Assessment of current reservoir sedimentation rate and storage capacity loss: An Italian overview

Sedimentation has a prominent impact on the functionality and lifetime of reservoirs and is a growing concern for stakeholders. Various parameters influence sedimentation caused by soil erosion. Here we have examined fifty Italian reservoirs to determine sedimentation rates and storage capacity loss. The reservoirs studied have an average age of 78 years as of 2021, with the highest loss of capacity observed, equal to 100%, for Ceppo Morelli. For the fifty Italian catchments covering north, south, central and islands of Italy, we found the mean annual sediment yield varying between 17 – 4000 m 3 /km 2 . year. Six of fifty reservoirs studied (Quarto, Colombara, Ceppo Morelli, Fusino, Vodo and Valle di Cadore) are already in a very critical situation in terms of storage capacity loss. Out of the fifty reservoirs, half of them will reach their half-life year by 2050. For example, for the Fusino reservoir located in northern Italy, we observed a loss of 90% of the storage volume as of 2020 with respect to its operation year 1974, compared to 6% in 2015 as available in literature. Modelling the sediment delivery ratio (SDR) is an open question, due to the lack of adequate data and uncertainties about the variability in hydrological, geomorphological, climate and landcover parameters. Here, we addressed the issue with a simplified multiple regression approach based on sediment delivery ratio values retrieved by the RUSLE model. We found different multi regressions for reservoirs belonging to the Alpine and Apennine regions. This analysis offers a starting point for the management and prioritization of adaptation and remediation policies necessary to address reservoir sedimentation.


Introduction
The utmost important concern for dam operation is the decrease of reservoir storage capacity owing to sedimentation in the reservoir (Cogollo and Villela, 1988;Evans et al., 2000;Petts, 1984;WCD, 2000). Reservoir sedimentation is one of the most challenging problems that affect hydropower production globally (De Cesare et al., 2001;Palmieri et al., 2001). The phenomenon by which the sediments are transported and deposited into the reservoirs by the streamflow is known as reservoir sedimentation (Morris and Fan, 1998). The soil erosion resulting from catchment anthropogenic activities is a major contributor to reservoir sedimentation. The quantity and type of sediment depend on the topographic character, geology, vegetation type, climate, and land use within the catchment (Schleiss, 2013). The suspended solids are deposited in the settling zone of the reservoir when the settling velocity is exceeded. The gradual process of sedimentation proceeds with different speeds that depend on a large number of factors, such as hydrology of the catchments and the characteristics of the river basin (Kondolf et al., 2014). The natural sediment cycle is a coherent continuum of mechanical and chemical processes along a river basin, consisting of three processes production, transport and deposition of sediment (Burt and Allison, 2010;Schleiss et al., 2016).
Reservoir sedimentation can lead to a wide variety of problems like water quality degradation, increase the flooding risk, decrease in the reservoir capacity and power generation, negative impacts on water availability for irrigation, wildlife, and navigation, and affecting the dam's life cycle, among others (Bohrerova et al., 2017;Poff and Olden, 2017;Schmitt et al., 2018;Wisser et al., 2013). The rate of loss to the storage capacity depends on the sediment yield of the river on which a reservoir is built, the morphologic factors of the reservoir, and its operational scheme (Billi and Spalevic, 2022;Kishore et al., 2021). Sediment influx and deposition pose a pronounced threat to the storage capacity of a reservoir (Asselman and Middelkoop, 1995;Fryirs, 2013;Meade and Moody, 2010) with the projected mean annual global loss of storage capacity rate in the range of 0.5-1%. According to the latest report of the International Commission On Large Dams (ICOLD, 2009), the global average value for sedimentation rate (and so storage loss) is around 0.96%. Palmieri et al. (2003) found that the global loss of reservoir storage capacity is around 0.8%, i.e., a rate of 45 km 3 per year of sediments trapped by the dams worldwide, resulting in a loss of revenue amounting to 13 billion $ every year. Nevertheless, the rates of sedimentation are not uniform worldwide. Along the lines of the estimates made by ICOLD, the Italian Committee for Large Dams (ITCOLD, 2009) estimated an overall loss of the reservoir volume of 30%, on a sample equal to 53% of large Italian reservoirs. In the study conducted by ICOLD (2009), it was reported that the majority of the existing reservoirs will completely lose their storage capacity within the next century if mitigation measurements are not implemented. However, the effect of sedimentation varies depending on the region and the climate considered, as erosion and depositional processes might differ. The situation is getting critical worldwide, and it will worsen over time. Reservoirs are a key infrastructure element of the water-energy-food nexus (via agriculture and electricity) in the realization of highly interlinked sustainable development goals (Quaranta et al., 2021;UN, 2015). Various technological, management and governance approaches for understanding sediment accumulation in reservoirs are mostly restricted to local or regional studies of single or a few reservoirs due to a lack of data (Fry et al., 2022;Schleiss et al., 2016).
One of the earliest pieces of available literature concerning the sedimentation issue in Italian reservoirs is Bazzoffi et al. (1996). A study of the sediment yield across forty Italian catchments covering north, south and central Italy found the mean annual sediment yield varying between 0.2 and 20 ton/ha/year (Van Rompaey et al., 2005). The reservoirs situated in north-eastern Italy across the Piave river have led to a decrease in the sediment load from 1 Mm 3 to 0.145 Mm 3 per annum (Surian et al., 2009). The Fusino reservoir located in northern Italy has lost 6% of the storage volume as of 2015 with respect to its operation year 1974 (Pagliari et al., 2017). A study carried out across 20 reservoirs located in the central Alps of northern Italy found the median rate of sedimentation to be 5.6 cm per year (Marziali et al., 2017). A decreasing trend of suspended sediment yield was observed for the last decade in a recent study of nine Italian rivers based on hydrological data (Billi and Spalevic, 2022). Most of these papers have investigated the denudation rates, rather than the sediment yield, using different approaches. Such studies of soil erosion in Italy were done recently for five experimental plots in southern Italy (Porto et al., 2022), Valchiavenna valley across Mera river (Maruffi et al., 2022), Basilicata Region (Gioia et al., 2021), Turano basin (Borrelli et al., 2014), Magra River Basin (Diodato et al., 2022), Oltrepo Pavese basin , Tronto river in southern Marche region (Bufalini et al., 2022) and lake Iseo (Rapuc et al., 2022). The knowledge about sediment yield in Italy is rather scarce and only a few have investigated reservoir sedimentation.
