Extreme Rainfall Events in the Southwest of Rio Grande do Sul (Brazil) and Its Association with the Sandization Process

Part of the sandization process in southwestern Rio Grande do Sul (Brazil) originates from daily torrential rains. However, it is believed that climate changes have been provoking more frequent and more intense rains in the region, a phenomenon which can change the dynamic of erosion/transport/sedimentation natural processes. The objective of this work is to identify the behavior of daily rainfall extreme events (in terms of their frequency, return time, tendency and genesis), relating them to both climatic change issue and enhancement of erosive processes. We have used daily rainfall data from meteorology stations of Brazilian National Water Agency (ANA) for the period between 1928 and 2017 and the percentile 99 was used to identify daily rainfall extreme value (71.5 mm). The upper values were categorized and their absolute and relative frequencies as well as their return time were identified. The temporal tendency of these events was evaluated by the Mann-Kendall test, considering the 90 years of the tensify these sandization processes, since they arise from the association between natural morphoscultural generating environmental problems for the region.


Introduction
Discussions on the occurrence of global climate change and its repercussions at regional and local levels have been a constant theme in scientific debates in recent decades. Problems related to planetary temperature rising, that are associated with the greenhouse gas emissions and the increase of extreme weather events (storms, extreme rainfall, prolonged droughts, heat/cold waves, for example) have guided work produced by several researchers around the world [1]- [11].
At the extreme events issue, [12] defined them as anomalies in relation to climatology, in time scales that can vary from days to millennia. For the authors, the increase in the occurrence of extreme short-term events in recent years has attracted the attention of climatologists. Climate projections proposed by the Intergovernmental Panel on Climate Change (IPCC), in its 5th Report (AR5), published in 2014 [13], has indicated an increment in frequency and intensity of heavy rainfall, heat and cold waves, dry periods, etc. (extreme events).
In southern Brazil (Paraná, Santa Catarina and Rio Grande do Sul states), the formation of rain is associated with several weather systems: frontal systems, cyclones and cold fronts, Mesoscale Convective Complexes (MCC), average-level cyclones (inverted comma-shape) and atmospheric blocking systems, in addition to the indirect actions of the South Atlantic Convergence Zone (SACZ) [14].
In Rio Grande do Sul state (RS), the occurrence of disasters (floods, windstorms, intensive rains, hails, overflows, inundations and landslides), which are associated to extreme meteorological conditions, was studied by [15]. During the months of October, November and December from the year 2003, the MCC were responsible for 90 incidents registered by the Civil Defense of the state, which reached about 16,500 people and caused 11 deaths throughout the state.
In the southwest of Rio Grande do Sul there are grassland areas (from the Pampa biome) that have sandy and uncovered soils, which are known locally as sand stretches (areais in portuguese). These areais are formed by the "reworking of little or non-consolidated sand deposits, that promotes in these areas many difficulties to vegetation fixation, due to sediment mobility by the action of water and wind"-a process called sandization [16] [17]. When these spaces are exposed to irregular and sometimes intensive rain-a regional hydroclimatic dy-namic, it triggers some morphoscultural processes which forms ravines and gullies [18] [19].
So, considering the climatic changing evidences, especially those associated to the occurrence of extreme events, and that intense rains in the region (extreme rains) are responsible for much of the morphoscultural dynamics (sandization) in southwestern Rio Grande do Sul, the purpose of this work is to analyze a long series of rainfall data, identifying the daily rainfall extremes, its frequency, its return time, its temporal tendency and the synoptic mechanisms which are responsible for the genesis of these events.  In order to achieve continuous and homogeneous data, it was tried to maintain the pluviometry stations of Alegrete (which belongs to INMET/ANA) as the main data sources, since they offered the largest set of continuous daily data. The information from other stations were used to fill gaps in main data sources.

Daily Rainfall Data
After applying statistical techniques to verify the homogeneity and consistency of the rainfall stations data-Determination Coefficient (R 2 ), Pearson Correlation (r), Double Mass Analysis, Mean Relative Error, Mean Absolute Error and Root Mean Square Error-the time sequence 1928-2017 was obtained.

1) Extreme rainfall events, Frequency and Return Time
The definition of extreme value for daily rainfall in the study area was obtained by using Expert Team of Climate Change Detection Indices (ETCCDI) recommendation, which uses the percentile 99 (R99) in the identification of extreme events.
After setting up the extreme value index, there were identified and categorized the daily pluviometry volumes which were higher than percentile 99 value through their frequency and relative frequency values (Equation (1)), being estimated also their time of return (Equation (2)). The Return Time (RT) for each extreme precipitation event was defined as the inverse of probability [20].
2) Extreme rainfall events tendency The evaluation of extreme events tendency was made by applying the Mann-Kendall Test (MK). The MK consists of a non-parametric test, recommended by World Meteorological Organization (WMO) to identify climate tendencies over long time series [3] [6] [21].
In this test we have adopted the hypothesis of stability for the time series (H 0 ), in which the values must be independent, and the distribution of their probability must remain the same.

