The Spatial Planning in Purworejo Regency-based Flood Hazard using Sentinel-1A Imagery

Flood disasters in Indonesia increased by 23% over a period of 70 years. In Central Java Province, the incidence of floods has also experienced a significant increase, including flooding in Purworejo Regency. Floods in Purworejo Regency occur in almost every rainy season in the same location. In 2020 and 2021, floods hit the same 13 villages. The utilization of the Sentinel-1A Imagery technology can quickly and accurately visualize the distribution of locations where there are indications of flood inundation. This research aims to do rapid flood mapping for spatial planning. The method used in this research is the processing of Sentinel-1A imagery, flood mapping in Purworejo Regency, validation test of processed Saentinel-1A imagery, and flood analysis with the Purworejo Regency Spatial Planning. The Sentinel-1A data processing is in the form of a combination process between Composite RGB and the Change Detection and Thresholding (CDAT) method. Sentinel-1A imagery was taken at two different times, namely when there was a flood and when there was no flood. The Flood Map Processed by Saentinel-1A imagery will be verified and validated. The validity test results using the Kappa coefficient are 0.841, and the accuracy value is 94%. The Kappa coefficient value and accuracy show that the flooding results from the Saentinel-1A image processing are valid and accurate. The results of the analysis between spatial planning and flood inundation in Purworejo Regency are > 50% of the flooded areas in the form of paddy fields and plantations.


Introduction
Flood disasters are among the highest natural disasters in Indonesia during these ten years.The total number of victims of floods was recorded to be the highest, namely 60% of the total disaster victims [1].Central Java Province is listed as the third province in Indonesia where floods frequently occur [2,3].Central Java Province ranks first in the number of victims of flood disasters in Indonesia [4][5][6].Purworejo Regency is one of the districts with the most flood disasters in Central Java Province [4].Floods in Purworejo Regency have continued to increase since 2019 -2021.
On March 5, 2020, floods hit two sub-districts in Purworejo Regency.The highest water level reaches 60-70 cm.The flood occurred due to heavy rain for more than 24 hours.Over 13 villages were inundated after the flood [7,8].One year later, on January 12, 2021, floods hit Purworejo Regency again.The flood inundation depth in 2021 is more severe than in 2000, which is 2 meters deep [9].In the same year, the Purworejo Regency Regional Disaster Management Agency (BPBD) conducted data collection on the affected areas along with losses and some other additional information.The data 1264 (2023) 012027 IOP Publishing doi:10.1088/1755-1315/1264/1/012027 2 collection results are in the form of tabular data and reports.Spatial data on the area of inundation has yet to be published, so one of the mitigation efforts from the spatial planning aspect cannot be carried out.
The flood mapping currently carried out uses various approaches and methods.BNPB [10] used a DEM data approach to study river morphology and flood events.Other research [11] maps floods using inundation data, field observations, and interviews with the community.Interviews were conducted to identify the frequency, height of inundation, and patterns of adaptation to flooding.The use of remote sensing methods, especially SAR data, for flood mapping has been carried out by many researchers [12].Use of SAR data, in this research using Sentinal-1A Imagery, is used for flood mapping to complement BPBD data, which is still in the form of tabular data.The use of Sentinal-1A Imagery in this research, apart from the relatively good level of accuracy, is also due to the availability, ease of access, and free nature of the data.
The mandate of the spatial planning law states that spatial planning must be able to deal with disasters [13].The spatial planning products are still not fully capable of being used as a tool for disaster management, as the results of research [14,15] in the City of Semarang explain that the spatial planning of the City of Semarang in 2011 -2031 has not been able to cope with flood disasters.The limited spatial data on floods is one factor affecting the quality of spatial planning products.Spatial planning is a form of natural disaster mitigation, especially floods [16][17][18].So, the availability of spatial flood data is needed to prepare spatial plans.The use of Senitnel-1A imagery can be used for rapid flood mapping in planning areas as preliminary data.
The fast and accurate flood inundation mapping has a positive impact on disaster mitigation efforts and the preparation of spatial plans.The availability of Sentinel-1A Satellite Imagery, processing methods, and processing software is the right choice to provide inundation maps.So, this research aims to use Sentinel-1A Imagery for fast and accurate flood mapping in Purworejo Regency.The results of the rapid flood mapping can be used for detailed regional planning, evaluating Purworejo Regency spatial plans and flood disaster mitigation plans in Purworejo Regency.