In a climate change context, the issue of sedimentation becomes more crucial, because climate change can lead to modifications in glacier volume and area, resulting in exposure to frozen soils, which in turn, increases the rate of erosion from those areas (Hallet et al., 1996;Leeder et al., 1998). A decrease in the water storage volume of a reservoir will have direct economic consequences on the revenue due to a loss in regulation capability (see Bonato et al., 2019;Gaudard et al., 2018;Patro et al., 2018 for the case of hydropower). For this reason, it is important to analyze the current situation and foresee future scenarios to determine possible impacts on dam operations and their adaptation to the issue of reservoir sedimentation. There is a consensus in the scientific literature about the necessity of data-driven studies and approaches regarding reservoir sedimentation but is proven to be challenging in recent decades, as few authorities monitor sediment in a regular manner (Kondolf and Yi, 2022;Landwehr et al., 2022).
Unfortunately, a huge threat to the effective operation of reservoirs is the reduction of their life span due to the rising sedimentation rate caused by large changes in their watersheds. Soil erosion in the catchment is an important parameter as the sediment yield depends on it. Thus, a systematic approach is needed to quantify and monitor the sedimentation and its impact on the dam-reservoir system. Soil erosion models can be grouped primarily into three types which are Empirical or Statistical (e.g., USLE, RUSLE, MUSLE), Conceptual (e.g., Agricultural Non-Point Source) and Physical based models (e.g., Water Erosion Prediction Project, Areal Non-point Source Watershed Environment Response Simulation) (Merritt et al., 2003). Physically based models are generally the most scientifically robust and flexible in both input and output and are based on an understanding of the physical processes that cause erosion and are therefore applicable to a wide range of soils, climatic and land use conditions (Yeon et al., 2021). But such models are difficult to parameterize as they are data-intensive and the amount of data needed for gross erosion is not readily available (Sotiri et al., 2021). Therefore empirical models are usually employed in situations with limited data and parameter inputs (Alewell et al., 2019). When dealing with the results obtained from empirical models, the challenges are encountered mainly in having a reliable assessment of the sediment input in cases where no monitoring station is available (Merritt et al., 2003). In this work for assessing the sediment yield from the catchment, the Revised Universal Soil Loss equation (RUSLE) based model was also used due to limited data availability. The main aim of the work is to provide an overview of the current situation of reservoir sedimentation in Italy. The novelties are: a) it presents a national-scale analysis of reservoir sedimentation in Italy, including fifty catchments covering north, south, central and even the islands of Italy, with relevant implications for the country's policymakers; b) it examines the health of the dams by reservoir sedimentation characteristics, reservoir variables and the average annual rate of sedimentation in the reservoir using in-situ observations; c) it quantifies the spatial variation of sediment delivery ratio and connects this to some catchment variables, making a distinction between reservoirs located in Alpine and Apennine regions. To monitor the dams' current situation, we examined the catchment and reservoir characteristics to understand the reservoir's loss of capacity, depletion rate, trap efficiency, sediment yield, and half-life. We have used a RUSLE model to estimate the sediment delivery ratio (SDR). Then we have proposed multi regressions (distinguished for reservoirs belonging to Alpine and Apennine regions) to have a quick and first approximation assessment of the sedimentation connectivity issue in situations with limited/absence of field data. The work is organized as follows: in section 2 we present the case study and the reservoirs considered and data available; in section 3 we describe the methodology used; in section 4 we give the results while in section 5 the discussion concerning the legislation context, finally in section 6 we provide the conclusions.

Reservoir inventory
Out of a total of 543 large dams currently working on the Italian territory, an analysis was done for fifty dams which are about 9% of the total dams and correspond to a total volume of 0.59 billion m 3 (approximately 4.4% of the total volume equal to 13.35 billion m 3 ). The reservoirs analyzed are distributed throughout Italy, mainly within the following ten administrative regions of Piemonte, Emilia-Romagna, Abruzzo, Lazio, Sicily, Tuscany, Marche, Veneto, Sardegna, and Lombardy. The fifty reservoirs are mainly located in two macro-areas: the Alpine one, located in northern Italy characterized by the Alps and that of Apennine relative to the other regions (central, south and islands of Italy) characterized by the presence of the Apennine range. Fig. 1 shows the geographical location of each reservoir, and all the reservoirs are numbered from 1 to 50 for better readability with Table 1. Table 1 also E.R. Patro et al. Journal of Environmental Management 320 (2022) 115826 provides the physical characteristics of study sites, while Table A1 (the letter A refers to Appendix A) reports the years for which reservoir volume information is available. We collected information about maximum reservoir volumes, height-volume curves in different years derived from bathymetric studies and information related to flushing activities if any, for about fifty reservoirs in Italy. Before the dams came into operation (Pre, 1975 period), the bathymetric survey was done manually (error ±10%). The reported information comes from the Italian dam authorities' office in Milan (https://trasparenza.mit.gov.it/) and dam operating companies such as ENEL, Edison and A2A. Conversely, most of the bathymetric surveys carried out post 2000 were done using either a) boat bathymetry on predefined sections (error ±3%); b) with differential GPS equipment synchronized with mono or  Table 1). E.R. Patro et al. Table 1 All parameters and quantities are considered for defining the state of sedimentation for the reservoirs examined (the reservoir locations are given in Fig. 1 multi-beam sonar on a small boat (error ±2%); c) ground topographic techniques (error depending on the equipment used); d) aerial photogrammetric techniques (error depending on the equipment used). The post 2000 surveys have an error quite in line with a value of 5%, to be considered a good precision value, as given in Estigoni et al. (2014).
Since the operational life of the reservoirs is quite long and the sedimentation is already being accounted for in their design, many dam operators do not carry out such studies continuously for long period, or with regular frequency (Patro, 2020).

Topographic data
A Digital Elevation Model (DEM) of Italian territory, with a resolution of 20 × 20 m 2 and WGS84 UTM 32N reference system (available at ISPRA website https://www.isprambiente.gov.it/en) was used to compute the catchment area and other topographic properties like average catchment elevation, slope, aspect and the Italian river network.
The input data namely bulk density ρ b , R factor, K factor, LS factor, P factor and C factor, for modelling the soil erosion were obtained from the European Soil Data Centre (ESDAC) (Panagos et al., 2012). Further details about the variables mentioned here can be found in Section 3 Methods.

Methods
This section gives an outline of the research methods utilized to understand the health of the reservoirs. The first part of this section describes the various variables computed for the evaluation of reservoir sedimentation based on the reservoir and catchment parameters of the study areas. The second part enumerates the computed variables to retrieve SDR. Fig. A1 shows an overview of the methodology adopted in the present study. All the variables discussed are then used in multi regression equations to represent the rate of reservoir sedimentation.
