Agricultural Tile Drainage: Mapping a Key Nutrient Transport Pathway

Abstract

Agricultural tile drainage has been used to remove excess water and facilitate agricultural crop production. However, nutrient delivery with drainage water from agricultural fields contributes to downstream algal blooms and hypoxic zones. Here, we simulate the nutrient loading, sources, and pathways across the US Great Lakes basin using the Spatially Explicit Nutrient Source Estimate and Flux (SENSEflux) model. An estimated tile drainage layer with GIS-based mapping has been used as one of the SENSEflux inputs. The results from SENSEflux have shown that tile fields are a major pathway of total nitrogen (TN) and total phosphorus (TP) transport, transporting 39% of TN and 29% of TP to the Great Lakes. The contribution is even higher during snowmelt season, and in some regions with a high density of tile drains, such as the Lake Erie basin. We also concluded that the lack of fine-resolution, spatially-explicit tile drainage maps make it challenging to model these agricultural landscapes. Thus, we have been developing an agricultural tile drainage map across the US Midwest at 30-m resolution using optical and radar satellite imagery along with soil- and climate-related variables within the Google Earth Engine (GEE) cloud-computing platform. A new regional training dataset includes point data that are manually identified from multi-resolution aerial imagery and compiled from other literature and agency sources that have been assembled to train a random forest classification. Aridity, subsurface soil moisture, normalized difference water index, and VV(vertical transmit, vertical receive) polarization from Sentinel-1 Synthetic Aperture Radar have shown the higher importance in the classification and improved the over accuracy (~90%). This provides insights into decision-making around tile drainage installation. We also compare the outputs to other currently available products to quantify the improvement in classification accuracy. The tile drainage maps are valuable inputs for characterizing the coupling of hydrologic and chemical/biogeochemical process modeling, informing sustainable water management practices, and providing environmental managers with needed information to reduce nutrient loads.

Luwen Wan

Postdoctoral Fellow

I am passionate about using data science, process-based modeling, remote sensing and machine/deep learning to explore how land use, climate change, and management impact water sustainability across diverse landscapes, and from watershed, regional to global scales.