Study Area
The location of this research study is the Bogowonto Watershed in Purworejo Regency (see Figure 2).

Materials and Methods
This study uses Sentinel-1A satellite image data, land use maps, and spatial plans for Purworejo Regency.Sentinel-1A Satellite Image is a radar image issued by the European Space Agency (ESA).This image can map flooding because it can penetrate clouds, fog, and rainfall to be used in bad weather conditions [19].Sentinel-1A Satellite Imagery has a high spatial and temporal resolution to map floods in more detail and accurately.Utilization of Sentinel-1A satellite imagery to map flooding using the difference in water level before and after the flood occurs [20,21].In addition, Sentinel-1A satellite imagery can monitor flood movements and identify areas isolated due to flooding.The Sentinel-1A satellite imagery used is March 14, 2022, shooting as wet data, and March 2, 2022, as dry data.The two data were selected based on the history of flood events obtained from BPBD.The research process is explained in Figure 3.
The method used in this study is divided into three stages, namely Sentinel-1A image processing, validation tests, and overlays with land use maps and spatial plans.Sentinel-1A Image Processing uses RGB composite techniques and Change Detection and Thresholding (CDAT).The RGB composite technique analyses the Sentinel-1A Image by stack terrain corrected image analysis.The analysis was carried out by overlaying the Sentinel-1A image when it was dry with the Sentinel-1A image when it was dry.The results of the composite band will show the distribution of flood inundation locations marked with a reddish colour in the image [22,23] (see Figure 4).Change Detection and Thresholding (CDAT) is a method for detecting changes in images or images.This method consists of two stages, namely the change detection stage and the threshold determination stage.The change detection stage compares two images taken at different times to detect differences or changes.The threshold determination stage is used to classify pixels that change or not [24].CDAT begins by taking two images or images at different times.Usually, the first image is taken initially before the change, and the second image is taken after the change.The second stage is the threshold determination stage, carried out by determining the threshold value used to classify pixels that experience changes.This threshold value can be set automatically or manually.Generally, the threshold value is determined by finding the optimal value or the one that gives the best detection results.After the threshold value is determined, the image's pixels will be labelled or a different class from those that have not changed [25].In the end, the change detection results can be visualized by producing a binary image or class image that shows the location of changes that have occurred in the image (See Figure 4).The second stage is the validation test.This stage was carried out to test the results of Sentinel-1A Image processing under field conditions.Locations indicated by flood inundation were directly surveyed.The survey results were compiled, and the suitability was calculated with the Sentinel-1A image processing results (see Figure 5).

The Processing of Sentinel-1A Imagery to Determine Flood Inundation in Purworejo Regency
Sentinel-1A Image Processing uses Composite RGB and Change Detection and Thresholding (CDAT) methods.To identify flood inundation areas using the RGB composite method by stacking.The stacking process produces overlapping images.The flood inundation area from this process uses composite RGB bands from the stacking process.The red band (R) uses polarization data when there is no flood, while the green (G) and blue (B) bands use polarization data when there is a flood.Areas that are inundated are drawn from bright red to deep red.The darker the red color produced; is assumed that the more severe the flood will occur at that location.
The use of the CDAT method to identify flood inundation is to compare changes in pixel values during floods and when floods do not occur.The CDAT method on Sentinel-1 imagery allows the identification of flooded areas based on the change in backscatter intensity between the two image capture times.The combination of RGB and CDAT composite methods produces a flood inundation area.The use of RGB and CDAT composite methods can increase the accuracy of the flood inundation area results.The flood inundation map resulting from the processing of the two methods is shown in Figure 6.

