Abstract:
Most natural disasters in Brazil are triggered by hydrological/climatological events. Landslides are among the types that have caused the most deaths in the country.The volume of rain associated with the high slope of the land and a history of urban occupations with precarious infrastructure has corroborated the occurrence of this type of disaster in the Ribeirão do Peixe basin, in Itabira in the central region of Minas Gerais. In this sense, in addition to cataloging in time and space, the occurrences of landslides between 1996 and 2022, this research aimed to analyze the distribution of rainfall in order to infer correlations of precipitation thresholds that may have triggered such geo-hydrological events.The methodological stages of this study were divided into three work fronts. The first refers to an analysis of the distribution of rainfall considering historical series and the index of maximum forecasts accumulated over five days (Rx5day) monthly and annually, also considering the existence of statistical significance using the non-parametric Theil-Sen method. In the second stage, documentary and georeferencing research was carried out in the collection of the Coordenadoria Municipal de Proteção e Defesa Civil de Itabira (COMPDEC) to interpret the distribution of landslides in space and time. And finally, the daily and accumulated rainfall (48, 72, 96 hours) and (5, 7 and 12 days) were evaluated. Some trends in the behavior of precipitation in the basin were observed. The annual rainfall volume of the 1977-2022 series (Conceição Station) appears to be decreasing. The maximum precipitation on 5 consecutive days appears to be increasing. However, except for the month of October, practically all months had no reported statistical significance. Of the 263 landslides recorded, the most affected neighborhood was Gabiroba with around 20%. The Madre Maria de Jesus neighborhood had the highest proportion (2.42 Dsl/Ha). In relation to relief units, 86.23% are in undulating or strongly undulating areas. The geological units with the highest number of occurrences were the granitic-gneissic terrains of the Borrachudos suite followed by the schists and phyllites of the Nova Lima Group.
All 263 events catalogued, 72 of them or 27% were recorded on a dry day (P<2 mm) and 191 events above 2mm. For at least 1 slip, the averages for the smallest frequency analysis intervals were 35mm for 48 hours, 44mm for 72h, 51mm (96h), 57mm (5d), 66mm (7d) and 76mm (12d). For 5 slips or more 99 mm for 48 hours, 107 mm for 72 h, 133 mm (96h), 151 mm for 5 days, 194 mm (7d) and 231 mm (12d). Through the graphical analysis it was not possible to establish equations that represented the randomness of the scattered data. From a decision-making perspective involving alert systems, 48-hour thresholds may represent greater predictability in practice. However, regarding monitoring, preparation and response activities to possible disasters, it may be important to use critical thresholds in a slightly longer time window, such as 7 days for example. The products generated here will be able to support short-term mitigation actions (alert, monitoring and evacuation plans) and long-term with the aim of composing a database and information for decision-making that helps reduce the risks of new landslides.