The relation between vegetation indices of the generative stage with productivity of paddy fields

Maros regency is one of the agricultural areas in South Sulawesi which is a large rice area. Sourced from BPS data (Central Statistics Agency), both rice production and productivity in Maros Regency have decreased fluctuatingly over the last three years. Knowledge related to rice productivity is very important to know in order to support food security. Geographic information system is one of the modern technologies that can be used in estimating rice productivity which then utilizes Sentinel-2 Image data. In its use, a spectral transformation is used to display aspects of plant conditions such as vegetation index NDVI (Normalized Difference Vegettation Index), EVI (Enchanced Vegetation Index) and SAVI (Soil Adjusted Vegetation Index). The purpose of this study was to determine the relation between the vegetation index value of the generative stage and the productivity of paddy fields in Alatengae village. The method used is a simple linear regression method in estimating the relation between rice productivity and the vegetation index NDVI, EVI and SAVI. From the analysis of the relation between the vegetation index and rice productivity, there is a very strong (positive) correlation, where the higher the value of the vegetation index, it will be higher the plant productivity.


Introduction
Indonesia is one of the countries where the majority of people use rice as a staple food.Rice comes from rice (Oriza sativa L.) which is an important type of cereal where almost 50% of the world's population uses it as a daily staple food.According to IRRI (2015), rice is estimated to grow and spread around 115 million hectares in all parts of the world [1].Sourced from BPS data (2019), both rice production and productivity in South Sulawesi fluctuated which can be seen from the total rice productivity in 2019 of 50.03 Ha with a total production of 5,054,167 Tons, decreasing to 48.23 Ha with total production of 4,708,465 Tons in 2020, while in 2021 it has increased by 51.95 Ha with total production of 5,152,871 Tons [2].
Maros Regency is one of the agricultural areas in South Sulawesi where this area is a large rice barn.Looking at data from BPS Maros Regency (2020), in 2020 rice production decreased by 20,229.53Tons, for rice harvest area it decreased by 1.11% or 482.25 Ha compared to 2019, then for rice productivity in 2020 has decreased by 8.85% compared to 2019 [3].Looking at these changes, it can be said that the data, especially food crops, namely rice in the Maros area, has experienced a large decline from various aspects.
Knowledge related to rice productivity in a region is very important to know for the central government as a policy maker in supporting food security and farmers.To support this, the use of modern technology in the form of remote sensing can be used as a way to support rice productivity 1230 (2023) 012145 IOP Publishing doi:10.1088/1755-1315/1230/1/012145 2 estimation in addition to using conventional methods.Geographic Information System (GIS), considered as a suitable remote sensing technology is also effective because it can save costs and can save time in finding information.One of the modern remote sensing technologies that can be used in agriculture is using Sentinel 2 satellite image data.
To determine the phase of rice plants requires a form of spectral transformation to display aspects of plant conditions such as density or other aspects and in this case can be detected using the vegetation index.There are several vegetation indices that are more widely used in estimating the growth phases of rice or rice productivity, namely the NDVI vegetation index (Normalized Difference Vegetation Index) and EVI (Enhanced Vegetation Index) which is a modified development of NDVI.Then the SAVI method (Soil-Adjusted Vegetation Index) which is a modification of the NDVI method by reducing the effect of soil background variations using the soil adjustment factor (L).The value of the results of this transformation of the vegetation index will be used as the basis for calculating a simple linear analysis between the relation between rice productivity and the vegetation index at the generative stage.
Based on the description above, this research was conducted to determine the relation between the generative stage vegetation index and the productivity of rice fields in Maros Regency

Objective of the research
The purpose of this research is knowing the relation between the value of the Vegetation Index of The Generative Stage with Productivity of Paddy Fields in Alatengae and to estimate the productivity of rice plant.
The benefit is to Provide an information about the right vegetation index for predict the productivity of paddy fields in the village of Alatengae.

Research methodology 3.1. Time and place
This research was carried out from October to December 2021, following the period of Sentinel-2A satellite imagery orbiting the Aletengngae Village Area, Bantimurung District, Maros Regency.As for the research location map it can be seen in the Figure1.