The sediment volume of a reservoir (V INT ) is the total volume of sediments deposited inside the reservoir in a given time interval (ΔT = T f − T i , in years) between a final time T f and an initial time T i . From an operational point of view, it coincides with the absolute value of the change in the reservoir filling volume that occurred in the interval of time ΔT and due to sedimentation. The loss of reservoir volume can therefore be estimated through: The percentage loss of reservoir capacity GI is important to understand the current health of the reservoir. It expresses the percentage occupied by sediments compared to the original capacity of the reservoir. In other words, it gives the loss percentage of total reservoir capacity, compared to the year of construction of the dam, defined as follows: where V orig indicates the capacity of the reservoir without any form of silt, which is relative to the year of construction. The GI parameter was considered a representative value for each reservoir of the degree of total sedimentation, obtained from the sum of the GI values obtained considering the single time intervals between subsequent bathymetric surveys.
The annual depletion rate (TI) is the ratio between the percentage loss of reservoir capacity and the number of years that elapses in an observation time interval. It is calculated as: The trap efficiency (TE) is the proportion of the incoming sediment trapped or deposited in a reservoir. Owing to the limited dataset, we followed the equation proposed by Brown (1943), which is simple because it uses only two parameters, yet is still practical.
where V refers to the reservoir volume and A is the catchment area. D is a parameter dictated by the reservoir characteristics and varies in the range of 0.046-1, with a mean of 0.1. We utilized TE index to evaluate the performance of the reservoir to retain the sediments between the period of the start of operation of the dam and the last bathymetry information available for the same. From the sediment volume, obtained as indicated in Eq. (1), it was possible to estimate the volume of sediments that accumulates on average every year upstream of the dam, in a given time interval ΔT. It is defined as the annual sediment input or "Sediment Yield" (SY), in m 3 /y and calculated as: By dividing this quantity by the catchment area, the specific contribution is obtained of annual sediment or "Specific Sediment Yield" (SSY) per unit of catchment area A, expressed in m 3 /(km 2 ⋅y): The analysis of each reservoir based on the yearly dataset available was carried out. Negative values were observed for some inter-year analyses of a few reservoirs (Poglia, Ceppo Morelli, Mignano and Fontanaluccia) while calculating the parameter concerning volume lost, loss of capacity, and annual depletion rate. This is due to the desiltation activities carried out in that reservoir prior to that year's bathymetric survey. In our assessment sediment yield excludes the amount of sediment that passed downstream of the reservoir. Therefore, the sediment yield analysis could be biased to some extent since some of these reservoirs have impoundments upstream, trapping some of the sediments and the exact volume of sediments flushed/dredged during the desilting activities are not available.
Using the above equations, we estimate the year at which half of the reservoir (t 1/2 ) will be infilled with sediments, also referred to as half-life year: The empirical model RUSLE (Revised Universal Soil Loss equation) was used to understand the reservoir sedimentation across the Italian regions. The unavailability of sufficient field data related to gross erosion guided the choice of the model. The RUSLE model (Renard et al., 1991) is formulated as below: ; LS is the slope length factor (dimensionless coefficient); C is the coverage factor (dimensionless coefficient, variable between 0 and 1); and P is the factor of erosion control practices (dimensionless coefficient, variable between 0 and 1). Various erosion models are available in the literature (Borrelli et al., 2018;Pandey et al., 2016). The choice of using RUSLE stems from a) popularity of this method due to the simplistic approach, b) applicability for hillsides or wide catchments due to inter-rill and rill erosion processes and c) appreciable performance in Italy (Covelli et al., 2020;Van Rompaey et al., 2005;Vente et al., 2006). The drawback associated with the RUSLE model is that it does not account properly the erosion due to tillage, landslide, gullying and river bank (Borrelli et al., 2018). Even though other models could perform better than RUSLE, they require a long-term sedimentation records dataset. The product of the listed factors excluding R (rain erosivity) provides a measure of the resistance of the environment to erosion. Over the last few decades, the Geographic Information System (GIS) has been widely used to manage the spatial distribution of land use/cover, vegetation cover, and extract automatically catchment characteristics and descriptors from the digital terrain model (DTM). Additional information such as slope, exposure, drainage area, soil composition or conductivity hydraulics assigned to the stratigraphy and others can be automatically imported into the model using remote sensing and GIS techniques. Applying these techniques allows determining the estimate of soil erosion and its spatial distribution with reasonable accuracy over multiple areas. The estimate of soil erosion was obtained by applying the RUSLE in the GIS platform (Arc GIS), which effectively manages the considerable amount of data required by the model: rainfall, use of soil, slope morphology etc.
