Keywords

1 Introduction

Precision agriculture developed in the early 1980, and precision agriculture is a subject jumping. Precision agriculture is the main system of modern agricultural production operation according to the spatial location, timing, spatial variability and quantitative to achieve, it is mainly the information technology to support, most of the data mining technology [1].

Precision agriculture is mainly composed of intelligent decision-making technology, field information collection technology, and intelligent equipment technology. Through the collection of farmland information for digital analysis and processing, intelligent decision-making, mechanization of farmland information and application of information, can increase crop production, increase efficiency and increase the income of farmers [2]. The main idea of precision agriculture is that agricultural production can be adjusted according to local conditions. It can be reasonably invested, not wasted, and managed scientifically. Precision agriculture has achieved orderly management in space and time, and improved the efficiency of agricultural production. Through the use of modern agricultural machinery operation scale, improve the agricultural production efficiency and the efficiency of agricultural production, greatly changed the peasant workers will always work hard in the fields only in traditional agriculture, backward production mode and the “inspiration” extensive industry management [3].

The productivity of the soil is mainly composed of soil fertility, soil fertility mainly refers to organic matter, nutrient content, soil texture, soil thickness and other factors in the soil, these factors have different effects on soil fertility, and have different constraints on soil fertility.

For the study of soil fertility classification, there are several situations that appear below: Zhang et al. [4] applied the mathematical method of principal component analysis and discriminant function discriminant to classify soil fertility. In addition to the mathematical model approach, quantification of soil fertility has also been studied, so far, however, quantitative indicators have not appeared, so that the size of soil fertility cannot be calculated. Liu and Liu [5] proposed that the soil per unit area under general conditions should be used to measure the amount of soil formed by plants to measure soil fertility. Because of the different types of soil use, the annual crop yield of the farming soil; the annual growth of timber for the forest soil; the grassland soil is measured by the annual grass growth, from an ecological point of view, soils of all types of use can be measured by solar energy (Joules), which is fixed by plants in the unit area throughout the year.

2 DBSCAN Algorithm

2.1 DBSCAN Algorithm Concepts

DBSCAN algorithm is a density based clustering analysis method. The algorithm defines clusters as the largest set of points connected by density, and divides the regions with high density into clusters. The kernel idea of clustering is to measure the density of the space of the point with the neighbor number in the neighborhood of a point [6]. For example, the object p is the center, and epsilon is the radius of the region, that is, within the epsilon neighborhood of p, including at least one positive integer Minpts objects, and p is the core object, fields containing Minpts objects are clusters, otherwise p is on the boundary of a cluster and is called boundary point [7, 8]. The following explains the relevant definitions:

2.1.1 Directly Density-Reachable

If the p is the core point, q in the - neighborhood of p, the p direct density of up to q (Fig. 1).

Fig. 1.
figure 1

Directly density-reachable

Full size image

2.1.2 Density-Linked

If an object o exists in the object set D, which causes the object p and q to be reachable from the object o about epsilon and Minpts, then the object p and q are connected to epsilon and Minpts density [9].

The definitions are parsed using Fig. 2 below, in Fig. 2, given the radius of the field epsilon, the minimum number is Minpts = 3. According to the above definition, point m, o, p, and r core objects because there are at least 3 objects in their respective epsilon fields. In addition, it is also observed that m is directly accessible from p and vice versa. q is direct density reachable from m, but m is not directly reachable from q, because q is not the core object. The density from q to m is up to m, and the direct density of p from p can reach up to q so that the density is reachable. Similarly, p and s are density reachable from o, and o is density accessible from r.

Fig. 2.
figure 2

Density-reachable and density-linked

Full size image

As a result, o, r, and s are density-reachable between each other.

2.2 DBSCAN Algorithm Progress

Input: n data objects, radius epsilon, minimum number Minpts.

Output: all clusters that reach the density requirement.

Algorithm processing flow:

  • Step1 Extracts an unprocessed point from the data object;

  • Step2 IF points out is the core point, THEN finds all objects reachable from the point density, forming a cluster;

  • Step3 ELSE The point that you take is the edge point (non core object). Jump out of this loop and look for the next point;

  • Step4 Loop Step1 to Step3 until all points are processed [10].

3 Maize Precision Fertilization

3.1 Data Acquisition

In Nong’an County of Jilin province Chen hometown pilot application, access to land information using the 3S technology, grid according to the distance of 40 m * 40 m. Soil sampling was carried out in the divided mesh, and 152 sampling points were obtained, which were collected for soil nutrient content, including four values of organic matter, available phosphorus, available nitrogen and available potassium.

