8776 Rate this article:

Mapping Crop Residue with Remote Sensing Data

Nate Oscar Monday, May 2, 2022

Problem

Over the last 150 years it is estimated that more than half the Earth’s topsoil has been lost due in part to soil erosion. Soil erosion leads to infertile land as well as increased pollution and sedimentation in waterways. Degraded soil is less able to hold onto water which increases the risk of flooding and drought. Soil health is a major concern to farmers who rely on it for their livelihoods, as well as for the global community in terms of food security.

Sustainable agricultural practices, such as low-till farming, go a long way toward maintaining healthy soil. In fact, high levels of crop residue cover (CRC) on a field, either from a cover crop or the stems, stalks, and remaining matter from last year’s crop, translates into better soil conservation. This is because CRC provides a barrier against soil erosion, reducing nutrient runoff and improving soil quality.

Left image: Area of study. Right image: Example Sentinel-1 Swath A single intensity image extends 250km, in this case from Belleville to Oakville. This makes ESA SAR data suitable for mapping very broad areas.

Solution

Timely and accurate estimates of crop residue and living crop coverage during the non-growing season (NGS) provides useful information about the health of the agricultural system in general, and more specifically about agricultural practices and impacts to the surrounding environment. This includes information about tilling practices, downstream nutrient loading estimates, and economic loss estimates for farm operators. Jeff Meyer, Information Technology specialist with the Lower Trent Conservation Authority, a community-based environmental agency in Southern Ontario, Canada, set out to use remote sensing data to map crop residue in the region.

Satellite remote sensing is a useful tool to estimate CRC efficiently on a regional scale. Meyer’s study used both optical and microwave Synthetic Aperture Radar (SAR) data to estimate CRC for the Lower Trent Region. SAR data is particularly useful for this type of study as it is not dependent on clear weather conditions to get an accurate picture of what’s happening on the ground. However, SAR is not always a go-to data type since it has an undeserved reputation for being difficult to work with.

NV5 Geospatial offers the software package ENVI® SARscape Analytics, which provides easy-to-use workflows for some of the most common SAR processing applications. "ENVI SARscape Analytics opened a window for us to operationalize SAR in a way we didn't think possible. Before it seemed abstract and experimental, but now we understand that's far from the truth," said Meyer.

Image: Optical vs SAR imaging

“ENVI SARscape Analytics opened a window for us to operationalize SAR in a way we didn't think possible"

- Jeff Meyer
Lower Trent Conservation Authority

Meyer acquired Sentinel- 1 and Sentinel- 2 images over a period of time "We made heavy use of the Sentinel Download tool and Time Series tool in ENVI SARscape Analytics to extract a wealth of information for this project,” said Meyer. The Sentinel Download tool provides automatic download of Sentinel-1 and Sentinel-2 data. The workflow lets users choose an area of interest, dates, and other filters, and downloads multiple Sentinel-1 and Sentinel-2 scenes to use instead of having to manually download them one at a time from a web service.

“ENVI SARscape Analytics greatly reduced the burden of downloading and processing Sentinel-1 data,” said Meyer. “In the future we are interested to explore the Flood Mapping tool in the software to get near near-real time observations during flood events. Flood warning and forecasting is an important part of what we do here, so getting actual observations is very useful for future planning."

Results

A methodology was introduced in the study that was based on Sentinel-1 and Sentinel-2 derived variables to provide a realistic index of winter shoulder-season soil cover conditions. The index was classified into four categories ranging from bare soil to full cover (dense residue or living crops). A total of 184 in-situ ground observations with geotagged photos were collected in three rounds during late-fall of 2020, and early-spring/late-fall 2021, to provide analytical support for the methodology. A robust accuracy assessment based on 31 sites was undertaken in the spring of 2022, concluding with an overall accuracy of 83.87%. The technique was designed to provide reliable field level information within the range of physiographic settings in the study area.

Image: Example profile of relative variation across an arbitrary linear transect ~400 metres. Samples are collected along the red line, from left to right in the COV intensity image, with extracted coefficient of variation values plotted. Letters are provided to help compare the image to the plotted values.

Summary

Overall, this study demonstrates the accuracy and efficiency of using SAR remote sensing products created with ENVI SARscape Analytics to provide timely information about farm field soil cover conditions on a regional scale. The study provides a relativistic index of soil cover by crop residues and living crops. The resulting maps can be summarized in numerous ways depending on the needs of the analyst and this study outlines a few examples.

For additional information, see the complete study, Crop Residue Mapping in the Non-Growing Season, created by Jeff Meyer and the Lower Trent Conservation Authority.