Advanced Modeling of Soil Erosion, Sediment Delivery and Nutrient Export from Iowa Watersheds

Date: 
Aug 2020

Issue

Soil sediment and sediment bound phosphorus export from agriculturally dominated watersheds is a global crisis that has far-reaching, potentially severe environmental and economic impacts. Multiple sources are known to contribute to surficial exports from these watersheds including in-field sheet and rill erosion, ephemeral gully erosion and streambank erosion. However, accurately quantifying suspended sediment and P export from watersheds is difficult.The Water Erosion Prediction Project (WEPP) may be utilized to help quantify sediment yields from agricultural fields. Sensitivity analysis and validation indicated that WEPP has high runoff and sediment yield accuracy for small watersheds. Still, scaling up WEPP Watershed to a HUC-12 scale may be difficult. The Daily Erosion Project (DEP) may be used to scale up WEPP, but these new outputs must be ground-truthed with field-collected data.

Objective

This project will help to identify and quantify the effects of ephemeral gullies on sediment and P loss. Ultimately, this project will work towards delineating sources of sediment and P export, i.e., hill slopes or ephemeral gullies, from Iowa watersheds, using Walnut Creek watershed in Jasper County, Iowa, as a test for soil erosion, sediment and nutrient export modeling.

Approach

The University of Iowa will perform a field investigation to collect and analyze actual soil sedimentation and export data at individual subcatchments within the study area to estimate sediment bound P exported from the watershed and calibrate related models. The USDA-Agricultural Research Service will provide and conduct an annual drone-based LiDAR field investigation in multiple land-use areas within the watershed to use as input data for the WEPP model. Iowa State University will complete WEPP models for the study area and scale up results using the DEP framework.

Award Number: 
2020-02

Project Updates

Note: Project reports published on the INRC website are often revised from researchers' original reports to increase consistency.

March 2024

FINAL REPORT

This project's primary purpose was to utilize a high-resolution WEPP Watershed model and in-stream turbidity and discharge data to accurately estimate sediment delivery ratios for a small watershed. By doing so, more accurate estimates of sediment export may be made compared to using a RUSLE erosion model or the dated NRCS sediment delivery nomographs.

In this study, we estimated soil erosion and sediment and nutrient export from a small watershed in Iowa during drought conditions. The WEPP Watershed model identified 10 events in 2022 that produced significant runoff and yielded sediment. The smallest event (13 mm of precipitation) produced 1,283 m3 of runoff at the outlet and eroded 2,884 kg of soil. Utilizing our in-stream sensor data, SDRs estimated using the WEPP Watershed model were 1.4% while RUSLE model estimates resulted in SDRs of 0.1%. Although the estimated SDR was higher using the WEPP Watershed model results, overall sediment yield was much lower (111 Mg compared to 1402 Mg).

Finally, we estimated that during the study, 66 g of total phosphorus (TP), 3 kg of total nitrogen (TN), and 28 kg of total carbon (TC) were exported from the catchment. We conclude that by utilizing the high-resolution WEPP Watershed model, accuracy and precision was gained that can help to inform researchers and land managers about best management practices to further reduce sediment and nutrient losses -- even during years of minimal rain, soil erosion and sediment export.

See the more complete Final Report to INRC for this project.  

Other activities

Two presentations. 

Findings have been submitted to the journal Applied Engineering in Agriculture.

July 2023

During the project period from January through June 2023, the project has continued on track and multiple significant accomplishments have occurred. In this latest period, we have developed a WEPP Watershed model with GeoWEPP for the upper reaches of Walnut Creek of the Des Moines River (HUC071000081505). We are using data from the latest statewide Lidar survey at 5-meter resolution for general basin properties like watershed area and hillslope steepness and the sub-meter drone lidar surveys for higher resolution properties such as channel morphometry. We are using crop management information from the Daily Erosion Project and Agricultural Conservation Planning Framework for crop rotations, field boundaries, and tillage intensity to parameterize the management files. Catchment boundaries used in GeoWEPP are shown in figure 1. We are currently troubleshooting difficulties with using Daily Erosion Project corn and soybean files (calibrated for current crop yields) within the GeoWEPP framework. We have default WEPP files (using lower yields) up and working and will use these to complete the runs during the next project period if we cannot get our preferred files running in a timely manner. The ARS has now completed their LiDAR flights over the study area with their drone and data processing is being finalized. In addition to incorporating the sub-meter drone data into the WEPP Watershed models, we intend to utilize the year over year drone data to attempt a mass-balance estimate of soil loss from each field. The live turbidity sensor and stream stage sensor have been redeployed for the 2023 field season. In addition, the IGS has leveraged their resources and installed a Nitratax at the site to measure nitrate levels in-stream. Data analyses are currently underway to determine sediment delivery and nutrient export for the study area in 2022. We will compare erosion and sediment delivery values for multiple different erosion models and make estimates of phosphorus export based on our field measurements of soil P and sediment delivery estimates. Data analysis and reporting will occur over the next project period. We intend to prepare a manuscript for peer review highlighting sediment and nutrient export from our study area.

