The identification of pests and diseases of rice plants using sentinel-2 satellite imagery data at the end of the vegetative stage

One of the factors that affect production yield of cultivation crops, including rice plants, is pest/disease attacks. Various types of pests/diseases are often found in rice cultivation such as rats, stem borers and rice leaf folders. The damage caused by these pests/diseases can lead to production decreases and crop failure. To prevent an increase in pest infestation, monitoring can be carried out by employing remote sensing technologies such as Sentinel 2 satellite imagery. The identification of rice plant pests/diseases can be done using Vegetation Indices such as NDVI and NDRE, due to their sensitivity to plant chlorophyll. The purpose of this study was to identify infected pest and disease of rice plants in paddy fields by using Sentinel-2 satellite imagery. The methods of this study were spatial analysis and simple linear regression analysis, by looking at the relation between Vegetation Indices and rice productivity. The research results showed that based on the low value of NDVI and NDRE at the end of vegetative stage, some of rice fields were infected with pests and diseases. From field observation, it was found that the pests and diseases included rats, rice leaf folders, borers and blasts. Pests/diseases infestation in paddy fields lead to a low Vegetation Index value which results in low rice production. Comparing the NDVI and NDRE, NDRE was better in detecting pests and diseases due to its sensitivity to the plants’ chlorophyll.


Introduction
Many factors affect crop yields in the cultivation crops, including pest/disease infestation.In rice paddy, pests and diseases contribute to the decrease of production in almost all growing season in Indonesia [1,2].Thus, as rice is the staple food of Indonesian people, this condition creates problem in fulfill the need of rice.Some significant pests and diseases in rice paddy field need to be monitored, such as blasts, rats, stem borers and rice leaf folders.Monitoring pests and diseases using conventional methods requires a lot of human labor, especially on large field.Therefore, exploring technologies for monitoring pests and diseases is necessary.One of technologies that can be used is remote sensing, where the monitoring can cover huge areas in a short time [3].
Sentinel 2 is one of satellites that can monitor pests and diseases.The satellite has a spatial resolution of 10 m and temporal resolution of 5 days.Sentine-2 consists of two twin satellites which scan earth's surface at a simultaneous angle of 180° and has 13 bands, including SWIR and VNIR [4].
Previous research has used Sentinel-2 to predict the distribution of brown leafhopper using Species Distribution Model (SDM) [5].The research results showed that there were differences in pixel values Identification of rice plant pests/diseases can be done by using vegetation indices which used some bands to calculate the reflectance of plant chlorophyll [6].Example of vegetation indexes are Normalized Difference Vegetation Index (NDVI) and Normalized Difference Red-edge Index (NDRE).NDVI showed the biomass parameter of plant [6], while NDRE can be used to identify chlorophyll content of plants.NDRE is known to be sensitive than NDVI, due to its red-edge band which can penetrate leave deeply than band red in NDVI [7].As it is essential to analyze the use of NDVI and NDRE in identify pest and disease outbreak, this study aims to identify infected rice paddy fields in Bantimurung District, Maros Regency, Indonesia by using Sentinel-2 satellite imagery.

Data acquisitions and processing
This study identifies the invasion of pests and diseases of rice plants at the end of their vegetative stage (35-55 days after planting.It was conducted through several stages, namely data collection, data processing, and data analyses.Primary data regarding pest and disease invasion were obtained through field observation and interviews.The primary data consists of pest and disease types, field location of infected pests and diseases, and production yields of selected fields.In the other hand, secondary data of Sentinel-2 level 1C satellite imagery data were downloaded from United States Geological Survey (USGS) Earth Explorer web page (https://earthexplorer.usgs.gov/).The first stage of Sentinel-2 imagery data processing was atmospheric corrections using semi-automatic classification tool in QuantumGIS (QGIS) software.Afterwards, NDVI and NDRE were generated by using raster calculator.

Data analysis
Data analysis was carried out using by comparing paddy fields attacked with pests and diseases with healthy paddy fields.The NDVI and NDRE imagery was classified using classification used in the previous research by by Rahaldi et al, 2013 and EOS data Analytics, 2021, which divided rice plants into four groups, namely bad, less healthy, healthy and very healthy (Table 1).The correlation between vegetation indexes and production yields was developed using linear regression analysis.

Results of vegetation index processing in the Vegetative Phase
The high and low value of the vegetation index is influenced by the health condition of the plant or the chlorophyll content of the leaves.Pests / diseases invasion will damage the structure of the leaf tissue [10].
The vegetative phase of the dominant rice plant was found in September and October 2021, in September the dominant rice was aged 0-30 DAP while from October 1, 2021 to October 21, 2021 the rice plant was in the final vegetative phase aged 31-50 DAP.In Figure 2, the NDVI transformation value is obtained from the mathematical combination of channel 8 (NIR) with channel 4 (Red) in the Sentinel-2 image at the end of the vegetative phase, namely October 1, 16, 21 and 26, 2021 using the formula NDVI = (NIR-Red)/(NIR+Red), while the NDRE transformation value is obtained by a mathematical combination of channel 8 (NIR) with channel 5 (Red Edge) with the formula NDVI = (NIR-Red Edge)/(NIR+Red Edge) at the same time.
The NDVI and NDRE values obtained have increased along with the increase in plant age caused by an increase in chlorophyll content so that the maximum index value at the end of the vegetative stage is at the age of 51 DAP (October 21, 2021) then decreases at the age of 56 DAP (October 26, 2021) because some paddy fields have begun to enter the generative stage.This is in accordance with the statement of Vitasari et al., (2017) which states that at the peak of vegetative growth there is a high level of greenness caused by the high content of chlorophyll.After this period, the greenness level will decrease, rice flowers will arise until yellowing (generative phase).Figure 2 shows the NDVI and NDRE imagery during the paddy plant vegetative stage [11].