3.2 Data Standardization

Since different data has different dimensions, it needs a unified dimension to compare the data, that is to standardize the data. normalization formula is:

$$ \text{g} = \frac{G - \hbox{min} }{\hbox{max} - \hbox{min} } $$
(1)

3.3 Soil Fertility Grading

The DBSCAN algorithm is used to cluster the processed data, that is, to classify the soil fertility, and each soil grade is called the cluster. In this paper, WAKA data mining tools are used to download DBSCAN data management package, set (field radius) 0.5, Minpts (epsilon field minimum point) is 10 of the cluster, the soil is divided into six cluster classes, that is, six levels.

According to the number of soil plots at each level, the four attributes of all soils in different levels are calculated, the average value is summed, and sorted according to the size of the sum, and the bigger one is ranked, and so on.

According to the sum of the average values of the attributes in the six clusters, the soil classification results are shown in Table 1.

Table 1. Soil classification result
Full size table

3.4 Calculation of Fertilizer Application

A fertilizer calculation model was established, and the nutrient balance method was used to calculate the amount of fertilizer needed for the nutrient content in different soil levels. We calculate the average of four attributes in different soil levels and substitute them into the formula of nutrient balance method, the required values here are the raw values before data processing, and the data values are shown in Table 2.

Table 2. Average properties of different soil classification
Full size table

Calculation of nutrient balance fertilization model by fertilizer application rate (2):

$$ sf = \frac{cl \times xs - cd \times ys}{hl \times ly} $$
(2)
sf::

Fertilization; cl: Corn yield target;

xs::

Grain corn nutrient absorption amount per 100 kg;

cd::

Soil nutrient determination;

ys::

Soil available nutrient conversion factor;

hl::

Fertilizer nutrient content;

ly::

Fertilizer utilization season.

Taking the mathematical model of fertilizer application rate as an example (The content of P2O5 was 46% in the application of diammonium phosphate):

$$ {\text{Phosphate}}\,{\text{fertilizer = }}\frac{{{\text{Corn target amount}}\, *\, 0. 0 7 { } - { 0} . 0 3\, *\,{\text{Soil nutrient content}}\, *\,{\text{Soil available nutrient conversion factor}}}}{{ 0. 4 6\, *\,{\text{Fertilizer utilization rate}}}} $$
$$ {\text{Soil available nutrient conversion factor}} = \left\{ \begin{aligned} \frac{{1578.8*{\text{Soil nutrient content}}^{{{ - 0} . 9 8}} }}{ 1 0 0} \hfill \\ \frac{{1068*{\text{Soil nutrient content}}^{{{ - 0} . 8 3 2}} }}{ 1 0 0} \hfill \\ \frac{{ 7 3 2 * {\text{Soil nutrient content}}^{{{ - 0} . 7 4 9}} }}{ 1 0 0} \hfill \\ \end{aligned} \right. $$
$$ {\text{Blank area yield}} = \frac{{ 0. 3\, *\,{\text{Soil nutrient content}}\, *\,{\text{Soil available nutrient conversion factor}}}}{ 0. 0 2 2} $$
$$ {\text{Fertilizer utilization rate}} = \left\{ {\begin{array}{*{20}l} {\frac{{ ( 4 3. 4- 0. 0 2 4 )\, *\,{\text{Blank area yield}}}}{100}} \hfill \\ {\frac{{ ( 3 6. 6- 0. 0 2 5 )\, *\,{\text{Blank area yield}}}}{100}} \hfill \\ {\frac{{ ( 4. 6- 0. 0 3 5 )\, *\,{\text{Blank area yield}}}}{100}} \hfill \\ \end{array} } \right. $$

Yield = 10000 kg/hm2

According to the nutrient balance method, the data of Table 2 are replaced by (2), and the specific amount of soil fertility at each level of soil as shown in Table 3 can be obtained.

Table 3. Soil nutrient fertilizer rate
Full size table

4 Conclusions

In this paper, the DBSCAN algorithm is used to classify the soil, and the classification results are applied to the corn precision fertilization decision, The experimental results of demonstration and application in Nong’an County of Jilin province Chen hometown. The average amount of fertilizer is 560 kg/hm2 (the average amount of fertilizer is calculated by the total amount of fertilizer divided by the total area), compared with the traditional fertilizer 608.42 kg/hm2, saving fertilizer 48.42 kg/hm2; The average yield of the experiment is 8313 kg/hm2, which is 930 kg/hm2 higher than that of the traditional output (the traditional values are derived from the statistical yearbook of China). Indeed, the purpose of reducing chemical fertilizer input, improving soil environment, increasing production and increasing income has been achieved.