Extreme rainfall erosivity: Research advances and future perspectives

Extreme rainfall erosivity, the capacity of intense rainfall to induce soil erosion, is vital for anticipating future impacts on soil conservation. Despite extensive research, significant differences persist in terms of understanding influencing mechanisms, potential impacts, estimation models and future trends of extreme rainfall erosivity. Quantitatively describing extreme rainfall erosivity remains a key issue in existing research. In this study, we comprehensively reviewed the literature to assess the relationships between extreme rainfall characteristics and rainfall erosivity, between extreme rainfall erosivity and soil erosion, estimation models and trend prediction. The aim was to summarize previous related research and achievements, providing a better understanding of the generation, impacts and future trends of extreme rainfall erosivity. Future research directions should include identifying the thresholds of extreme rainfall events, increasing research attention on tropical cyclones in terms of rainfall erosivity, considering on the impact of extreme rainfall erosivity on soil erosion, and improving rainfall erosivity estimation and simulation prediction methods. This study could contribute to adapting to global climate change and aiding in formulating soil erosion prevention and environmental protection recommendations.

The influence in rainfall erosivity calculation by using different temporal resolution in Mediterranean area

Rainfall erosivity (EI30) is crucial to characterize the rainfall effect on soil erosion based on storm intensity. Its calculation is highly sensitive to the time resolution used, in which using rainfall data at fixed time intervals (ΔΤ) >30-min results in large underestimation. Therefore, there is a need to explore the difference and correlation between EI30 calculated at different ΔΤ. One-minute rainfall data from 2006 to 2022 were collected from 6 stations over the Basilicata region in southern Italy to compute the maximum 30-min rainfall intensity (I30), total kinetic energy of storm (KE), EI30 and erosivity density values, for a total of 2516 storm events. These data constitute the actual values of I30, KE and EI30 and will be used as reference data. Underestimation of all the considered parameters were systematically evaluated using data aggregated at 5-, 10-, 15-, 30- and 60-min fixed interval. For ΔΤ ≤ 15 min the parameter responsible of the greatest underestimation turns out to be KE, whereas for coarser temporal resolution (ΔΤ > 30 min) I30 plays a dominant role in underestimating EI30. The use of coarse temporal resolutions also leads to >5 % loss of erosive events, especially those characterized by middle to high intensity/low duration (ΔΤ ≤ 45 min) events, as well as to an underestimation higher than 30 % in the estimated rainfall erosivity. The results show that an accurate estimation of the rainfall erosivity requires the use of rainfall data with a fixed time interval of length lower than 10 min.

Improvement of sediment yield index model through incorporating rainfall erosivity

Soil erosion and sediment yield in watersheds are comprehensively affected by land use/cover changes and climatic factors. The current sediment yield index (SYI) model incorporates parameters of area (A), delivery ratio (DR), and curve number (CN), which reflect the character of underlying surface conditions, while the impact of rainfall intensity on sediment yield could not be properly considered. This study aims to improve the current SYI model by introducing rainfall-related factors such as rainfall erosivity (R) and applying it to estimate the sediment yield of river basin. Taking the Dongjiang River basin, South China, as a case study, the performances of the improved simplified SYI model (SYI-CN + R) were compared and demonstrated at multi-spatiotemporal scales. The results showed that (1) compared with the SYI model which only has the parameter CN (SYI-CN), the model SYI-CN + R achieves better simulation performances at yearly (the efficiency coefficient (CE) is 81% in the whole basin and 62% in the sub-basin) and half-month (CE is 69% in the whole basin and 57% in the sub-basin) time scales. (2) On the basin scale, the simulation performance in the whole basin is better overall compared to that in the sub-basin, and the model SYI-CN + R at the half-month time scale is more suitable for the sediment yield simulation in the Dongjiang River basin, with higher value of correlation coefficient (CC) of 87% and 83% for the whole basin and the sub-basin, respectively. And (3) the values of CN and R have an obvious spatial gradient in the whole basin, showing an increasing trend from northeast to southwest as a whole, with larger values concentrated in the lower reaches and smaller values in the middle and upper reaches. This study extends the application and improves the performance of the SYI model, and provides a basis for soil and water conservation in a river basin with fewer observation data.