Tools and materials
The tools used in this research was a laptop which contained software QGIS and stationery.The materials used in this research are primary data such as data from interviews with farmers, field survey data, and field observation documentation while for secondary data in the form of Sentinel-2A Imagery for Maros Regency, acquired from October to November which can be accessed on the website.https://earthexplorer.usgs.gov/,and shp data map of paddy fields.

Research procedure
The research procedure for the relation between vegetation index of the generative stage with a productivity of paddy fields consists of three main stages, namely the preparation stage, the field data collection stage, and the data analysis stage.The stages of the research will be presented in a flow chart by looking at the stages below.

Preparation stage.
This stage consists of a literature study which begins to be carried out browsing for research related literature searches.

Data clumping stage.
In this stage it is carried out to see the condition of the land and the condition of the plants such as land area, age of rice plants and other data.

3.3.2.1.
Data pre-processing stage.This stage begins by downloading Sentinel 2A image data with the acquisition time of October to November on the link https://earthexplorer.usgs.gov/.All of this data was taken in 2021 where the rice planting period is in the generative phase, which is between 46-66 days after the rice is planted.Then make an atmospheric correction on the Sentinel-2A image to change the radians value to a reflection value.This process serves to improve the results of the accuracy of the vegetation index model.The correction phase is carried out in one of the following waystoolsin QGIS named Semi Automatic Classification Plugin.After the correction process is carried out, the next step is to carry out cropping image data in the QGIS application.

Data processing stage.
This stage begins by processing the vegetation index data from the NDVI and EVI transformation, then making a visualization map of the rice growing phase based on the NDVI, EVI and SAVI transformations.After that, a linear regression model is created to analyze the relation between rice productivity and the generative stage vegetation index.According to Innarossy (2019), there are index value criteria for plants.Can be seen in table 1 is the range of NDVI, EVI and SAVI vegetation index values used [4].

Regression and correlation analysis.
Regression analysis is one of the methods used to calculate the results of independent variables on the dependent variable.In principle, this analysis is an analysis that is used in finding the curve that describes the bond of a data set.The results of the regression analysis are displayed in graphical form.Making graphs is done with the help of software Microsoft Excel 2010.This stage is intended to find out how closely the relation between rice productivity and the generative stage vegetation index.
The regression analysis used in this study is linear regression.This regression is used to determine a linear function (straight line) that is adjusted to a predetermined set of data points (xn,yn).The stage begins with analyzing the relation between the vegetation index and land productivity where the influence of the dependent variable (y) in the form of actual rice productivity data and the independent 1230 (2023) 012145 IOP Publishing doi:10.1088/1755-1315/1230/1/0121454 variable (x) in the form of the vegetation index value is then predicted using the linear regression method.The equation that will be obtained through linear regression is: y= a (x) + b (1) Where: y= bound variables, x= free variable a, a= constant, b= regression coefficient As for the correlation analysis carried out in this study aimed at knowing the strength or weakness of the relation between the two variables.The correlation value is also known as the correlation coefficient which is expressed in the symbol r.The value of r is always located between -1 to 1.When the value of r = +1, it means that there is a perfect positive correlation between variable x and variable y otherwise if r = -1, it means a perfect negative correlation between variable x and variable y, while r = 0 , means there is no correlation between x and y.The closer to 1, the stronger the correlation while the closer to zero the lower the correlation between the two variables.The strength of the correlation value can be seen in table 2 below.2019), it can be seen that the it is included in the high greenery.This value indicates that at that time the dominance of rice plants was in the generative phase range from 0.3515 to 0.8368.2019), it can be seen that this value is included in the high greenery category.This value indicates that at that time the dominance of rice plants was in the generative phase in the range of 0.1528 to 0.6664 [4].2019), it can be seen that this value is included in the high green category.This value indicates that at that time the dominance of rice plants was in the generative phase in the range of 0.1213 to 0.5955 [4].