December 2022

During the project period from July to Dec. 2022, the project has continued on track with multiple significant accomplishments. Project team members have been working on some upgrades for the Walnut Creek watershed during the last six months that will improve the ability to better represent the land surface in the DEP. First is an increase in DEP input sampling intensity, which allows creation of  ‘oversampled’ versions that contain 2-100x more random flow paths than a standard DEP run in each DEP sub-catchment. A DEP sub-catchment generally represents 20-100 hectares, allowing better representation of the variety of hillslopes in a sub-catchment over the standard 1 random hillslope per sub-catchment. This is up and running now. The second upgrade is an update to include the latest statewide DEMs from 2020. These DEMs are 2x better resolution (0.7m average GSD; 4x the points) than the old 2008 DEMs but are not as good as the drone data (0.1m average GSD). The DEMs are generated at 1m this will be incorporated with the increased sampling intensity above to give a robust DEP estimate of runoff and soil loss in the top of the Walnut Creek watershed.

Brian Gelder has also been working with Allisyn Stanfield and Kevin Cole to develop some better ways to process the drone data, so it better fits the needs of DEP. One option is to use PDAL, a point data processing library that can filter to bare earth fairly well. The other option they’ve been pursuing is using CloudCompare to do some point classification that should allow us to incorporate their DEMs into the project’s model of the top of Walnut Creek. In the fall of 2022, the ARS completed an additional LiDAR flight over the study area with their drone. The area was flown at 50 meters above ground and with a point density of at least 200 points per square meter. Initial screening indicated appropriate overlap and coverage of the study area. A fourth flight is planned for spring 2023.

The live turbidity sensor and stream stage sensor collected data continuously through late November when the sensors were removed for winter. Data analyses are currently underway to determine sediment delivery and nutrient export for the study area in 2022.

June 2022

During this project period, the project has continued on track and multiple significant accomplishments have occurred. A one-year no-cost extension will allow for 2 additional years (fall and spring) of drone LiDAR flights. Flights were delayed in year one when acquisition of the ARS’s drone took longer than expected. In the spring of 2022, a second data collection flight occurred with the ARS’ LiDAR capable drone. The Walnut Creek study area was flown on 4/11/2022. The area was flown at 50 meters above ground and with a point density of at least 200 points per square meter. Initial screening indicated appropriate overlap and coverage of the study area. A third flight is planned in fall 2022 and a fourth flight in spring 2023.

The project team is working to generate super high resolution (0.5 m XY) elevation models from the ARS LiDAR drone data captured by Kevin Cole. The data generally aligns well with the previously captured statewide datasets. However, filtering the trees fully from the elevation surface has proven difficult. See Figure 1 for a visualization; the light gray areas show good alignment between the surfaces, and the black areas show the drone surface being too high. Additional filtering mechanisms are being investigated to remove these points from the elevation surface or manually remove them via classification.

The project team has also been working to represent the monitored watershed in WEPP Watershed, a multi-hillslope implementation of the WEPP sheet and rill erosion model with erodible channel connections between the hillslopes. A simplified representation is shown in Figure 2 with the channel ending at the sampling location. Additional WEPP hillslopes and channels are being added to represent hillslopes, waterways and other channels conducting flow to the channel shown as C1. Next, this model will be fed weather data from the nearest Daily Erosion Project climate file.

The live turbidity sensor was reinstalled in early April 2022. Further, a pressure transducer is currently installed in the stream to quantify stream discharge in conjunction with the stream stage/discharge model that we developed last year. The sensor is named WQP0001 and can be monitored here: https://iwqis.iowawis.org/app/?iwqis=/sensors-map. In winter 2022, soil samples were analyzed for bulk density and nutrient concentrations. Nutrient stocks were determined. Initial estimates of exported soil sediments from the study watershed in 2021 were determined based on RUSLE soil erosion modeling and TSS estimates.

A third round of drone LiDAR flights is anticipated to take place in fall 2022. The turbidity sensor will be deployed in the stream through fall 2022. For the modeling effort, the team will work on implementing hillslopes in the desired locations in the coming quarter and creating channel files and channel erosion estimates.

