Evaluating the quality of remote sensing-based agricultural water productivity data
Due to the COVID-19 crisis the PhD defence of Megan Blatchford will take place (partly) online.
The PhD defence can be followed by a live stream.
Megan Blatchford is a PhD student in the research group Water Resources (WRS). Her supervisor is dr.ir. C.M.M. Mannaerts from the Faculty of Geo-Information Science and Earth Observation (ITC).
The role of water in agriculture is essential and plays an important role in food security. We need to better understand how to optimise water use in agriculture to meet both global food security and water management efficiency goals. Earth observation with satellite offers the opportunity to map, monitor and better understand the dynamics of agricultural water. However, the applicability and accuracy of these observations need to be better understood. This dissertation aims to better understand the suitability of large remote sensing-based datasets for the monitoring of crop water productivity (CWP). CWP being an indicator of agricultural water efficiency.
The dissertation is composed of seven chapters.
Chapter 1 is introductory and describes; the importance of CWP in agriculture, how CWP can be monitored and the importance of understanding the accuracy of remote sensing based CWP estimates.
Chapter 2 provides and introduction to the The FAO portal to monitor WAter Productivity through Open access of Remotely sensed derived data dataset.
Chapter 3 assesses the accuracy of remotely sensed CWP against the accuracy of estimated in-situ CWP. The accuracy of CWP based on in-situ methods, which are assumed to be the user's benchmark for CWP accuracy, and the CWP derived from remote sensing methods are reviewed. CWP from non-locally parametrised studies can achieve accuracy within the error bounds of in-situ methods. However, it was shown that Uncertainty varies significantly dependent on method and application.
Chapter 4 evaluates the accuracy of a large remote sensing-based evapotranspiration dataset, the denominator in the CWP indicator, over Africa. The evaluation uses multiple methods to compensate for sparse ground-truth data. The dataset showed mixed results at point, daily scale to annual, basin scale, which were similar to those reported by other literature in Chapter 2. The research highlighted the risks in using these large datasets in CWP monitoring, due to the high variation in accuracy of a large dataset.
Chapter 5 quantifies the suitability of varying remote sensing-based resolutions for application in agricultural productivity. Three performance indicators, adequacy, equity and productivity are tested in five irrigation schemes for three spatial resolutions, 250m, 100m and 30m. This is frequently resulting in large differences in the irrigation performance assessment criteria for inter-plot comparisons. This is particularly noticed in irrigation schemes with the smallest field sizes. This highlights the importance of selecting the spatial resolution appropriate for to scheme characteristics when undertaking irrigation performance assessment using remote sensing input.
Chapter 6 quantifies a sample size to improve processing time of validation activities of large continental CWP datasets. A progressive sampling approach, as typically applied in machine learning to train algorithms, combined with two performance measures, was applied to estimate the required sample size. The proposed approach can significantly reduce the processing time while still providing a statistically valid representation of a large remote sensing dataset. This can be useful as more high-resolution remote sensing data becomes available.
Chapter 7 reflects on the implications of the CWP concept and how the concept itself may at times contradict the global food security and water management efficiency goals. The CWP concept can be a useful one, however, it can also be misleading. On both a local and global scale, users must reflect on the goals and select or prioritise indicators, improving CWP may not always be the most useful indicator. We need to focus on the sustainable improvement in CWP rather than the increase in CWP for sustainable agriculture.
Future studies should show how the interactions between carbon and transpiration can better be utilised as a tool in understanding the quality of agricultural water monitoring datasets.