Vegetation index
Analysis of the vegetation index of rice plants attacked by pests and diseases with healthy rice plants using NDVI and NDRE transformations can be seen in Figure 3 and Figure 4.The NDVI value of rice plants in the final vegetative phase attacked by pests/diseases with healthy rice plants in the planting season of September-November 2021 can be seen in Figure 3. Healthy rice plants have an NDVI value of 0.7620 to 0.8064, while pest-infested rice plants have an NDVI value of 0.3814 to 0.4788.The difference in value between healthy rice plants and rice plants attacked by pests is caused by changes in shape and color in rice plants.The NDRE value of rice plants in the final vegetative phase attacked by pests/diseases with healthy rice plants in the September-November 2021 plant season is very different from the NDVI value, this is because the NDRE Index is more sensitive to plant chlorophyll so it has a lower value than NDVI, rice plants attacked by pests have an index value of 0.2951 to 0.3897 while healthy plants have an index value of 0.5726 to 0.6598.
In rice plants that are healthy and attacked by pests, there are different treatments, farmers who have healthy plants spray 5 to 7 times during the planting period, thus preventing the widespread spread of pest attacks, while plants attacked by pests/diseases only spray 2 to 4 times during the planting period.

Relation of vegetation index to rice crop productivity
The relationship of vegetation index with rice plant productivity can be known through the equation in the form of the following graphic Figure 5  The calculation of the research data between the NDVI value and rice productivity above resulted in the equation y = 11.583x-3.6186where from this equation has a coefficient value of variation R² = 0.7943 which means that 79.43% of the variable y (rice production) is influenced by the variable x (vegetation index).With an R² value obtained > 0.5, it can be seen that the correlation relationship between the vegetative late-stage vegetation index through the NDVI transformation above can be categorized as good.This is in accordance with the statement of Nurwatik (2015) which states that from the regression equation with the value of the coefficient of determination R 2 > 0.5 has a good correlation [11].0.3000 0.3500 0.4000 0.4500 0.5000 0.5500 0.6000 0.6500 0.7000 Field Productivity (Ton/Ha) Vegetation Index Value (vegetation index).Thus, it can be seen that from the results of the correlation relationship between the vegetative late-stage vegetation indices through the NDRE transformation above has a good correlation.When compared with the NDVI vegetation index value, the relationship of the NDRE vegetation index with rice crop productivity is greater with a value of R² = 0.8077 while the transformation of the NDVI vegetation index has a value of R² = 0.7943.Therefore, the NDRE index transformation value has a better relationship with rice crop productivity when compared to the NDVI index transformation value.

Testing the correlation of rice crop productivity
Productivity correlation testing was carried out to determine the relation between productivity in the field and productivity based on indices using the regression equation of each vegetation index at a planting age of 51 DAP, the results of these tests can be seen in Figure 7. Rice productivity based on estimates and measurements of paddy yields in Figure 7 has a fairly strong relation.This is known based on the points that describe the relation close to the linear line, on the NDRE graph there are points that are on the linear line, but there are some points that are overestimate and underestimate, overestimate shows that the estimated productivity is higher than the productivity in the field while the underestimate shows the production estimate is lower than the productivity in the field.Estimates of rice productivity using the regression equation of each vegetation index can be seen in Table 2. Some rice fields have an estimated productivity that is less than field productivity and some also have productivity estimates that exceed field productivity.The difference in production results with production estimates is due to other factors that can affect rice production.This is in accordance with the statement of [12] which states that the small coefficient of determination causes the model made to be less accurate to estimate rice productivity according to reality as in the field.This is influenced by other factors.Other factors that can affect the level of rice productivity include the level of health of rice plants, irrigation systems, types of rice, pests, rice planting systems and fertilization.

Conclusion
Sentinel 2 satellite imagery can be used to identify pests/diseases in rice plants at the age of 30-50 HST by using the NDVI and NDRE vegetation indexes with pest-infected NDVI index values ranging from 0.11 to 0.42 while NDRE values range from 0.1 to 0.4.And the NDVI index is better to use because it has a good correlation with rice productivity.

Figure 2 .
Figure 2. (a)NDVI and (b) NDRE imagery at the end of the vegetative stage.

Figure 3 .
Figure 3. Comparative chart of NDVI values of infected and healthy rice plants.

Figure 4 .
Figure 4. Graph comparing the NDRE values of infected and healthy rice plants.

Figure 5 .
Figure 5. Graph of the relationship of NDVI transformation vegetation index with productivity of paddy fields at the end of the vegetative stage (51 DAP).

Figure 6 .
Figure 6.Graph of the relationship of NDRE transformation vegetation index with Productivity of paddy fields at the end of the vegetative stage (51 DAP).

Figure 7 .
Figure 7. Relation between estimating productivity and productivity in the field (51 DAP).
of several Sentinel-2 bands for infected areas.The increase of pixel values in band 2, 6, 12, and 8A influences the SDM output which lead to the identification of brown leafhopper outbreak.

Table 2 .
Difference between estimated productivity and field productivity.