Abstract of PhD Thesis

Remote Sensing as a Monitoring Tool for Slurry Spreading on Grasslands for Compliance with European Legislation

Ross Donnelly, University College Dublin, 2015

Under the Nitrates Directive (91/676/EEC) there are strict conditions on the spreading of slurry during the winter. Monitoring slurry spreading on farmlands can be difficult and not always feasible due to the limited number of inspectors and the large amount of farms Compliance with this legislation must be monitored and remote sensing may offer a means of doing this. Slurry that is spread on short (under 5cm) grasslands will have a high percentage of exposed soil. A field experiment was set up to analysis the spectral response with a proximal remote sensor. This experiment was set up first on the 7th of July 2012 and again on the 10th of August 2012. A randomised block design was used with plots that received treatments of slurry (5000ml or 2500ml per m2) and dirty water (5000ml per m2) which were monitored over 26 days. The data were tested with numerous vegetation indices (Normalized Difference Vegetation Index, Enhanced Vegetation Index, and Soil Adjusted Vegetation Index) and the Modified Chlorophyll Absorption Ratio Index was found to be able to significantly detect the different between slurry (5000ml per m2) and the other treatments (potable water/control/dirty water). Using data from the first experiment a Slurry Index (SI) was developed to detect the presence of slurry on short grass and tested on the 2nd experiment. The SI was successful in detection of slurry on Day 1 and 2 of the second experiment. Remote sensors will have difficulty detecting slurry that is spread on grasslands with large amounts of exposed soil compared to slurry that is spread on grasslands with growth over 20 cm. In order to check every farm during the closed spreading season would require a lot of farm visits. To test if remote sensing can be used to aid in monitoring, imagery was obtained from SPOT sensors. The imagery were used to locate and detect slurry spreading events that were cross-referenced with data obtained from farms about the exact location and date of these events. The NDVI was calculated for pixels from fields with slurry and fields with no slurry. The difference between the fields was shown to be statistically significant (P<0.001). Testing the model for prediction indicated it was about 70% accurate. The accuracy of the model could be improved by greater access to slurry spreading records kept by farms. In Ireland using remote sensing during this period is greatly affected by cloud cover. Average cloud cover measured in oktas over the 5 year winter periods at the main airports was analyzed at solar noon. There is a high probability of cloud cover of 6-8 oktas however some months in the winter have a probability of 40% of cloud cover in the 0-5 oktas range. Remote sensing will not replace traditional methods of monitoring but when conditions are suitable this method could be an additional tool to utilize.