A new application of deep neural network (LSTM) and RUSLE models in soil erosion prediction

Rainfall variation causes frequent unexpected disasters all over the world. Increasing rainfall intensity significantly escalates soil erosion and soil erosion related hazards. Forecasting accurate rainfall helps early detection of soil erosion vulnerability and can minimise the damages by taking appropriate measures caused by severe storms, droughts and floods. This study aims to predict soil erosion probability using the deep learning approach: long short-term memory neural network model (LSTM) and revised universal soil loss equation (RUSLE) model. Daily rainfall data were gathered from five agro-meteorological stations in the Central Highlands of Sri Lanka from 1990 to 2021 and fed into the LSTM model simulation. The LSTM model was forecasted with the time-series monthly rainfall data for a long lead time period, rainfall values for next 36 months in each station. Geo-informatics tools were used to create the rainfall erosivity map layer for the year 2024. The RUSLE model prediction indicates the average annual soil erosion over the Highlands will be 11.92 t/ha/yr. Soil erosion susceptibility map suggests around 30 % of the land area will be categorised as moderate to very-high soil erosion susceptible classes. The resulted map layer was validated using past soil erosion map layers developed for 2000, 2010 and 2019. The soil erosion susceptibility map indicates an accuracy of 0.93 with the area under the receiver operator characteristic curve (AUC-ROC), showing a satisfactory prediction performance. These findings will be helpful in policy-level decision making and researchers can further tested different deep learning models with the RUSLE model to enhance the prediction capability of soil erosion probability.

Letter to the editor regarding Li et al. (2022) Identifying ecosystem service bundles and the spatiotemporal characteristics of trade-offs and synergies in coal mining areas with a high groundwater table, Liu et al. (2021) Ecosystem service multifunctionality assessment and coupling coordination analysis with land use and land cover change in China's coastal zones, and Zhang et al. (2021) Spatial relationships between ecosystem services and socioecological drivers across a large-scale region: A case study in the Yellow River Basin

Three studies used empirical equations to calculate the rainfall erosivity factor R, and all three equations appeared to be incorrect. All of the studies were published in the journal Science of the Total Environment, and none of them accurately cited the sources of the incorrect equations that were used in them. We were able to track down the original equation as well as the source of the equation. Additionally, it was discovered that the original equation contained an incorrect conversion factor, which needs to be corrected.

Spatiotemporal evolutionary analysis of rainfall erosivity during 1901-2017 in Beijing, China

Rainfall erosivity is regarded as one of the main factors affecting soil erosion. Based on 117-year monthly precipitation data of Beijing from 1901 to 2017, the  spatiotemporal evolutionary analysis of rainfall erosivity in Beijing were analyzed by using Theil-Sen median analysis (Sen), the Mann-Kendall (MK) trend test, R/S analysis method, cumulative anomaly method, MK mutation test method, Pettitt test, and wavelet analysis. The results showed that the average annual rainfall erosivity in Beijing ranged from 1080.6 to 6432.78 MJ • mm/(hm² • h • a), with an average value of 3465.06 MJ • mm/(hm² • h • a), showing a gradual decrease from the southeast to northwest. Regarding seasonal distribution, 86% of rainfall erosivity was mainly concentrated in summer. In the past 117 years, the annual rainfall erosivity in most areas of Beijing showed a downward trend, but its future trend also showed an increasing trend, indicating that Beijing, especially the northern part, was facing greater potential pressure from soil erosion. Through cross-validation of various methods, the abrupt change interval of rainfall erosivity in Beijing from 1901 to 2017 was from 1994 to 1997. The change in rainfall erosivity in Beijing had a strong oscillation in 32 years and a small periodic change in 15 and 7 years. The results will provide a decision-making basis for soil erosion control and water/soil conservation planning. Additionally, they will benefit and ensure national agricultural and food security.