Processing of the savi vegetation index (soil adjusted vegetation index)
The transformation map of the NDVI, EVI and SAVI vegetation index can be seen in the figure 2, 3 and 4 below.Based on the map resulting from the transformation of the three indices above, it can be said that October 26, 2021 will be the peak of the generative period, this can be seen from the average spectral value and the description of the greenness of the plants shown on the map.This is in accordance with Johan's statement (2017), that the initial generative phase is when the rice plant enters the beginning of 1230 ( 2023  Based on the calculation of research data between the NDVI value and rice productivity above, the productivity value equation is equivalent to 34.462 (NDVI) -17.893 which from this equation has a coefficient of variation value (R2) 0.9017 which means that 90.17% of the NDVI contribution to rice productivity while the rest is another factor that affects rice productivity.Thus, it can be seen that from the results of the correlation relation between the generative stage vegetation index through the NDVI transformation above can be categorized as very strong this is according to the Jumiagra interval table (2019), for correlation interval values (r) with intervals above 0.80 to 1 it identifies the level very strong correlation [6].Based on the calculation of the research data between the EVI value and rice productivity in the field, the productivity value equation is equivalent to 133.07 (EVI) -61.38 which from this equation has a coefficient of variation value (R2) 0.9385 which means that 93.85% of the EVI contribution to rice productivity while the rest are other factors that affect rice productivity.Thus, it can be seen that from the results of the correlation between the generative stage vegetation index and rice productivity through the EVI transformation above it can be categorized as very strong, this is according to the Jumiagra interval table (2019), for correlation interval values (r) with intervals above 0.80 to with 1 identifies a very strong positive relation between the two variables [6].

The relation between the SAVI vegetation index and the productivity of paddy fields.
By correlating the relation between EVI and rice productivity, it will produce an equation in the form of Figure 7 below: Based on the calculation of research data between the SAVI value and rice productivity above, the productivity value equation is equivalent to 88.283 (SAVI) -32.534 which from this equation has a coefficient of variation value (R2) 0.9231 which means that 92.31% of the contribution of SAVI to rice productivity while the rest are other factors that affect rice productivity.Thus it can be seen that from the results of the correlation relation between the generative stage vegetation index through the SAVI transformation above it can be categorized as very strong, this is according to the Jumiagra interval table (2019), for the correlation interval value (r) with an interval of 0.80 to 1 identifies the level of relation very strong correlation [6].
Based on the three graphs above, it can be seen that the value of the coefficient of determination (R2) is highest in the EVI vegetation index, followed by the SAVI vegetation index and finally the NDVI vegetation index.The high R value2the EVI vegetation index is affected because the EVI vegetation index can clarify the appearance of the image area with the addition of a blue band and can also correct the NDVI value due to atmospheric influences.This is in accordance with Mufti's statement (2018), that EVI is included as a development vegetation index from NDVI which is intended as an option in overcoming NDVI deficiencies which uses the blue channel to correct NDVI values that experience deficiencies due to atmospheric aerosols [7].
Then for the SAVI vegetation index which has a coefficient of determination (R2) which is higher than the NDVI, this is affected because the SAVI vegetation index can suppress the influence of the soil background which the NDVI vegetation index cannot so that the resulting value is greater.This is in accordance with the statement of Arindi (2018), which states that SAVI is included as an improvement vegetation index from NDVI which can correct the reflection of light generated from the ground by using the soil background canopy adjustment factor (L) [8].
By looking at the results of the three graphs above, it can be seen that increasing the value of the vegetation index has an effect on increasing total productivity.This is in accordance with Nafi's statement (2017), that under normal conditions the vegetation index value and rice plant productivity have a positive relation, the impact of this positive relation is that if there is an increase in the NDVI value it will also be followed by an increase in the productivity value of the rice plant [9].