The dataset used to implement the RUSLE model in GIS is obtained from the European Soil Data Centre (ESDAC) (Panagos et al., 2012. The RUSLE model was applied to the fifty catchments corresponding to the reservoirs previously analyzed. In the GIS environment, each of the factors of the RUSLE formula represents a layer of information. These are combined according to the RUSLE formula to calculate soil loss as shown in Fig. A2. The Sediment Delivery Ratio (SDR) concept was originally introduced by Brown in 1950 to estimate the sediment load (Wu et al., 2018). SDR (dimensionless) is an important indicator for studying the relationship between on-site soil erosion and downstream sediment yield. It can be used directly to indicate the amount of sediment eroded from slopes. SDR was calculated for the catchments under study, along with the parameters (variables independent or explanatory) that may influence the SDR due to their relationship with the process of erosion, transport, and deposition in the catchment. SDR is defined as: where: SSY (Specific Sediment Yield) represents the annual average sediment load per unit of area; GE (Gross Erosion) is the annual average soil erosion over the area considered. Eq. (9) also utilizes the bulk density ρ b to convert the volume of sediments stored (SSY) in the reservoir to its equivalent mass. The value of bulk density was derived individually for each catchment separately. The average dry bulk density for all the fifty case studies is ρ b = 1.04t/m 3 , which is of the order of magnitude of some values found in literature: an average value of 1.8 g/cm 3 bulk density for the soil around the Italian territory (Carnicelli and Costantini, 2013); while a value of 0.865 t/m 3 was adopted in a study of forty Italian river catchments considered by Van Rompaey et al. (2005). Many factors influence the SDR and consequently regulate the processes of erosion, transport, and sedimentation in the reservoir. In the present study, we wanted to establish an empirical relationship between SDR and the representative variables of climatic, morphological, and hydrological catchment characteristics, and land use from multiparametric statistical regressions. Among the climatological/hydrological variables were taken into consideration are: a) Rainfall erosivity factor R (considered from RUSLE model input); b) The average annual precipitation, assuming the average annual precipitation value measured at the station closest to the dam, as representative of the average annual precipitation over the whole catchment (obtained from https://www.isprambiente.gov.it/it/banche-dati); c) Catchment area A (obtained from catchment characteristics); d) Average elevation E (from DTM); e) Average slope of the catchment (from DTM); f) LS factor "Length Slope factor" (considered in the RUSLE model); g) Drainage density, that is the ratio between the total length of the watercourse within the catchment and the catchment area (obtained from catchment characteristics); h) Relief Length Ratio: that is the ratio between the difference in height between the point at the maximum elevation in the catchment and that at minimum elevation, and the catchment area (obtained from catchment characteristics and DTM); i) Circularity ratio: defined as the ratio between the catchment area A and the area of the circle of equal to the catchment perimeter P: R c = 4πA P 2 (obtained from catchment characteristics); j) Uniformity ratio or compactness factor: it is the ratio between the catchment perimeter P and the circumference of the circle of area A: R u = P 2 ̅̅̅̅ ̅ πA √ (obtained from catchment characteristics); k) Hypsometric integral H si : defined as the ratio between the height difference between the average elevation and minimum elevation of the catchment, and the difference in height between the maximum and minimum elevation of the catchment: H si = Hmean− Hmin Hmax− Hmin (obtained from DTM); l) Slope of the hypsographic curve P 60 calculated as the slope of the line passing through the point at maximum elevation and the point at relative elevation 60% of the hypsographic curve (obtained from DTM); m) "C factor" land cover considered in the RUSLE model (considered from RUSLE model input). All these thirteen parameters considered above were calculated for each catchment under study using ArcGIS© using the DTM catchment characteristics and the RUSLE model input. These parameters will be used in multi regression equations (see section 4.2) to represent the pattern of reservoir sedimentation.

Sediment yield assessment of some Italian reservoirs
We report in Table 1 the sediment yield, the loss of capacity and the half-life year of the reservoir in terms of infilling of sediments for the fifty Italian reservoirs under consideration.
The period of observation considered for each reservoir corresponds to the time interval (number of years) between the year of construction and the year of the last bathymetric survey available. The mean value of the observation period of all dams considered is 69 years. From Table 1, the average value of GI for all dams turns out to be equal to 38%. In some dams, a very high GI is observed, in particular, 12% of reservoirs show GI ≥ 80%, which means that 12% of the dams observed have already run out 80% of their initial volume, according to the data available from the last survey (see Fig. A3). Between these, there are also the dams in which GI = 100% has been reached.
Following the classification criterion by ITCOLD (2009) only two dams appear to have a loss of reservoir capacity of less than 5% and therefore they can be considered without significant reservoir sedimentation. For the rest of the reservoir (96% of the total sample of 50 dams), GI is greater than 5% and therefore they are considered with significant reservoir sedimentation. It is interesting to note that, if we want to identify two homogeneous macro-areas, specifically that of the Alps and the Apennines, the percentage loss of reservoir capacity is on average higher in the Apennine macro-area (GI = 43%) than that observed in the Alpine macro-areas (GI = 33%). Also, the only two dams examined with GI <5% fall within the latter macro-area (see Table 2). Carrying a t-test on the two samples of GI, the difference observed in percentage loss of storage capacity between the two regions is not statistically significant (p > 0.05).
But the difference attributed to the two macro-areas is also supported by the similar findings available in literature based on soil erosion studies in Italian rivers (Harris et al., 2008;Marchetti, 2002;Winterberg and Willett, 2019). A hypothetical explanation of the different GI in the two macro-areas could lie in the different geomorphological characteristics of these areas. The Alpine chain consists mainly of metamorphic and igneous rocks, while that of Apennine is sedimentary rocks of turbiditic and carbonate type (Argnani et al., 2004). Also, note that the percentage loss of reservoir capacity is generally small for reservoirs at higher altitudes with smaller catchment areas and is usually characterized by low erodible soils. The same is high for reservoirs at lower altitudes and with basins progressively larger and characterized by more erodible soils. The presence of a greater number of reservoirs of the second type within the sample analyzed in the macro-area Apennine could be another hypothetical explanation. The TI values observed for the dams examined, are aligned with the median volume lost due to sedimentation of 0.41%, reported by Marziali et al. (2017) for twenty reservoirs in Italy. A value of TI = 1% indicates that every year a hundredth of the original reservoir volume is lost by sedimentation and therefore conversely, if the TI remains constant over time, the dam will lose its entire storage capacity over the 100 years of operation (ITCOLD, 2009). It is evident that the higher TI, the more quickly the dam loses its storage capacity due to filling. On the other hand, the higher TI of a given reservoir, also an effective indicator of the behavior of the catchment area concerning the sediment output and the corresponding river sediment transport. Reservoirs in which values of TI ≥ 2% are observed underlie the fact that this catchment shows a high production of sediments a. Consequently, reservoirs in these catchments may have an operational life of ≤50 years. None of the fifty dams examined has TI > 2%. Using the TI obtained for each dam, we can predict the useful life remaining for each dam along with the half-life year, t 1/2 . From Table 1, it can be seen that most of the dams will exhaust the total storage capacity over a time span greater than 80 years, but also some of the dams will exhaust it in shorter time intervals (see Fig. A4).
In percentage terms, the analysis leads to the following results. It is interesting to point out that 4% have already exhausted the total volume; 14% have will run out in 20 years; 6% in a time interval between 20 and 40 years; 8% between 40 and 60 years; 6% between 60 and 80 years; and 62% of the dams examined will take more than 80 years to exhaust the total storage capacity. The above analysis is further supported by observed values of specific sediment yield (SSY), as shown in Table 1. The values found to range from as low as 17.32 m 3 /(km 2 ⋅year) to a maximum of 4168.31 m 3 /(km 2 ⋅year); the mean value observed for all the dams examined is even 451 m 3 /(km 2 ⋅year). A dependence of SSY on the catchment area has often been observed in the literature (Bachiller et al., 2019;De Vente et al., 2007;Dendy et al., 2013;Verstraeten et al., 2003). In comparison to the study of forty Italian reservoirs (Van Rompaey et al., 2005) that reported the mean annual sediment yield ranging from 8 m 3 /(km 2 ⋅year) to 830 m 3 /(km 2 ⋅year), these estimates were lower than ours.