Assessment of rainfall erosivity (R-factor) during 1986-2015 across Nepal: a step towards soil loss estimation

Rainfall is a main cause of soil erosion which varies spatially and temporarily. R-factor is an erosive power of the rainfall that is responsible for soil detachment and subsequent displacement. Mathematically, it is expressed as a sum of the product of kinetic energy and maximum 30-min rain intensity. A precise assessment of R-factor needs higher temporal resolution rainfall data (sub-hourly) for a period of several years, which is rarely available. Many empirical approaches are used to predict R-factor as a function of mean monthly and annual rainfall amount. In this study, we used Loureiro and Countinho (Journal of Hydrology 250:12-18, 2001) approximation approach to estimate R-factor and explore its intra-annual variability using 30 years (1986-2015) of daily rainfall data from 280 stations distributed across Nepal. This study employs different intra-annual variability indices and calculates erosivity density (ED) and weighted erosivity density (WED). The country average mean annual R-factor (MAR), annual ED, and WED are found to be 9434.8 MJ mm ha^-1 h^-1 year^-1, 4.39 MJ ha^-1 h^-1,and 1.61 MJ ha^-1 h^-1, respectively. On a monthly scale, July is the highest erosive month followed by August (> 2000 MJ mm ha^-1 h^-1 month^-1). Likewise, November is the lowest erosive month followed by December (~ 50 MJ mm ha^-1 h^-1 month^-1). Spatial distributions of these indices show clear delineations of areas with different erosivity patterns at different time of the year. In addition, this study explores inter-annual variation, temporal evolution, and trend estimation of R-factors over the country (for the first time). Significant rising trends are observed in the western region of the country. We found that the mean soil erosion for Nepal is estimated at 21.01 ton ha^-1 year^-1. The smallest R-factors are observed in the north-western region of the country and the maximum values are observed at mid hills and southern plains of the country. Our study could be an initial but important step for effective soil conservation, land use planning, and agricultural production.

Variation characteristics of rainfall erosivity in Guizhou Province and the correlation with the El Niño Southern Oscillation

Rainfall erosivity is an important indicator that can be used to measure the ability of rain to cause erosion and is connected with the El Niño Southern Oscillation (ENSO) through the transmission of rainfall. This work aimed to explore the characteristics of rainfall erosivity in Guizhou Province and determine its correlation with ENSO. Rainfall erosivity was calculated from daily rainfall data from January 1960 to December 2017. The analyses were conducted using a daily rainfall erosivity model, inverse distance weighted (IDW) interpolation, linear regression analysis, Mann-Kendall test and correlation analysis. The long-term (1960-2017) average rainfall erosivity was 5825.60 MJ·mm·ha^-1·h^-1 in the study area and showed a high temporal variability with the estimates from the linear trend line ranging from -449.5 MJ·mm·ha^-1·h^-1/10a to 496.8 MJ·mm·ha^-1·h^-1/10a. According to rainfall and erosive rainfall, an uneven spatial distribution of rainfall erosivity was observed with an increasing trend from south to north. Temporal distribution of monthly rainfall erosivity was consistent with that of seasonal rainfall erosivity, and concentrated in the summer months (June to August). As the representation indices of ENSO phenomena, the Oceanic Niño Index (ONI), Southern Oscillation Index (SOI) and multivariate ENSO Index (MEI) were selected for correlation analysis with rainfall erosivity. During El Niño events, the ONI, SOI and MEI showed significant correlations (>95% confidence level) with rainfall erosivity, while during La Niña events, only the ONI and MEI were significantly correlated with rainfall erosivity, but no significant correlation was detected during the neutral period or for the entire study period. The degree of rainfall erosion is proportional to the ENSO duration; the longer the ENSO duration, the greater the rainfall erosivity. These findings could help predict soil erosion and be used to develop further adaptation measures to prevent water and soil loss.

Contribution of soil erosion to PAHs in surface water in China

In China, the total annual atmospheric emission of 16 polycyclic aromatic hydrocarbons (PAHs) reach up to approximate 100 thousand tons, part of which is preserved in soils. In this study, the contribution of soil erosion to PAHs in surface water nationwide was quantified. The results indicate that a major portion of the annual PAHs emission is lost from soils via rainfall erosivity and subsequently transported to the ocean. The national annual flux of PAHs from soil to surface water by the natural physical forces of water measures up to ~70 thousand tons, which accounts for ~62% of the annual emission of PAHs with 19% entering the sea directly. In general, both the soil erosion intensity and flux of PAHs for the regions located in the Southeast of China are over those in the Northwest of China, with the regions being divided into two different parts by the famous geographic "Hu Huanyong line", reflecting the intensive impact of human activities on environmental degradation. Comparative analysis suggested that there must be a big fraction of PAHs lost during transmission due to the river sedimentation and lake dispersion. This study closes a major gap in the national budget of PAHs and provides critical information in the context of regional environment risk assessment.