3) Genesis of extreme events: satellite image and synoptic analysis
From the dates of occurrence of extreme rainfall events, it was identified its genesis by using winds, moisture flow and accumulated precipitation informa-American Journal of Climate Change tion, which were obtained through data from European Center for Medium-Range Weather Forecasts (ECMWF), from ERA/Interim reanalysis data (https://www.ecmwf.int/en/forecasts/datasets/archive-datasets/reanalysis-dataset s/era-interim), as well as from Geostationary Operational Environmental Satellite (GOES) 13 infrared images, that were provided by the Center for Weather Forecasting and Climate Studies, from Brazilian National Institute for Space Research (CPTEC/INPE) (http://satelite.cptec.inpe.br/home/index.jsp).

1) Frequencies and Return Time of Extreme Events
From the application of percentile 99 to the precipitation data, the volume of daily rainfalls of 71.5 mm were considered as episodes of extreme precipitation events. However, as shown in Figure 3, such events were very common over the 90 years of the study. Thus, we chose to categorize the daily rainfall events between 71.5 mm (percentile 99 value) and the maximum value during the review period (183.9 mm), extracting frequency, relative frequency and return time variables, cf. Table 1.
It is found that daily rainfall events between 71.5 mm and 80 mm can be repeated at least twice a year. Meanwhile, events between 80 -90 mm, 90 -100 mm and 100 -110 mm, equally intense, can return (statistically) in up to two years. Incidentally, torrential daily rainfall characterizes themselves as common in the study area.
Thus, considering that sandization processes ( Figure 2) have hydroclimatic phenomena (intensive rains, erosive processes-erosion, transport and deposition-and fluvial dynamics) as part of its genesis, intense daily rain events connected to these processes are part of the regional climate dynamics.
Defined by [18] as hydrological summer, the period from September to November is characterized by heavy rains, which are responsible for direct surface   On the other hand, in April and May (during the autumn) (Figure 6), the tendency of these events was not considered significant by the MK assessment.
However, events of this nature (greater than 80 mm) occur relatively frequently in autumn months.
3) Extreme rain events in the southwest of Rio Grande do Sul and its relationship with LLJ and MCC The increase in the amount of intense rainfall events observed in the months of November and December can be associated with the performance of the   Considering that the mean rainfall recycling in Amazonian environment is 22% [23], the remaining moisture for rainfall originates from Atlantic Ocean, by the action of the trade winds [22]. Isotopic analysis (using O 18 ) done by [24] show that about 44% of the water vapor that enters in Amazon region through these winds goes out from Amazon basin to conditioning the rainfall in other regions of South America [25].
This moisture transport is made by low-flowing winds (between 850 and 700 hPa)-LLJ, which, conditioned by the Andean topography, intensify the rains in the Prata basin and in the southeast of South America. The intensification of both frequency and speed of these winds in spring/summer/autumn increases the occurrence of extreme rainfall events in the Prata basin region and in the southeast of South America [26]. Finally, for [27], the increase in global temperatures observed in the second half of the 20th century contributed to the intensification of the western and southern Amazonian moisture fluxes through the LLJs, increasing the frequency of intense rainfall events in the Prata basin. So, it is possible to relate the increase of the frequency of extreme precipitation events in the study area to the increment of LLJ from Amazon region, which promoted the formation of a larger number of MCC phenomena.

Conclusions
The reconstruction of a long time series of daily rainfall data and the application of a statistics set (percentile 99, absolute frequency, relative frequency and return time) over that data allowed identifying extreme rainfall events for the southwestern region of Rio Grande do Sul. Daily rainfall data have shown much higher pluviometry events than the value considered as extreme. In addition, the phenomenon of extreme rainfall has been repeated more frequently and more intensively over the last 20 years of the series.
The assessment of the tendency for the increasing on heavy rain days in November and December suggests that such events have become more frequent in recent years and it is considered significant for changes in weather behavior-a possible evidence of climate changing.
The origin of these intense rains can be attributed to the formation of large convective systems (MCCs). They are formed by the atmospheric runoff and the low level moisture transport originated in the Amazon region; as ongoing climate changes (the Amazon deforestation, the increment of the amount of aerosols, the heating of atmosphere and oceans, etc.) intensify, it is believed that LLJs tend to become more intense, carrying more moisture and promoting MCC in greater quantity and magnitude.
More intense and more frequent extreme rainfall events in the areais region, in the southwestern portion of Rio Grande do Sul, intensified the natural processes of morfoescultural dynamics, enhancing geomorphological processes F. Sanches et al.