Based on the information obtained in Table 1, Fig. 2 shows the relationship between catchment area, specific sediment yield, average annual depletion rate, and reservoir volume for the fifty reservoirs E.R. Patro et al. Journal of Environmental Management 320 (2022) 115826 studied. The catchment area of the fifty dams examined varies from a minimum of 2.10 km 2 to a maximum of 2642.01 km 2 . With the increasing catchment area, a net decrease in the SSY is observed (see Fig. A5 and Table A2). While for the catchment area and the annual depletion rate (TI) shown in Fig. 2, the relationship observed, in this case, is inverse to the previous one, with the increase in catchment area the TI also increases. The reservoir volume for the reservoirs examined varies from a minimum of 0.09 Mm 3 to a maximum of 141 Mm 3 . The trend was also analyzed between the TI and the reservoir volume, a decreasing trend was observed with TI as the reservoir capacity increased. A similar trend was also observed in studies conducted in the USA comprising around 1000 reservoirs (see Dendy and Champion, 1978;Dendy et al., 2013). Therefore, the rate of sediment yield is usually lower for larger catchments (Birkinshaw and Bathurst, 2006). In comparison to the global studies, the sediment yield should be higher for small catchments, our results reciprocate the same.
The results support that with the increase in reservoir size the depletion rate decreases. For the fifty reservoirs examined, we also evaluated the percentage change (decrease) in the trap efficiency TE in the current situation with respect to the beginning of operation in Table 1. Ten reservoirs (20%) had a significant decrease in TE greater than − 50%, nine reservoirs (18%) have a percentage decrease in TE in the range of − 25 to − 50%, while only 31 reservoirs (62%) had a decrease in TE less than − 25%, From Table A1, we have calculated that at the start of dams operation, out of the fifty reservoirs, sixteen (32%) have <50% TE, eight (16%) have 50-70% TE, seven (14%) have 70-90% TE, and nineteen (38%) have ≥90% TE. In comparing with the present situation, we see that twenty-three (46%) have <50% TE, four (8%) have 50-70% TE, five (10%) have 70-90% TE, and eighteen (36%) have ≥90% TE. The TE estimation could be improved further if the flow and sediment data are available (Najafi et al., 2021;Rahmati et al., 2020).

Statistical analysis of reservoir sedimentation
The soil erosion map of the whole Italian territory was obtained based on the RUSLE model. From this, it was possible to obtain the average value of annual soil loss over the area of each catchment for the fifty reservoirs considered in our study (see Table A3). In Eq. (9), we used the values of SSY reported in Table 1 derived from Eq. (6). The values of GE are obtained from the application of the RUSLE model and reported in Table A3. The bulk density ρ b for each catchment was retrieved from ESDAC. Therefore, these values were used to convert the SSY measurements from [m 3 /(km 2 ⋅year)] to t/(ha ⋅year). The results are shown in Table A3.
The fifty reservoirs were classified into two groups according to the geographical location of the respective catchments: one group includes Alpine catchments (located in northern Italy), and the other group includes Apennine catchments (located in central-southern Italy and islands). Twenty catchments belong to the first group and twenty-four to the second one. Six case studies were not included in these groups because have significant anthropogenic interference: Casteldoria, Poglia, Ponte Cola, Dazare, Panigai, and Fusino. Table 3 provides the classification of the fifty case studies in the two groups.
Multiple linear regression in the log-log plane was used to examine the relationship between SDR and the thirteen parameters. Initially, all thirteen explanatory variables were used to fit the model regression and all variables with a non-significant correlation with SDR were removed from the analysis. This skimming helps to increase the quality of the model, which is inversely related to the number of independent variables (Bachiller et al., 2019). The level of significance was set at 0.05. The values of the sample regression coefficients are calculated with the least-squares method. The F-test for the significance of regression was also carried out. Overall F-test was found to be statistically significant for both the reservoir groups (Alpine and Apennine) at significant levels of 0.01 and 0.05 while considering all reservoirs only at a significant level equal to 0.05. In addition, when considering all reservoirs together the coefficients (all except for the coefficient of A) failed to be individually significant at the mentioned levels. This was not the case when the test was done separately for the Alpine and Apennine reservoirs.
For each regression, different indicators are calculated: the coefficient of determination (R 2 ), the Root Mean Square Error (RMSE) and the Mean Absolute Error (MAE) to quantify the goodness of the model. Once the regression analysis was completed, the Jackknife resampling technique was also applied (Kottegoda and Rosso, 2008). The coefficients obtained from the Jackknife procedure were used to estimate the values of SDR and compare them with the corresponding RUSLE based SDR values, both in the calibration phase and in the validation phase. Grouping the reservoir samples into two macro-groups was also preferred because the performance of the regression considering all the reservoirs was not satisfactory. In the case of the Alpine group, eleven reservoirs were selected to calibrate the model and the remaining nine to validate it, while in the case of the Apennine group, twelve reservoirs in the calibration phase and twelve reservoirs in the validation phase were used. In Table 3, data are distinguished for the calibration and validation phases, respectively for the Alpine and Apennine groups. Of the thirteen parameters, four variables showed a significant relation with SDR, for the two groups: the slope of the hypsographic curve P 60 [-]; the circularity ratio R c [-]; the average elevation of the catchment E [m]; the catchment area subtended by the reservoir A [km 2 ]. Table 4 gives the Jackknife estimates of coefficients and intercept of regressions both for Alpine and Apennine groups. In contrast, Table 5 provides the values of indicators used to quantify the goodness of the regressions. Figs. 3 and 4 show the comparison between SDR evaluated with the RUSLE model and that one calculated through the multi regression, respectively for Alpine and Apennine groups.

Table 3
Values of SDR and reservoirs used in the calibration phase and those used in the validation phase, are distinguished in Alpine and Apennine groups (the reservoir locations and names are given in Fig. 1  The results obtained show a clear distinction between the catchments located in the Alpine geographical area and those located in the Apennine geographic area, in response to complex erosion phenomena, transport and deposition that determines SDR. What emerges is that the regression model, although developed on an empirical basis, seems to capture well the behavior of the catchments located in the Alps while the same cannot be said for the catchments located in the Apennines. In addition to the model results, the graphical interpretation of the variation of the hypsographic curves of the catchments analyzed for both the Alps and the Apennines, concerning the variation of SDR, confirms the different behavior of the two groups of catchments. Fig. 5, shows the hypsographic curves of the catchments of the Alps and those of the catchments of the Apennines. It is possible to observe that the hypsographic curves in the case of the Alps seem to be qualitatively arranged homogeneously for the range of values of SDR identified (Fig. 5a), while in the case of the Apennines the variability of the SDR does not correspond to a homogeneous distribution of the hypsographic curves (Fig. 5b).

