A novel way of identifying agricultural water drainage systems and their impact on catchment hydrology

Water drained from agricultural lands is getting more attention as its valuable water is lost from groundwater storage. The historical location of buried agricultural water drainage systems is not known very well. Hence, first, finding the location of those infrastructures is critical. Several methods have been applied in the past, including decision tree classification (DTC), remote sensing based, using radar, etc. However, all the methods neglect the primary cause of the drain application, which is groundwater. Hence, a novel approach is introduced that considers groundwater in the identification procedure. We used two case studies for drain identification, one from Ontario, Canada, and another from Belgium. Furthermore, a physically based and fully distributed modeling approach (SWAT+gwflow) is conducted to investigate the impact of these drainage systems in the catchment hydrology of the Kleine Nete watershed, Belgium.

The result of the drainage system identification has indicated the pitfalls of the already existing methods where accuracy as low as 17% was recorded. On the other hand, the additional filtering based on groundwater head enables us to find an additional 19.4 km² area. Therefore, the use of groundwater level as an additional filtering technique is vital for increasing the accuracy of tile drain/ditch network identification. Next, drains have been shown to affect hydrology, where a 15% decrease in groundwater evapotranspiration, a 50% reduction in groundwater saturation excess flow, and a 39% decline in groundwater discharge to streams are observed.