Agricultural Droughts Monitoring of Aceh Besar Regency Rice Production Center, Aceh, Indonesia – Application Vegetation Conditions Index using Sentinel-2 Image Data

Monitoring the agricultural drought of paddy rice fields is a crucial aspect of preparing for proper action in main - taining food security in Indonesia. The Aceh Province is one of Indonesia’s national rice production centers, especially Aceh Besar Regency; it includes three central districts; Indrapuri, Kuta Cot Glie, and Seulimeum. Sat-ellite-Sentinel 2A data have been tested to monitor the drought levels of around 2,803 Ha in the three districts in this study. This study aimed to determine the drought level in Indrapuri, Kuta Cot Glie, and Seulimeum districts, Aceh Besar Regency’s paddy rice fields using Sentinel-2A data imagery. The vegetation conditions index (VCI) of Sentinel-2 data was utilized to identify a vegetative drought level in the area for the 2018, 2019, 2020, 2021, and 2022 growing seasons. The vegetation inertia index is derived from the Normalized Difference Vegetation Index (NDVI). The results show that the VCI looked volatile, but the trendline increased by four percent, from 92.56 in July 2019 to 96.08 in July 2021. Most areas on the dates investigated found that the no drought category was still dominant. The designated data analyzed found that the June 2022 data tend to be distributed to the drought in extreme, severe, moderate, and mild increases compared to the previous data investigated. This figure shows an increasing drought in the study area, and the average drought index is in the category of mild drought. In addition, there has been a trendline decline in the value of NDVI in recent years, causing agricultural land for paddy rice fields to be slightly vulnerable to drought.


INTRODUCTION
Paddy rice production is a crucial aspect of food security in Indonesia. One of the treating factors that may affect production is agricultural drought in some paddy rice production centers. As the primary providers of the worldwide food supply chain, the agriculture practices of paddy rice In Indonesia are vital in safeguarding food security global (Accorsi and Manzini, 2019). Meanwhile, the spread of the COVID-19 pandemic has forced significant tasks to food production and consumption patterns, highlighting the vital need for reconsidering the sustainability of present approaches to agriculture and the food industry. One of the problems for maintaining food security is avoiding the drought on agricultural land during the growing season for paddy rice fields. The causes of droughts on agricultural land can be meteorological, agricultural, hydrological, and socioeconomic droughts (Wilhite, 2005). However, in recent years meteorological drought has been dominant in hampering agriculture practices (Bageshree et al., 2022;Guo et al., 2019).
Drought is a significant natural hazard, especially in agricultural practices for paddy-rice fields, and can cause significant losses to crop production and water supply (Surmaini et al., 2015). In recent years, droughts have been frequent due to the El Nino phenomenon and have had a tremendous impact on agricultural land (Syahrial et al., 2017). This phenomenon has further triggered several threats to Indonesian agriculture practices for paddy rice fields, especially in Indrapuri, Kuta Cot Glie, and Seulimeum Districts, Aceh Besar Regency, Aceh Province, Indonesia. These areas are among national rice production centers.
Identifying drought levels is essential for evaluating the likelihood of drought and its severity for national food security awareness (Gitz et al., 2016). The impact of drought can be well mitigated if detected earlier; this can be done through remote sensing data measures and monitoring regularly. However, monitoring regular based field visits is time-consuming and costly. Using remote sensing data, drought can be effectively monitored over large areas. Satellite-borne remote sensing data provide a synoptical interpretation of the Earth's surface and can therefore be used to monitor drought events spatially (Kogan, 1997(Kogan, , 1995Liu and Kogan, 1996;Sholihah et al., 2016). Several remote sensing indices have been developed and applied to monitor the drought spatially (Huang et al., 2018). Among the indices, the NDVI as a vegetation health check has become one of the most commonly used for monitoring drought events (Akinyemi, 2021;Walter-shea, 2002). To improve the approach, combining vegetation index and temperature is recommended. Combined NDVI and ground surface temperature (ESG) provide a strong correlation and valuable information for identifying agricultural drought as an early warning system (Lee et al., 2021;Olii et al., 2022). Some vegetation indices for drought monitoring using remote sensing data have been introduced. One of them, vegetation health index (VHI) has demonstrated outstanding capabilities (Sholihah et al., 2016). The result has shown better relevance in detecting drought (Akinyemi, 2021;Walter-shea, 2002 Drought monitoring during the dry season is essential to see the possible adjustment for the growing season during this time. The dry season of paddy rice planting season is at the end raining season which ranges between April-September. Thus, June and July are critical months for paddy rice growing season in many parts of paddy rice growing in Indonesia rice production centre, including in Indrapuri, Kuta Cot Glie, and Seuleumun districts of Aceh Province Indonesia. In 2021, more than 600 hectares of paddy rice fields were experiencing drought in Kuta Cot Glie district (Indonesia, 2021). The harvest in the regency failed over an average area of 800 hectares. This data provides that drought is a serious matter. During this time of planting for the dry season, there were likely risks against the drought and a high risk of crop harvesting failures.
Obtaining data directly during the dry season is challenging; however, remote sensing data can be utilized to obtain data even though the harvest time has been done. One of the data freely available for evaluating drought is Sentinel-2 images data. Sentinel-2A images are considered in this paper because of the open access policy, good resolution, availability in a short period, and relatively good spatial range for drought monitoring. Ensuring the utilization of satellite image data can be used for drought monitoring for agriculture land, this study aimed to determine the drought level in the paddy rice fields of Indrapuri, Kuta Cot Glie, and Seulimeum Districts, Aceh Besar Regency, Aceh, Indonesia, by utilizing remote sensing data of Sentinel-2A to assess the drought in the regions by applying derived vegetation indices. In addition, a comparison was made between long-term droughts in 2018, 2019, 2020, 2021, and 2022 to determine drought-level conditions in the three districts as the rice production center in Aceh Besar Regency. As droughts are an unpredictable event in agricultural practices, and are recurrent as natural phenomena due to a deficiency in precipitation, and also water storage during the growing season, providing the information with rapid analysis using remote sensing data is an important aspect for continuing an area as production centre. This study tried to provide the information needed by both the government and farmers in a timely manner in making decisions in paddy rice planting activities.

Study area
The Indrapuri, Kuta Cot Glie, and Seulimeum districts of Aceh Besar Regency were selected as research areas (Figure 1). These areas are located in the northern part of Aceh Besar Regency, Aceh-Indonesia Province, which is geographically located between 5°10'-5°40'N and 95°20'-95°45'E. Administratively, the three districts bordering the Strait of Malacca to the North, the Seulawah Mountains to the East, the Bukit Barisan Mountains on the Southern border, and Banda Aceh City are located to the west. The landform in the study area is dominated by flat plains between 0-50 m above sea level. Almost 50 percent of the land use in the study area is paddy rice fields, about 184,486 ha of the total 397,104 ha in the three districts. These areas are one of the centers of national rice production in Indonesia.

Data
The Sentinel-2A multispectral imager includes 13 spectral bands with a plot width of 290 km and a spatial resolution of 10 m (three visible and near-infrared bands), 20 m (6 red edge/shortwave infrared bands), and 60 m (3 red edge/shortwave infrared bands) were utilized. The objective of this study is intended to monitor the variability of ground surface conditions, its wide plot width, and high re-visit times (10 days with one satellite and five days in an entire constellation with the twin satellite Sentinel-2B) to support the monitoring of vegetation changes during the growing season. It also provides data and applications for operational land monitoring, emergency response, and security services.  Seulimeum districts paddy rice field, Aceh Besar-Indonesia Regency. Scenes S2A of 100% cloudfree were selected and downloaded from the Copernicus Sentinels Scientific Data Center (https:// scihub.copernicus.eu/dhus/). Orthorectified and radiometrically corrected images (level 1C processing) were used for the image analysis. Since the study area represents only a small part of the traces of each scene, it was assumed that constant atmospheric conditions and no atmospheric corrections were applied. Each scene contains 13 spectral bands with a native spatial resolution of 10 m (blue, green, red, and near-infrared bands), 20 m (red edge bands), or 60 m (shortwave infrared bands); all tape is re-sampled to a resolution of 10 m for further processing.