Flood Area RGB Composite results
Flood Area CDAT results Flood Inundation Map Legend:

Flood Inundation Validation and Accuracy Test
The processed flood inundation map in sub 3.1 is then tested for validation.A validation test was carried out to assess the accuracy of the Sentinel-1A image processing process.The validation test was carried out using observation and interviews with residents.The observation points are spread evenly throughout the flood inundation area (see Figure 7) and an example of the results of the observations is in Table 1.Data from field observations and interviews are calculated to obtain accuracy and validation of flood inundation processing results.Calculation of validation and accuracy is done by standard Kappa coefficient and confusion matrix.The use of the Kappa coefficient standard aims to identify the level of agreement between processing results and field results [26,27].The provisions used state that if the Kappa value is close to 1, then the processing results are declared valid.The results of validation processing stated that the flood inundation map in the Bogowonto Watershed, Purworejo Regency, was valid.The Kappa efficiency value obtained is 0.841 (see Table 2).The probability value (Approximate Significance) is 0.000 and less than 0.05.This value states that the Kappa value is significant.Calculation of the accuracy of flood inundation processing results is done by comparing the processed flood inundation with a reference standard.The standard reference used is field/primary flood event data.The level of accuracy is the result of a comparison between the amount of accurate data and the total observed data.Calculation of the level of accuracy is described in Table 3.The total overall accuracy value is 0.94 or 94% with an accuracy level of processing results of 92%.These two values explain that the results of the flood inundation map processed by Image Sentinel-1A in the Bogowonto Watershed, Purworejo Regency, are accurate.Based on the results of validation and accuracy calculations, it can be concluded that the flood inundation map is correct.

Analysis of Flood Inundation with Land Use and Spatial Planning in Purworejo Regency
The aspect of flood disasters is very important to be used in the spatial planning process.As explained in the background section spatial planning must be a tool for disaster mitigation.The results of the flood inundation mapping from Sentinel-1A imagery will be overlaid with land use maps and spatial plans.This process is carried out to apply and evaluate the current land use and future spatial plans.
The overlay between the flood inundation map and the existing land use will produce areas that are prone to flooding.In addition, the results of the overlay produce the type and area of inundated land use.The results of the overlay between the flood inundation map and the existing land use are shown in Figure 8.  4. The next analysis is the overlay of the flood inundation map with the Purworejo District spatial plan.The purpose of this analysis is to identify spatial plans with flood threats.The Overlay results can be used as an initial reference for planning, especially disaster mitigation.The results of the overlay of the flood inundation map with the spatial plan for Purworejo Regency are shown in Figure 9.

Conclusion
Existing land use and land use plans that are inundated by flooding in Purworejo Regency are food agricultural land areas.The form of flood control for food-agriculture land areas is different from residential areas or other built-up areas.The extent of the spatial plan that was inundated due to the flood disaster in Purworejo Regency indicates that the consideration of spatial flood data has not been properly applied.So, the use of Sentinel-1A Imagery for mapping flood-prone areas is very useful in the spatial planning process.

Acknowledgments
The author would like to thank Mr. Dr. Anang Wahyu Sejati, ST., MT, and Mrs. Mitha Asyita Rahmawaty, ST., MT for their ideas and discussions.This paper is the result of a collaboration between UNDIP vocational college research funding in 2021 and the author's independent funding.

Figure 2 .
Figure 2. The Study Area

Figure 5 .
Figure 5. Process of taking points in the field using Avenza Map

Figure 6 .
Figure 6.Flood Inundation Map as a result of Composite Processing of RGB and CDAT

7 Figure 7 .
Figure 7. Observation Locations for Validity Test

Figure 8 .
Figure 8. Map of Flood Inundation Analysis with Existing Land Use in Purworejo Regency Most of the flood inundation locations are located on the banks of the Bogowonto River and agricultural land is located at the mouth of the river.The type and extent of the use of the existing inundated land are shown in Table4.

Figure 9 .
Figure 9. Map of Flood Inundation Analysis with Spatial Planning in Purworejo Regency Types and extent of spatial plans that are threatened by floods Most of them are areas of food crops, plantations, and production forests.In detail shown in table 5 and figure 10.

Figure 10 .
Figure 10.Percentage of area of land use plan that is flooded.

Table 1 .
Example of Field Observation to Test Validity

Table 3
Calculation of Processing and Field Data Accuracy

Table 4 .
Flooded land use area

Table 5
Types and Areas of Flooded Land Use Plans