Assessment of flooding risk in Lima, Peru, through change detection based on ERS-1/2 and Sentinel-1 time series

Catastrophic floods, that happened in Lima in 1997–1998 and 2017–2018 (years selected for this study), caused hundreds of fatalities and significant economic loss. To test the hypothesis that information mined from satellite synthetic aperture radar (SAR) images can provide valuable inputs into the common workflow of flooding hazard assessment, the complete archives collected over the Rímac River basin by the European Space Agency’s ERS-1/2 missions and the European Commission’s Copernicus Sentinel-1 constellation were screened. SAR backscatter color composites and ratio maps were created to identify change patterns in the study years. A total of 197 changes (32.10 km2) due to flooding-related backscatter variations and 212 (26.40 km2) due to anthropogenic processes were highlighted. The areas inundated during the flooding events in the study years mostly concentrate along the riverbanks and plain, where gentle slope (≤5°), and the presence of alluvial deposits, also indicate greater susceptibility to flooding. Through geospatial integration with ancillary data (topography, geology, urban footprint, etc.), a risk classification map of Lima was produced. The map highlights the sectors of potential concern along the Rímac River, should flooding events of equal severity as those captured by SAR images occur in the future.


34
In recent years, Lima, the Peruvian capital, has experienced severe and catastrophic floods [1].

35
These events became more frequent, especially in the coastal area of the Peruvian mainland, as a 36 consequence of El Niño. A chronological long-term ground-based data analysis has allowed the 37 identification of major flooding events occurred in the last century along the Peruvian coast, 38 specifically in the areas near Lima. In particular, two main flooding events that occurred in 39 1997-1998 and 2017-2018 were selected as the focus of this study because, according to the 40 Emergency Events Database [2], they caused significant damage to urban infrastructure.

41
The aim of this research was to assess whether a flood risk map of the city could be generated 42 based on the evidence of flooded areas during these events, as recorded in satellite Synthetic The 3rd International Electronic Conference on Geosciences, 7 -13 December 2020 The 3rd International Electronic Conference on Geosciences, 7 -13 December 2020 Aperture Radar (SAR) images. The basic processing workflow in order to achieve this goal consisted 44 in pre and post-processing of SAR data, generation of RGB composites that showed "where" the 45 change patterns occured, and the ratio maps providing the information on the magnitude of such 46 changes. These products jointly with three key spatial hazard datasets (terrain slope, alluvial deposits and land cover) and ancillary data related to topography, geology, urban footprint, roads 48 and population, allowed us to undertake an integrated evaluation of the hazards and a risk analysis.

49
Key findings from this integrated analysis are presented in this paper. For the detailed analysis

68
The SAR processing workflow ran in the Sentinel Application Platform (SNAP) v.6.0 software, 69 using the Sentinel-1 toolbox (S1TBX), and included: (i) radiometric calibration and precise 70 co-registration of the SAR scenes; (ii) terrain correction and geocoding to map coordinates; and (iii) 71 generation of SAR amplitude color composites and ratio maps.

72
Hazard and risk analysis was conducted by combining SAR amplitude change patterns and 73 geospatial data related to topography, geology, alluvial deposits, urban footprint, new urban 74 development, roads and infrastructure, population at district level.

76
A total of 197 changes attributed to the "Flooding" group were highlighted, for a total of 32.10 The 3rd International Electronic Conference on Geosciences, 7 -13 December 2020 The 3rd International Electronic Conference on Geosciences, 7 -13 December 2020 The flooded areas concentrate mainly in the district of Lurigancho-Chosica (orange polyongs in 81 Figure 2). This is mainly due to the large residential and commercial sectors that were flooded

120
The areas identified at a high risk of flooding are those that in the years 1997-1998 and 121 2017-2018 were directly affected by the flooding event, and where satellite data and/or ground 122 evidence suggest that material loss occurred (e.g., collapse of riverbanks), slope was ≤ 5° or falling 123 between 6° and 10°, and urban fabric is onto alluvial deposits. The medium risk was attributed to 124 non-urban areas that could be affected by flooding events as a result of the combination between 125 slope with values ≤10° and presence of alluvial deposits. In the end, urban areas where neither 126 previous evidence of flooding nor changes in the satellite data were found, and where it is very 127 unlikely that they would be inundated due to their slope characteristics (between 15° and 20°) 128 and/or local geology, were classified at low risk.

161
Conflicts of Interest: The authors declare no conflict of interest.