January 2022

During July to Dec. 2021, the project continued on track with significant accomplishments. Following final calibration of the ARS’s LiDAR capable drone, the updated UAV LiDAR system was used for test flights in October and LiDAR processing steps were further refined to produce high resolution digital elevation models. The Walnut Creek study area was flown on 11/09/2021 and 12/14/2021. The area was flown at 50 meters above ground and with a point density of at least 200 points per square meter. Initial screening indicated appropriate overlap and coverage of the study area.

Modelling progress was somewhat delayed during this project period since drone LiDAR data was not available when expected. We anticipate this to change now that the data has been acquired. A second round of drone LiDAR flights are scheduled for spring of 2022.  

Progress was made progress on related projects that should help speed implementation. Algorithms have been developed and tested that extract more reasonable gully elevation estimates than the previous simple punch-and-fill approach that ACPF/DEP uses. This will ensure gullies have a more realistic continual downhill slope than the previous stair step pattern where the steps of same elevation could be quite long within channels. Algorithms have also been developed to compare DEP estimated runoff and erosion with gauge measurements for STRIPS sites. These algorithms can be easily ported to other locations and will allow us to compare the measurements at the gauge location with WEPP watershed estimates for the site.

Researchers have also been building a way to add site specific management practices into DEP using the ISU GIS Facility’s BMP database. This will involve adding a new dataset to the DEP data sampling process and developing a WEPP management file for each practice in the BMP database and one for any additional practices we want to encode, like STRIPS.

For most of this period, the live turbidity sensor remained installed in the stream at the study site. Data was collected from April 2021 to late September 2021 when the stream level decreased to a point where data collection was not possible. The sensor transmits live data when installed and may be monitored here: at https://iwqis.iowawis.org/app/?iwqis=/sensors-map (sensor: WQP0001). The turbidity sensor will be redeployed in the stream in the spring of 2022.

In November 2021, following crop harvest, soil samples were collected for nutrient and bulk density analyses. Results from these analyses will allow for the estimation of eroded and exported nutrient stocks.

June 2021

During Jan. to June 2021, significant progress has been made towards the goal of modeling soil erosion, sediment delivery and nutrient export in agricultural subbasins of Walnut Creek watershed in Jasper County, Iowa. Due to COVID shutdowns, the expected arrival date of the ARS LiDAR capable drone was pushed back considerably. In early spring, the drone was received by Kevin Cole. Four test flights were conducted near Ames in the spring of 2021. The ARS is currently evaluating the point cloud data collected and determining the appropriate flight parameters and data processing steps to create a high-quality DEM for use in terrain analysis. The drone manufacturer may need to recalibrate the LiDAR sensor to produce the high-quality data that we require. Plans are now underway to attempt drone surveys this fall once crops are harvested.

The modeling components of the project are still on schedule. The DEP framework has been altered and will enable the team to create one-off versions of the DEP data structure to build and test this project’s custom setup of DEP without influencing the production runs of DEP. We also altered the DEP framework to create DEP hillslopes in defined locations so we can simulate hillslopes in desired locations to create the hillslope part of the hillslope/channel combinations needed to run WEPP watershed and estimate gully erosion. A team member has been working on custom exports of DEP model run results so simulations can be analyzed on a site by site or year by year basis. Finally, during this project period, a “live” turbidity sensor was installed in the stream at the study site. It has been collecting data since early April. Sensor WQP0001 can be monitored at: https://iwqis.iowawis.org/app/?iwqis=/sensors-map.

This work contributed to advances in sediment delivery and erosion modeling shared with IDALS.

January 2021

The project kicked off with a team planning meeting in August 2020. In November, the research team met in Walnut Creek watershed and confirmed the location of three study area catchments/agricultural fields. Due to COVID shutdowns, the expected arrival date of the ARS LiDAR capable drone was delayed and flights were not conducted in the fall as originally anticipated. A pre-thaw flight is planned for February 2021 and another before planting in April 2021.

During this reporting period, a Python script was developed that will take current Daily Erosion Project (DEP) inputs of rainfall, elevation, soils and ACPF management, along with a point location of interest (monitoring station, gully head), to generate a DEP flowpath that simulates the hillslope flowing to that point. This will enable modeling of that hillslope in WEPP and validating model estimates versus measured data. This work was recently completed – researchers have not been able to validate any STRIPS2 watersheds yet. This was the first step in building a WEPP watershed implementation within DEP. WEPP watershed estimates runoff and erosion across multiple hillslopes and routes this to a ‘channel’ where channel erosion and deposition may be estimated. These WEPP/DEP watershed estimates will be much closer to estimates of delivery to a stream or waterbody.

Category: