaking and senescence stages – helped identify and label different crop types, especially for the early and late planted maize classes, for which their growth stages in the early or late part of the growing season was most critical for identification. Mapped intercropping (i.e. mixed crop farming) and double growing cycles in smallholder farming systems. The best model for predicting crop types combined Sentinel-1 monthly and Sentinel-2 at bimonthly intervals. Detailed field size delineation was automated to assess how field size relates to cropping patterns in smallholder farming systems, and estimates were efficiently obtained with deep transfer learning. We found that monocropping was positively associated with field size in the Nigerian lower Guinea Savannah.
A novelty is harnessing the monthly NDVI time-series to visually identify spatio-temporal phenology stages to actively label different crop types. For example, temporal characteristics of farming activities from farmer interviews and farm visits – field clearing, crop emergence, crop peaking and senescence stages – helped identify and label different crop types, especially for the early and late planted maize classes, for which their growth stages in the early or late part of the growing season was most critical for identification. Mapped intercropping (i.e. mixed crop farming) and double growing cycles in smallholder farming systems. The best model for predicting crop types combined Sentinel-1 monthly and Sentinel-2 at bimonthly intervals. Detailed field size delineation was automated to assess how field size relates to cropping patterns in smallholder farming systems, and estimates were efficiently obtained with deep transfer learning. We found that monocropping was positively associated with field size in the Nigerian lower Guinea Savannah.