Processing rice productivity correlation test result
The results of the rice productivity correlation test were carried out by applying the three regression equations for each vegetation index generated by the rice productivity regression model.The samples used in the model test were 5 samples from several paddy fields.This correlation test yields 90.17% for productivity based on the NDVI index, 93.85% for productivity based on the EVI index and 92.31% for productivity based on the SAVI index.From the regression results, the results of productivity estimation are also obtained.To see the values of the average deviation between actual productivity and the estimated results, it can be seen in the table 6.Based on the deviation value generated, each vegetation index above has varying values where there are values that are less and more than the values in the field.The deviation value indicates how far the data value deviates from the estimated productivity and actual data, where if the deviation value is smaller then the value can be said to be good or in accordance with what is in the field, and vice versa.This is in accordance with the statement of Lubis (2019), which states that the smaller the resulting deviation value, the more uniform the data, in this case the data meant by uniform is between the data in the field and the estimated data [10].
There are several factors that can cause the estimated value of the productivity measurement results to be different from the actual conditions, namely factors of different plant conditions such as the level of plant health or can be caused by cloud disturbances at the time of image taking.This is in accordance with Nafi's statement (2017), that the estimated value more or less with that in the field can be influenced by differences in plant age, health level, differences in plant types planted, planting methods,or interference (clouds) during the process of taking satellite imagery [9].

Conclusion
Based on the results of the research obtained, it can be concluded as follows: 1.The results of the regression analysis of the relation between the vegetation index and rice productivity have a very strong (positive) correlation, where the higher the value of the vegetation index will be followed by an increase in crop productivity.2. The best result of processing the vegetation index between NDVI, EVI and SAVI in predicting rice productivity in the generative phase is the EVI vegetation index.3. The results of the correlation of rice productivity yield a correlation of 90.17% for the productivity of the NDVI vegetation index, 93.85% for the productivity of the EVI vegetation index and 92.31% for the productivity of the SAVI vegetation index.4. The average deviation of the estimated productivity with field conditions ranges from 0.36 Ton/ha or 5.02% for the NDVI vegetation index, 0.24 Ton/ha or 3.63% for the EVI vegetation index and 0.28 Ton /ha or 4.19% for the SAVI vegetation index

Figure 1 .
Figure 1.Map of paddy fields in Alatengae village.
where there is an increase in plant photosynthesis which causes the value of the vegetation index to increase[5].4.2.Correlation test of the relation between the vegetation index and the productivity of paddy.fields.4.2.1.The relation between the NDVI vegetation index and the productivity of paddy fields.By correlating the relation between NDVI and rice productivity, it will produce an equation in the form of Figure5below:

Figure 5 .
Figure 5.The relation between the NDVI transformation vegetation index with productivity of the paddy fields in 56 days after planting.

Figure 6 .
Figure 6.The relation between the EVI transformation vegetation index with productivity of the paddy fields in 56 days after planting.

Figure 7 .
Figure 7.The relation between the SAVI transformation vegetation index with productivity of the paddy fields in 56 days after planting.

Table 1 .
Criteria for the value of the vegetation index NDVI, EVI and SAVI.
4. Results and discussion4.1.Results of vegetation index processingThe following are table shows the results of the processing of the NDVI, EVI and SAVI vegetation indices using sentinel 2A imagery on 16 and 26 October and 5 November 2021 to be precise in Alatengae Village, Maros Regency.4.1.1.Processing of the NDVI vegetation index (normalized difference vegetation index).

Table 3
above shows the minimum, maximum, average and standard deviation values of the processed spectral values of the NDVI vegetation index for sentinel 2A imagery.It can be seen that the image calculation results for the NDVI vegetation index on October 26 2021 with a plant age of 56 HST have the highest average spectral value among the other images, namely 0.7031, thus based on the Innarossy criteria table (

Table 4
above shows the minimum, maximum, average and standard deviation values of the processed spectral values of the EVI vegetation index on sentinel 2A imagery.It can be seen that the image calculation results for the EVI vegetation index are almost the same as the NDVI where in this case the highest average spectral value is on October 26 2021 with a plant age of 56 HST, has the highest average spectral value among the images another, namely 0.4959, thus based on the Inarossy criteria table (

Table 5
above shows the minimum, maximum, average and standard deviation values of the processed spectral values of the SAVI vegetation index on sentinel 2A imagery.It can be seen that the image calculation results for the SAVI vegetation index are the same as the transformation of the NDVI and EVI vegetation indices where in this case the highest average spectral value was on October 26 2021 with a plant age of 56 HST which is around 0.4629, thus based on the Inarossy criteria table(

Table 6 .
The mean deviation between the actual rice productivity and the estimation result of paddy fields.