Drought-sensitive spectral indices
The available Sentinel-2 data is still at level 1C. To perform vegetation index calculations, it is required to use imagery that has been corrected by the atmosphere at least until surface reflection (surface reflectance) to avoid atmospheric effects. Therefore, in this study, atmospheric corrections were curried out to Sentinel-2 images using Sen The study used the Sentinel-2 imagery data provided by the European Space Agency (ESA), which covered the entire dry season in the study area. The Sentinel-2 satellite captures Earth's surface data with 13 spectral bands and three spatial resolutions of 10 m, 20 m, and 60 m. This is the processed data and only needs to be converted to reflectance values before calculating the drought index. With time-series data, a classification approach using Red and NIR bands of Sentinel-2 scenes has been used to view drought index results calculated from remote sensing data against drought in rice fields.
Drought was identified from multitemporal sentinel-2A satellite imagery of five different years; July and December 2018, July and December 2019, July and December 2020, July and December 2021, and June 2022. VCI was developed to monitor vegetation conditions (Kogan, 1995) to describe changes in the NDVI value due to weather conditions to monitor the effects of drought on vegetation. The NDVI layer was overlaid to notice differences in vegetation health  (Liu and Kogan, 1996). VCI is based on the relative NDVI change concerning the minimum historical NDVI value. The VCI compares the current Vegetation Index (VI), such as NDVI or Enhanced Vegetation Index (EVI), to the values observed in the same period in previous years within a specific pixel. Thus, the VCI index can be derived from NDVI as in Equation 1. NDVI is calculated by band 4 and band 8 from Sentinel data.
where: VCI j -the value of the vegetation condition index for the j-date; NDVI j -the NDVI value for the j-date; NDVI max and NDVI min -the maximum and minimum NDVI values in the data set.
This study proposed the following classification scheme for drought monitoring (Table 1).

RESULTS AND DISCUSSIONS NDVI value distribution
Drought patterns in the three districts as amongst the paddy rice production center of Aceh Besar regency over the period 2018-2022 were analyzed using the NDVI approach. The NDVI value is an index value that can be used as an indication of the vigor level of the plant during the growing season. It is acknowledged that Indonesia experiences two growing seasons, wet and dry seasons. The wet season starts in October and ends in March, with the peak season in December/January. The dry season starts in April and ends in September, with the peak growing season in June/July. Thus, NDVI value distribution may refl ect the health or drought experienced at the study site. The diversity of NDVI shows spatial variations in vegetation health in the study area and the dynamic values of NDVI annually, as shown in Figure 2. The fi gure depicts the spatial distribution of NDVI value of multitemporal data from July 2018 until 2022. July is considered a dry growing season. It shows there is a shift of NDVI value for diff erent years investigated, the 2019 data of NDVI with high value concentrated  in the north-west of high value NDVI. Similarly, the 2021 data also show the same pattern of high NDVI value.
Even though there is a low correlation between the data observed between 2018 and 2021 ( Figure 3) as considered dry season, the vegetation health index can be distributed spatially in the study area from the lowest on July 2021 with the value NVDI of approximate 0.67 and the highest NDVI value in July 2018 of approximately 0.95. This fi gure indicates that in July data, the health of vegetation index is still high, confi rming that healthy paddy rice conditions still dominate the area. A similar result has been demonstrated by a study on the phenology of paddy rice in the area (Sugianto et al., 2020) and also similar studies in another part of Indonesia during the dry and wet season using the enhanced vegetation index (EVI), normalized difference vegetation index (NDVI), land surface water index (LSWI) and normalized diff erence water index (NDWI) of MODIS data series (Sari et al., 2010), showing variation growth stage during the dry season. Figure 4 shows the spatial distribution of NDVI value of multitemporal data from December 2018 until 2022. December is considered a wet growing season. Figure 4 refl ects the spatial distribution of NDVI values for diff erent years from 2018 until 2022.
Similarly to the July data, there is a low correlation between the data observed between 2018 and 2021 ( Figure 5). As the wet season is considered, the vegetation health index can be distributed spatially in the study area from the lowest on December 2020 with an NVDI of approximately 0.64 and the highest NDVI value in December 2019, approximately 1.0. This fi gure indicates that the shift of VCI in December data to observe the health of vegetation is high, confi rming that healthy paddy rice conditions dominated the area. Figure 5 shows that the NDVI value decreases during the observation period. Without a rice planting calendar at the test site, drought monitoring using NDVI alone as a critical parameter is not enough to evaluate the drought levels on a regional scale.