Computing structural and functional flow and sediment connectivity with a new aggregated index: A case study in a large Mediterranean catchment

Hydrological connectivity in large catchments is influenced by natural and human-induced heterogeneities and dynamic processes. In this study, a new aggregated index (AIC) based on topography, C-RUSLE factor, RUSLE2 rainfall erosivity, residual topography and soil permeability, was proposed to model structural and functional flow and sediment connectivity (FSC). It was tested in a large Mediterranean catchment (Vero River, NE Spain, 380 km²) with contrasted physiographic and climatic conditions (19 land uses and 15 types of lithology). Twelve weather stations were used and simulations were done at 5 m of pixel resolution using a LiDAR-derived DEM and the D-Infinity algorithm. Structural FSC (FSC-st) was computed with both an updated version of Borselli's index (IC) and the AIC. Values of connectivity with AIC followed a normal distribution with a wider range of values compared with the non-normal distribution obtained with Borselli's approach. The differences in the values of FSC-st between the different land uses were similar with the two indices and in agreement with the soil erosion rates reported in comparable landscapes. The spatial characteristics at sub-catchment scale were better reflected with AIC although values of FSC-st in the river and outlet were similar between both indices. Functional FSC (FSC-fn) was computed with AIC during 96 months (September 2009-August 2017) characterising the spatio-temporal dynamic at catchment scale (18% of coefficient of variation). FSC-fn was higher in September, October, June and July and lower during the period December-February. Variation of connectivity in the stream was higher than in the hillslopes. Modelling testing with river flow was satisfactory between November and March, and during the months with high discharge values and weak during the summer, suggesting different runoff and sediment responses over the year. The new AIC appeared as a suitable tool for geomorphic and hydrological studies at catchment scale.

Rainfall erosivity in Slovenia: Sensitivity estimation and trend detection

Slovenia is one of the EU countries with the largest values and largest amounts of variability in rainfall erosivity, with maximum annual values exceeding 10,000 MJ mm ha^-1 h^-1 yr^-1. Five-minute rainfall data was analysed from 10 Slovenian rainfall stations with data-length availability longer than 25 years with a maximum data length of 69 years and a total data-station length equal to 443 years. Trends in the rainfall erosivity R-factor were detected for four different sub-samples using monthly, half-year, and annual rainfall erosivity values. The results indicate that rainfall erosivity trends for the selected Slovenian stations are mostly statistically insignificant, with the selected significance level of 0.05. However, a larger share of identified trends are positive than negative. The maximum annual rainfall erosivity values were obtained for one specific mountain station. Furthermore, a sensitivity analysis regarding the rainfall erosivity factor R calculation showed that the rainfall threshold parameter (12.7 mm) that is used to remove the small-magnitude rainfall events in order to reduce the computational burden can attribute up to 10% of the average annual R-values in cases where this threshold is not used. Other parameters have, on average, a smaller impact on the calculated rainfall erosivity. Furthermore, the application of local kinetic energy equations resulted in, on average, about 20% higher annual rainfall erosivity values compared to the equation that is proposed by the Revised Universal Soil Loss Equation (RUSLE) manual and was not developed specifically for this region.

Computation of rainfall erosivity from daily precipitation amounts

Rainfall erosivity is an important parameter in many erosion models, and the EI30 defined by the Universal Soil Loss Equation is one of the best known erosivity indices. One issue with this and other erosivity indices is that they require continuous breakpoint, or high frequency time interval, precipitation data. These data are rare, in comparison to more common medium-frequency data, such as daily precipitation data commonly recorded by many national and regional weather services. Devising methods for computing estimates of rainfall erosivity from daily precipitation data that are comparable to those obtained by using high-frequency data is, therefore, highly desired. Here we present a method for producing such estimates, based on optimal regression tools such as the Gamma Generalised Linear Model and universal kriging. Unlike other methods, this approach produces unbiased and very close to observed EI30, especially when these are aggregated at the annual level. We illustrate the method with a case study comprising more than 1500 high-frequency precipitation records across Spain. Although the original records have a short span (the mean length is around 10 years), computation of spatially-distributed upscaling parameters offers the possibility to compute high-resolution climatologies of the EI30 index based on currently available, long-span, daily precipitation databases.