Discussion
The Italian peninsula has a long-standing history of hydrological and geological instabilities due to geological and geomorphological characteristics of the land, in addition to climatic forcing which entails a very high exposure to the risk of floods and landslides (Bosco et al., 2015;Panagos et al., 2012;Terranova et al., 2009). In the present work, we studied fifty Italian reservoirs extended to a wider geographic area covering, the Alpine and Apennines mountain ridge with different hydrological, geological and relief conditions observable from north to south and the islands of Italy. Many Alpine catchments are considered to have steep slopes and perennial snow covers. The Apennines catchments Table 4 Jackknife estimation of coefficients and intercept of the regression model log(SDR) = a⋅log(P 60 ) + b⋅log(R c ) + c⋅log(E) + d⋅log(A) +intercept for both Alpine and Apennine groups.   E.R. Patro et al. Journal of Environmental Management 320 (2022) 115826 have low slopes and relatively higher anthropization. From the analysis of the fifty reservoirs carried out six reservoirs (Quarto, Colombara, Ceppo Morelli, Fusino, Vodo and Valle di Cadore) are already in a very critical situation and have already completely silted. Further six reservoirs (Boschi, Alanno, Tavernelle, La Penna, San Lazzaro and Ozola) could also lose their entire storage capacity within the next three decades if proper remedial measures are not carried out. Of these twelve reservoirs, eight belong to the Apennine region particularly Marche and Emilia Romagna with a percentage reduction of trap efficiency in tune − 30% to − 88% compared to the beginning of dam operation. While the remaining four belong to the Alpine chain with a percentage reduction in trapping efficiency greater than − 50%. The evaluation of trap efficiency TE highlights the need of monitoring the health of reservoirs and further studies for specific reservoirs to help optimize the sediment collection from both technical and economical points of view. The reservoirs which underwent a drastic reduction in TE need further investigation to prioritize and develop efficient sediment control plans. Although many reservoirs have successfully trapped sediments, they could be more effective and efficient.
With many reservoirs reaching the end of their original design life, sedimentation is becoming an increasingly important issue in reservoir management. Sediment yield modelling has been increasingly used to evaluate the impacts of various variables controlling sediment dynamics at the catchment scale. An estimation of eroded soil is vital for hydraulics, hydrology, ecosystem, and agricultural studies. Historical insitu reservoir and sediment data are not publicly available due to the associated confidentiality with it. In this work TI, TE, SDR, SY and SSY, the values were calculated as an average of the years available and excluding negative values. The sediment yield analysis could be biased because some of these reservoirs have impoundments upstream of it, trapping some of the sediments and the exact volume of sediments flushed/dredged during the desilting activities are not available. Many studies support the existence of an inverse relationship between the SSY measured at the catchment's closing section and the catchment area. This relationship has often been directly employed to predict the SSY both regionally and globally. There are also cases in which this correlation is instead positive and positive and negative in combination or very poor (Bachiller et al., 2019). This study should be used as the starting point for exploring various variables affecting reservoir sedimentation since they are associated with uncertainties that are not easy to define. The risk of reservoir sedimentation scales directly with the size of the directly controlled catchment variable.
The amount of sediments reaching annually to the catchment outlet due to soil erosion is influenced by many reasons such as the length of overland flow and catchment area (Boyce, 1975;Hrissanthou, 2011). For proper sediment management, it is important to understand the sediment connectivity which describes the spatiotemporal variation of sediment transport across the catchment (Najafi et al., 2021). This connectivity consists primarily of two types functional and physical. A significant amount of research is being carried out in the last two decades to study the structural and functional sediment connectivity (Bracken et al., 2015;López-Vicente and Ben-Salem, 2019;Schmitt et al., 2016;Zingaro et al., 2019). This work gives a brief overview of the structural sediment connectivity and the issue of reservoir sedimentation, derived mostly from the physical characteristics of catchment and reservoir. The sediment transfer processes from sediment sources to sinks have rarely been considered in operational water studies because they imply some major modelling challenges in magnitude, transport time, and delivered grain size (Schmitt et al., 2016).
RUSLE was developed as a tool for long-term soil loss calculation in agricultural areas, therefore, the approach is not expected to be well suitable for erosion assessment in mountain catchments (Borrelli et al., 2016;Renard et al., 1991). One of the most discussed limitations of RUSLE is its inability to represent mass movements (to landslides, avalanches, or soil displacement by snowmelt), gully erosion and stream bank erosion (Gianinetto et al., 2019). As a result, phenomena such as sediment deposition and sediment routing are not considered (Alewell et al., 2019). Apart from this RUSLE suffers from other limitations since it does not takes into account such as a) seasonal dynamics of precipitation by considering both rainfall and snow, b) effect of snow cover dynamics c) the seasonal dynamics of vegetated areas d limited interactions between input factors (Geitner et al., 2021;Phinzi and Ngetar, 2019). The error margins of the RUSLE model result can fluctuate considerably and this work should be considered comparative and not absolute since the application of RUSLE over the alpine territory could present huge uncertainties (Berteni and Grossi, 2020).
Improved estimation of factors in RUSLE based erosion model could reduce the uncertainty up to some extent in Alpine catchments. Gianinetto et al. (2019) used the D-RUSLE (Dynamic Revised Universal Soil Loss Equation) model for an Alpine catchment which considers the snow cover and land cover dynamics (improved C-factor). They found reasonable agreement with the JRC-RUSLE erosion estimates (Panagos E.R. Patro et al. Journal of Environmental Management 320 (2022) 115826 et al., 2015 and added the advantage of seasonal erosion evaluation. Kaffas et al. (2021) used a USLE-based model for an Alpine basin with a modified LS-factor to account for rock in the soil surface which is non-erodible. They found model validation to be improved by 10%. Improving the factors to reduce the uncertainty in the erosion models is often difficult due to the need for extensive field/satellite measurements.
Nevertheless, even if sediment yield at the basin scale is a product of all sediment producing processes Geitner et al. (2021) argued that simulating all the processes going on in a basin -both Apennine and Alpine one-is not possible and models that try to consider different processes often become too complex to be correctly applied in real cases. Despite these shortcomings and the lack of data to validate the models in many mountain areas, the RUSLE model can be regarded as an effective tool to estimate soil erosion and its distribution in the Alpine and Apennine regions of Italy.