VCI value index distribution
The VCI value was obtained from the sentinal-2 imagery analysis of 5-multitemporal images data. It is acknowledged from the previous study that dry soils are usually visualized with low VCI. The 5-multitemporal images of the paddy rice fi eld of the study area are depicted in Figure 6. Figure 6 shows that the VCI increased significantly from July 2019 to July 2021, but decreased slightly in June 2022. In general, dry conditions have high VCI and low NDVI. VCI spatial variations of July and December 2018, July and December 2019, July and December 2020, July and December 2021, and June 2022 from presented in Figures 6 and 7. Figures 6 and 7 demonstrate the spatial distribution of VCI for diff erent years of observation. The 2022 data is dominated by extreme and severe drought. Similarly, for 2018 and 2020, the data are dominated by severe drought. However, for 2019 and 2021, the data show mild and no drought spatially in the study area. As drought is considered a common phenomenon in agriculture, knowing the drought is essential for evaluating the growing season shift to reduce the potential loss of paddy rice production in the study area.  for 360 hectares and 1200 hectares, respectively. However, no drought is still dominant, accounting for an average of 8000 hectares from July to August 2020, then dropping to 6800 in January 2021, then increasing from June till November 2021. In April 2021, it also showed a decreased area with no drought. Table 2 shows the variation of the distribution of total area based on the class draught from extreme to no drought for 2018 until 2022 growing season; dry season (June-July) and wet season in December. Other research in Indonesia found that from 1992 to 2016 drought Interestingly, the designated dates of data analyzed found that June 2022 data tend to be distributed to the drought in the extreme, severe, moderate, and mild increased compared to the previous data investigated. It was assumed that due to climate change, June 2022 will start with extreme weather. Further investigation will be conducted for the incoming year to see the trend of drought occurrence in the area.
The result of this study shows that during the dry season little or no extreme drought occurred in the paddy rice production center in the region. However, this finding may contribute to an early warning system of drought as part of the food security initiative (Shams Esfandabadi et al., 2022). Even though using Sentinel-2 data, which has 10-m bands, likely causes a severe salt-andpepper effect (Xiao et al., 2021), this study has shown the promising result of vulnerability assessments of paddy rice field assessment of agricultural drought. Vulnerability assessment in the region for paddy rice fields is a crucial component of drought risk assessment, as they support the design of mitigation actions to target sectors or more sensitive populations (Wilhite, 2005). Drought is one of the causes of loss of production in agriculture practice (Ray et al., 2018). Moreover, drought occurrence may be avoided if potential drought data are provided an early warning.

CONCLUSIONS
This study has demonstrated the extent of Spatio-temporal analysis of drought for agriculture for paddy rice fields in five years of the growing season in the Indrapuri, Kuta Cot Glie, and Seulimeum Districts, Aceh Besar Regency by utilizing satellite-based remote sensing data. Severe drought was shown in November 2021, which then increased in terms of the total area to moderate drought and mild drought, accounting for the highest in November for 360 hectares and 1200 hectares, respectively. No drought category was still dominant from July to August 2020, dropping in January 2021, then increasing from June to November 2021. In April 2021, it also showed a decreased area with no drought category.
The distribution of drought from 2018 data until 2022 data indicated that only a tiny area in June 2022 had an extreme drought; the rest of the data investigated no extreme drought. Almost all dates investigated in the study area experienced moderate drought, with the highest in June 2022. The rest of the areas are experiencing mild drought. Most of the areas on the dates investigated found that no drought was still dominant. Interestingly, the designated data analyzed found that the June 2022 data tend to be distributed to the drought in extreme, severe, moderate, and mild increases compared to the previous data investigated. The index used in this study has successfully proven to identify the degree of the agriculture drought in the paddy rice production centers. This result can explain the drought severity in the study area by analyzing suitable vegetation. The results of VCI estimates can contribute to monitoring the occurrence of agricultural drought as an early warning system.