Variable Rate Phosphorus and Potassium_OneSoil blog_cover

How we calculate variable rate phosphorus and potassium

Determining fertilizer rates from the results of an agrochemical soil survey is an inefficient method. So we learned to restore data about the yield using satellite imagery. That’s how our free calculator of fertilizers for differentiated application in the OneSoil platform works.

About the importance of phosphorus and potassium

Along with nitrogen, phosphorus and potassium are the main elements needed for plant nutrition. Potassium plays an important role in the synthesis of proteins, it stimulates photosynthesis and increases the ability of plants to tolerate frost. Phosphorus is needed for energy processes in plant cells. It helps roots to grow and affects the quality of seeds.

Unlike nitrogen, phosphorus and potassium are accumulated in the soil. With the increase of fertilizer rates, a farmer can increase their concentration in the soil. However, there is an optimal level of phosphorus and potassium for each crop. After reaching it, these elements won't be absorbed by plants. Sometimes their high content brings negative effects. For example, if the potassium level is too high, plants begin to absorb magnesium and calcium more slowly.

To save money and collect higher yields, it is important to calculate variable fertilizer rates correctly for each zone.

The principles of calculating fertilizers

Most methods are based on measuring the soil nutrient level and on calculating the planned yield. From the yield, the farmer estimates the removal of nutrients, that is, how much nutrients plants will take from the soil. Knowing this, you can calculate the dose of fertilizers that will provide the needed level for this desired yield.

Usually, the logic is as follows:
  • If the level of phosphorus and potassium is below the optimum one, it is necessary to apply more fertilizers than the predicted removal.
  • If the level of phosphorus and potassium is above the optimum one, then it is necessary to apply fewer fertilizers than the predicted removal.
It sounds simple, but the problem is that the nutrient level and productivity greatly differ from zone to zone.

The level of phosphorus and potassium can be estimated only by conducting an agrochemical soil analysis. This is an expensive method, and the selection of soil samples requires time and special equipment. Many commercial companies offer such a service. But it is still tricky: there is a clear methodology for carrying out the analysis itself, but there are different views on how to properly select soil samples.

How soil samples are taken

There is point sampling and grid sampling. In the first case, the sample is taken at a specific point in the field and its coordinates are fixed. Then by the results of laboratory analysis, experts make predictions about the nutrient level between sampling points. In the second case, the field is divided either into equal rectangles or into "cells" defined by maps of yield, relief, electrical conductivity or other indicators. In each of these zones, one composite soil sample is taken.
Why point sampling doesn’t work

Here are two maps of Belarusian fields 20 hectares in size. We made them from the results of a survey conducted on May 20, 2018. On the left, there is a map obtained at a sample density of 100 meters. On the right, the density is 50 meters. Obviously, the maps are different. If a farmer would apply fertilizers from the map with a density of 100 meters, he would increase the spatial variability of phosphorus and potassium. But sampling with higher density is expensive and difficult and few farmers can afford it. Hence, point sampling is quite ineffective.
Maps of potassium distribution_OneSoil blog
Map of potassium content made from point sampling. On the left — density of 100 meters, on the right — 50 meters.
Why grid sampling doesn't work

Since 2014, we are testing the effectiveness of the allocation of field zones with various methods that are practiced by agro consulting companies. First of all, the allocation of zones using maps of electrical conductivity, relief, and humus content, as well as using soil maps and a grid of rectangles.

We surveyed dozens of fields in Belarus, Slovakia, Lithuania, Ukraine, and Russia. For some fields, a detailed agrochemical survey was conducted. Our data demonstrates that the obtained zones on the fields are heterogeneous in the level of phosphorus and potassium regardless of the method.
For instance, there are areas of high and low level of potassium and phosphorus even across a zone 1.5 hectares in size with a homogeneous humus level.
The same is true for zones allocated by all other methods. Since all methods of soil sampling are ineffective, we calculate the rate of phosphorus and potassium by yield data.

Why do we calculate variable rates by yield?

It is easier to understand how yields have changed than to measure different nutrients levels. Many companies equip agricultural machines with special sensors. They record the mass of grain that goes to an elevator and link this value to the geo coordinates.
Yield map_OneSoil blog
Yield map
We collected data on field productivity in different countries in Europe. Most of them are in Belarus: in 2015 — 2018 we surveyed 39 thousand hectares. In Ukraine and Russia — 2.5 thousand hectares, and about 1,000 hectares in the Czech Republic, Spain, Lithuania, Italy, and Slovakia. Using this data, we can say that the yield varies greatly from site to site on most fields.
For example, on a field of winter wheat, there can be both areas with a yield of 1.5 tons per hectare and areas with a yield of more than 8 tons per hectare.
The most difficult thing here is to find information. We rarely met farmers who had data on crop performance spanning more than 2 years. Some don’t have special sensors on their equipment, some don’t activate them. Even if there are sensors, combines should be calibrated, and the collected data should be correctly processed. So at every stage something can go wrong which makes collecting yield data a complex procedure.

To solve this problem, we learned to restore data about yields for previous years with satellite imagery.

How we learned to restore crop yield data

We used our database on fields with a total area of 40,000 hectares and satellite imagery for several years. We compared actual yields and vegetative indices at different stages of plant development. This allowed us to model the relative yields. In other words, we don’t know how many tons have been collected for every part of the field, but we know how productive this part was (in percent) compared with the whole field.
For example, from a whole field 10 tons were harvested, and from a specific part — 2 tons. It means its yield in relative value is 20%. This is the type of information we can determine.
Here is the yield data for 4 years: for 2014 and 2016 it was modeled by us, and for 2015 and 2017 — actual data, collected by a combine. Despite the difference in weather and in cultivated crops, most parts of the field are stable in terms of yield.
Yield real and simulated_OneSoil blog
Maps of modeled and actual data
It is clear that in areas with a high yield more nutrients are being removed, and in areas with a low yield — less. Considering this fact, the farmer can precisely determine the variable rate of phosphorus and potassium for each part of the field.

Not to forget, there are other factors that affect yield. For example, a farmer applies phosphorus and potassium to a field aiming to harvest 8 tons per hectare, but in certain parts, he collects 2 tons per hectare from year to year. There is hardly a problem in the shortage of phosphorus and potassium there.

Lack of water, low humus level or suboptimal level of soil acidity also reduce yield. Therefore, it is important to not only accurately determine the variable rate but also to find and eliminate yield-limiting factors.

How the OneSoil calculator of phosphorus and potassium works

Log in to OneSoil platform and select the field for which you want to calculate the variable rates of fertilizers.
Specify the desired crop and planned yield — its value should correlate with the actual field indicators. You can calculate fertilizer rates for most grain and oilseeds crops.
Specify previous crops for the last three years. Based on satellite images, we will determine the zones of stable yield for each of them. This allows to identify three zones with average indicators.
We will automatically calculate the removal of nutrients for each zone and then determine the fertilizer rates.
Select the type of your onboard computer and download the task file with prescription map.
Do you like this post?
Related Articles