SDR model developed from the RUSLE model is based on an empirical relationship relating the SDR with morphological characteristics of the catchments and implies numerous uncertainties including temporal discontinuity and spatial variability characterizing the land cover, climatic, hydrological, and geomorphological variables involved. The general limitation of this approach is dictated by the local influence of climatic and geomorphological characteristics which make the developed relationship unsuitable in geographical areas different from the area where the relationship was calibrated. The developed SDR model performed better than SDR SIM (Spatially Invariant Model) developed by Diodato and Grauso (2009) based on 25 Italian river basins. The SDR SIM as shown in Eq. (10), accounted for the roles played both by catchment morphology (Catchment area A, average elevation E, mean catchment slope Slope) and the annual average rainfall amount (P) (Diodato and Grauso, 2009).
The performance of this model on our dataset exhibited R 2 of 0.52 and 0.008 for the Alpine and Apennine groups respectively. Researchers have produced many empirical SDR models, with each trying to choose the best formulation based on the data availability.
By 2050, the mean soil loss rates for European Union are projected to increase by 13-22.5% for three representative concentration pathways RCP (2.6, 4.5 and 8.5) scenarios compared to the 2016 baseline (Panagos et al., 2021). Based on the RUSLE model coupled with future rainfall erosivity, Panagos et al. (2021) projected that the mean soil loss for Italy will not change much in the period 2016-2050. The change (%) of soil losses in the period 2016-2050 for Italy is < +5% under different RCPs. Although the projected increase in soil erosion for Italy is not that substantial, such projections are always associated with significant uncertainties as they are based on long-term climate model projections across different emissions pathways etc. Furthermore, their trend is not easy to foresee and hence estimation and prediction for future storage capacity get more difficult. Constructions of new check dams and interventions of removing sediments are key actions that must be counted. But to have a resilient dam with effective mitigation procedures it is essential to have a proper monitoring system. The results from our work could be further complemented by combining them with remote sensing technologies. Optical satellite images, synthetic aperture radar images and satellite altimeter data could play a crucial role in investigating the spatial and temporal changes in reservoir sedimentation. Using remote sensing techniques comes with drawbacks such as coverage, revisit frequency, and resolution; still, its advantages are enormous because the proper technique can be utilized for near-real-time monitoring of reservoirs.

Italian legislation framework
The results of this analysis must not be fully understood without referring to the (past and present) legislation context. Based on the latest Italian dam regulatory framework from 2006 onwards (refer to Appendix B and https://trasparenza.mit.gov.it/), at the end of the concession/lease of the reservoir, it must be returned to the State with a useful reservoir completely free of sediments. This law constitutes an additional burden for the dam operators because the previous regulations before 2004 said that the dams must be returned upon completion of the concession period in a state of regular operation. It should be noted that the discharge requirement at the end of the concession should meet the conditions requested by the public administration, which is an obligation set for all dams, regardless of the economic and environmental consequences that could arise.
The costs associated with the loss of useful storage capacity derive basically from two contributions: one is given by the non-utilization of the water resource and the other is connected to the recovery of reservoir capacity. In the case of reservoirs allocated for water usage for agriculture and human needs, the reduction of reservoir volume caused by sediments translates into a directly proportional loss of consumption. For hydroelectric reservoirs, on the other hand, it is necessary to distinguish between run-of-river and storage systems. In the first case, the reduction of the reservoir volume by sediments does not determine consequences for hydroelectric production; while in the second case, the loss of reservoir does not cause an appreciable decrease in overall energy production; rather the loss manifests itself in the production of valuable energy as it reduces the capacity of regulation of the reservoir.
The results of this analysis must not be fully understood without referring to the (past and present) legislation context. Based on the latest Italian dam regulatory framework from 2006 onwards (https://tras parenza.mit.gov.it/), at the end of the concession/lease of the reservoir, it must be returned to the State with the useful reservoir completely free of sediments. Often the recovery of the reservoir capacity involves a reduction in water allocation. It should be noted that often in the reservoirs that have created situations of environmental/landscape/touristic value with the birth of ecosystems of particular value that would be irremediably lost or seriously damaged by removing sediments. Therefore, it is believed that existing regulations do not always represent the best solution; however, it would be appropriate to evaluate each case with the necessary flexibility. It is also worth noting that the economic burden necessary to comply with the complete desilting requirement is often unsustainable with the real size of the related business. This means that many dam operators will face financial difficulties.

Conclusions
Considering the increasing significance of the reservoir's storage capacity in future climatic conditions, many important concerns regarding the management of sediments may arise. Based on the fifty reservoirs studied, almost 34% have reached their half-life year, and this percentage will become 50% by 2050. The mean annual sediment yield varied between 17-4000 m 3 /km 2 . year for the reservoirs studied. We report the maximum volume that can be stored nowadays within the reservoir, and the volume that could be stored during the inception of the dam (i.e., reservoir practically without sediments). According to ambiguities in Italian sediment management legislation (https://tras parenza.mit.gov.it/), and the lack of proper awareness of this issue, the dam operators in Italy have not taken into adequate consideration the reservoir management strategies to recover the lost storage capacity for many decades. Data availability and calibration requirements often determine the researcher's choice of one model or another. This work, therefore, should be treated as an investigation of the evolution of reservoir sedimentation in Italy and its interpretation in terms of the need for a comprehensive strategy and legislation at a national level. Then we have modelled the SDR, bearing in mind that the studies on E.R. Patro et al. SDR are still in a preliminary stage and there is no established model available yet to standardize the results. We used statistical multi regression analysis and its performance was tested for Alpine and Apennine reservoirs separately. The performance of the regression model was satisfactory for the Alpine group and weak for the Apennine group. The results are a starting point and highlight the lack of data available for such analysis to be conducted with certainty. The mechanisms contributing to reservoir sedimentation are well known but their physical contents are not yet fully apprehended theoretically.
The subject of reservoir sedimentation is getting a decisive push from various stakeholders. The recent 2022 drought in Italy showed the necessity of healthy reservoirs; so as to store enough water for the dry spell, particularly in rivers with high hydrologic variability seasonally and/or annually such as the Po River. A decisive restart of sediment management activities to pursue the best possible way to safeguard dams and respective reservoirs. It is also necessary that the new sediment management regulations contain criteria and guidelines capable of guiding the design and execution of sediment management operations as well as environmental monitoring of the same. However, one issue remains challenging: climate-related uncertainties come on top of other uncertainty sources, which affect the results of risk analysis models and their effectiveness. The effect of global warming and climate change on the sustainability of reservoirs needs assessment as well as the role of the reservoirs in the carbon cycle. In many regions, reservoirs will become vital infrastructures for climate change adaptation. This represents a major roadblock to adaptive decision-making and requires dam operators to adapt their standard practices. The economic and societal importance of water storage makes sedimentation in reservoirs an active and expanding field of research, needing an appropriate knowledge transfer from researchers to entities involved in the issue and vice-versa.

