Agriculture Canada has funding to update the Disease Risk Tool (DiRT1) it developed in 2016 to include crops beyond canola.
DiRT1 combines information from satellites and user inputs into a prototype web application that can be used to investigate the accuracy of crop-disease forecast models.
“In this first prototype we also integrated geospatial data from Environment Canada on temperature, precipitation and the application allowed users to provide information on seeding density and disease history in the field or in the region,” said Heather McNairn of Ag Canada.
McNairn specializes in using space for agricultural applications and she said DiRT1 and DiRT2 are not designed to forecast or identify when a disease is present in the field.
Instead, they pull together the risk factors and produce a geospatial-information tool to query disease risk.
“Having said that, once you pull together all of this geospatial data of course it provides a really good tool to investigate the accuracy of existing models and to test new models as well,” McNairn said.
Sclerotinia in canola has many risk factors that create conditions that enable the disease to get a foothold, and DiRT 1 focused on risk factors that can be monitored with satellite-based data.
“From satellites, we can tell you what the cropping history is of a particular field, we can identify crop phonology and, in this particular case, when canola is flowering we can provide an estimate of how wet the soils are,” McNairn said.
To obtain the cropping history the annual crop inventory is used, which provides estimates of what’s growing in every field across Canada every year.
The annual crop inventory has been active for a decade, so there is now 10 years worth of data available the DiRT tool can use.
To determine how wet the soils are, Ag Canada works with Environment Canada, which has a land surface model that assimilates satellite data and one of its outputs is surface soil moisture.
“What the land surface model does is it provides an estimate of volumetric soil moisture every single hour at a 100 metre resolution,” McNairn said.
“This is an experimental model. So, of course we spent time working with Environment Canada to validate those surface soil moisture maps.” McNairn said.
However, she said just knowing how much water is in the soil isn’t enough so Environment Canada created a secondary model to help gauge how saturated the soil is relative to its field capacity.
“The third product that we developed was a soil moisture persistent product. Because, of course, it’s not just how wet and how saturated soil is, but how long the soil stays wet,” McNairn said.
“It tells us over the last two weeks how saturated, how close to field capacity was that soil.”
The information on cropping history and on how wet the soils are is combined with information on the crop’s growth stage, a method developed by Canadian industry A.U.G. Signals.
“It assimilates satellite data into this machine learning algorithm and what the algorithm does is it provides an estimate of the growth stage of any particular field on a daily time stamp,” McNairn said.
She said an important piece of the new project is to build a high-resolution data pipeline for the geospatial data, and produce a web application that could run in near real-time.
“We want to extend it to other crops like wheat and soybeans because we have a lot of that information in terms of cropping history and soil moisture. It’s a matter of applying it to other diseases and other crops,” McNairn said.
“The vision is to create a pipeline to pull in all of this geospatial data at 100 metre resolution and provide frequent updates and near real-time in terms of disease risk for any particular disease in the Canadian Prairies.”