A new methodology for estimating rainfall aggressiveness risk based on daily rainfall records for multi-decennial periods

The temporal irregularity of rainfall, characteristic of a Mediterranean climate, corresponds to the irregularity of the environmental effects on soil. We used aggressiveness as an indicator to quantify the potential environmental impact of rainfall. However, quantifying rainfall aggressiveness is conditioned by the lack of sub-hourly frequency records on which intensity models are based. On the other hand, volume models are characterized by a lack of precision in the treatment of heavy rainfall events because they are based on monthly series. Therefore, in this study, we propose a new methodology for estimating rainfall aggressiveness risk. A new synthesis parameter based on reformulation using daily data of the Modified Fournier and Oliver's Precipitation Concentration indices is defined. The weighting of both indices for calculating the aggressiveness risk is established by multiple regression with respect to the local erosion R factor estimated in the last decades. We concluded that the proposed methodology overcomes the previously mentioned limitations of the traditional intensity and volume models and provides accurate information; therefore, it is appropriate for determining potential rainfall impact over long time periods. Specifically, we applied this methodology to the daily rainfall time series from the San Fernando Observatory (1870-2010) in southwest Europe. An interannual aggressiveness risk series was generated, which allowed analysis of its evolution and determination of the temporal variability. The results imply that environmental management can use data from long-term historical series as a reference for decision making.

Linking sedimentary total organic carbon to 210Pbex chronology from Changshou Lake in the Three Gorges Reservoir Region, China

The influences of total organic carbon (TOC) and total nitrogen (TN) on Lead-210 (²¹⁰Pb) dating have recently been of increasing concern in lacustrine research. Sediment core from Changshou Lake in the Longxi catchment was investigated for influence of TOC on ²¹⁰Pb dating. Lead-210 excess (²¹⁰Pb_ex), Cesium-137 (¹³⁷Cs) activities, TOC, TN, and particle size were measured. We proposed a dating index based on ¹³⁷Cs chronology and particle size distribution of the lake sediment profile and rainfall erosivities calculated from Longxi catchment metrological records. Increasing trends in TOC and TN were specifically caused by commercial cage fish farming after 1989. The statistically significant correlation between ²¹⁰Pb_ex activity, TOC (0.61, p = 0.04) and TN (0.51, p = 0.04), respectively explained post-1989 ²¹⁰Pb scavenging. The ²¹⁰Pb_ex activity was closely related with coupled peaks of TOC and TN from mass depth 5-10 g cm^-2. Higher TOC/TN ratio (8.33) indicated submerged macrophytes and native aquatic algal growth as main source of carbon from enhanced primary productivity because of massive fertilizer use and coherent climate warming. The study supported key hypothesis on vital role of fertilizer usage and algal derived TOC in controlling sedimentary ²¹⁰Pb_ex activity at Changshou Lake sediment. ¹³⁷Cs profile and erosive events as time markers provided reliable and consistent sedimentation rate of (1.6 cm y^-1). ²¹⁰Pb_ex activity decayed exponentially after peak at mass depth 5.68 g cm^-2. Therefore, violation of ²¹⁰Pb dating primary assumptions made it inappropriate for sediment dating at Changshou Lake. TOC content must be considered while using ²¹⁰Pb as dating tool for lake sediment profiles.

Use of disdrometer data to evaluate the relationship of rainfall kinetic energy and intensity (KE-I)

Determination of rainfall kinetic energy (KE) is required to calculate erosivity, the ability of rainfall to detach soil particles and initiate erosion. Disdrometers can measure rainfall KE by measuring raindrop size and velocity. In the absence of such devices, KE is usually estimated with empirical equations that derive KE from measured rainfall intensity (I). We evaluated the performance of 14 different KE-I equations to estimate the 1min KE and event total KE, and compared these results with 821 observed rainfall events recorded by an optical disdrometer in the inner Ebro Basin, NE Spain. We also evaluated two sources of bias when using such relationships: bias from use of theoretical raindrop terminal velocities instead of measured values; and bias from time aggregation (recording rainfall intensity every 5, 10, 15, 30, and 60min). Empirical relationships performed well when complete events were considered (R(2)>0.90), but performed poorly for within-event variation (1min resolution). Also, several of the KE-I equations had large systematic biases. When raindrop size is known, estimation of terminal velocities by empirical laws led to overestimates of raindrop velocity and KE. Time aggregation led to large under-estimates of KE, although linear scaling successfully corrected for this bias.