Availability of data
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

B.1 Italian National and Regional legislations
To properly frame the sedimentation issue within reservoirs, it is important to recall the regulatory framework. In Italy, the evolution of the regulatory framework was quite articulated, starting in the seventies of the last century until today. Below is given the list in chronological order of the main national regulations that have been gradually issued and subsequently amended. The overview is based on the documents available from the website of the Italian Ministry of Infrastructure and Transport (http://dgdighe.mit.gov.it/leg) and ITCOLD (https://www.itcold.it/bollettini-itcold/). The issuing of the Decree of June 30, 2004 by the Ministry of the Environment and Land Protection, defined the criteria to be followed by dam managers for the drafting of the reservoir management project according to Art. 40 of Legislative Decree 152/99 and the subsequent Decree of December 29, 2003 n.391, concerning the benchmarks of classification of the ecological status of lake/reservoir water bodies. The management of the sediments for the maintenance of the useful reservoir capacity is the subject of art. 114 of the Legislative Decree.152/06 and subsequent amendments. Article 114 provides for the preparation of a management project for each reservoir, prepared based on the criteria established by decree of the Ministry of the Environment and the Protection of the Territory and the Sea, in agreement with the Ministry of Infrastructure and Transport and in agreement with the Ministry of productive activities and with that of agricultural and forestry policies. At present, this decree has not yet been issued and therefore reference is still made to the Ministerial Decree 30/06/04 as required by Legislative Decree 152/99. Currently, also article 43 of Decree Law 201/2011 converted in Law 214/2011 provided for the obligation to draw up the project for the management of large dams by December 31, 2012, and the identification by the competent authorities (Ministry of Infrastructure and Transport, Ministry the environment and the protection of the territory and the sea and Regions) of the dams for which the removal of accumulated sediments, is a priority. At present, some Italian Regions have issued Regional Regulations for dams not subjected to the provisions of Presidential Decree 01/11/59, n.1363, as required by article 1, paragraph 2 of the DM 30/06/04. These regulations for some specific issues are the only sources of information and technical indications. So far, the regions that have approved these regional regulations are the Autonomous Province of Trento (Water Protection Plan -Regulations of implementation, approved with DGR of 30/12/04); Autonomous Province of Bolzano (decree of the president of the province January 21, 2008, n. 6); Veneto (Council Resolution no.138 of 31/01/06); Valle d'Aosta (Water Protection Plan, attachment F, approved with Resolution of the Regional Council n • 1788/XII of 08/02/ 06); Piedmont (DPGR 29/01/08, n.1/R); Sardegna (resolution of the regional council March 4, 2008 n. 13/12); Tuscany (Artificial reservoirs -Elements for sustainable management, 2009); the province of Lucca (Guidelines for the preparation of management projects drawn up according to art. 114 of Legislative Decree 152/06 following Ministerial Decree 269/04); Abruzzo (DGR 242 of April 11, 2011 andLR June 27, 2013 N.18); Sicily (DDG n. 710 of May 07, 2012). In Lombardy, Lazio, and Friuli Venezia Giulia, guidelines are drafted but not yet published. In Lombardy, pilot experiences conducted in the Region were published in the Quaderni della Ricerca series n. 90, while in Basilicata the experiences carried out were published in the Quaderno n. 6 of the AdB Basilicata and the Studies and Research series -vol. 4. In Umbria exists the Regional Law 25 of December 10, 2009 which excludes the small reservoirs from the obligation of the present management project.
The regulatory frameworks contain cognitive surveys carried out on the hydrographic basin, on the chemical-physical characteristics, on the water quality of the reservoir, and sediments at the bottom of the reservoir. Based on this information, the framework defines a forecast of the sediment removal activities aimed at the one hand at maintaining functional efficiency over time and on the other hand to maintain or recover any useful storage capacity of the reservoir lost due to sedimentation. These regulations contain the specifications of the measures of prevention and protection of the aquatic ecosystem, of fishing activities, and of water resources stored and released downstream of the dam, which must be implemented during the operations themselves. It specifies, for example, a) the chemical-physical characterization of the sediments and water and sediment, b) what are its potential destinations in case of removal; c) in the case of flushing or bleaching set, the limits of suspended solids that must not be exceeded in the downstream water body, d) establish the measures for the mitigation and prevention, e) the monitoring plan, f) the authorization process of the reservoir management.
Based on the latest regulatory framework (2004), at the end of the concession/lease of the reservoir, it must be returned to the State with the useful reservoir completely free of sediments. The latest law constitutes an additional burden for the dam operators, until then they were required to comply with RD n. 1775/1933, according to which the dams must be returned upon completion of the concession "in a state of regular operation". It should be noted that the discharge requirement at the end of the concession should meet the conditions requested by the public administration, which is an obligation set for all dams, regardless of the economic and environmental consequences that could arise.
It should be noted that often in the reservoirs that have created situations of environmental/landscape/touristic value with the birth of ecosystems of particular value that would be irremediably lost or seriously damaged by removing sediments. It is therefore believed that such a rule does not always represent the best solution; however, it would be appropriate to evaluate each case with the necessary flexibility. It is also worth noting that the economic burden necessary to comply with the complete desilting requirement is often unsustainable about the real size of the related business. This means that many dam operators will face financial difficulty.
The costs associated with the loss of useful storage capacity derive basically from two contributions: one is given by the non-utilization of the water resource and the other is connected to the recovery of reservoir capacity. It is considered useful to highlight that also the recovery of the reservoir capacity involves, in most cases, a lack of use of water allocation. In the case of reservoirs allocated for water usage for agriculture and human needs, the reduction of reservoir volume caused by sediment translates into a directly proportional loss of consumption. For hydroelectric reservoirs, on the other hand, it is necessary first to distinguish between run-of-river and storage systems. In the first case, the reduction of the reservoir volume by sediment does not determine consequences for hydroelectric production; while in the second case, the loss of reservoir does not cause an appreciable decrease in overall energy production; rather the loss manifests itself in the production of valuable energy as it reduces the capacity of regulation of the reservoir.