Portal de Periódicos da CAPES

Detection of satellite data-based flood-prone areas using logistic regression in the central part of Java Island

2019; IOP Publishing; Volume: 1367; Issue: 1 Linguagem: Inglês

10.1088/1742-6596/1367/1/012086

1742-6596

G Pratidina, Suroso Suroso, Purwanto Bekti Santoso,

Water and Land Management

Abstract The history of natural disasters recorded in BNPB (2019) explains that the total number of natural disaster events in the central part of Java (Central Java Province and Special Region of Yogyakarta Province) ranks highest in terms of the number of frequency of occurrences nationally. Of the total natural disasters that have occurred in Central Java, the number of floods is ranked third after the landslide and tornado disaster, which is around 1500 disasters. Various factors that can cause flooding cannot be eliminated. However, what is more, necessary is how to control the impacts caused by floods so that they can be managed and monitored appropriately. One effort to overcome the problem of the threat of flooding is to develop a detection model for flood-prone areas. In this study, the detection of flood-prone areas was carried out by using a logistic regression method that takes into account the variables that cause flooding such as elevation, land slope, river distance, flow accumulation, rainfall, and runoff coefficients. The results of the modelling, obtained coefficients of the variables/parameters mentioned earlier, namely intercept (5.05766 – 16.13210), rainfall (-0.01547 – 0.04075), elevation (-0.02173 – -0.00592), slope (-0.28108 – -0.01940), runoff coefficient (-9.10476 – 7.15039), river distance (0.00038 – 0.00783), and flow accumulation (-9.26342E-06 – 0.00309). The level of success in this modelling testing was 93.47826% -98.26087% of 329 flood event data points and not floods.