Assessing the Impact of Often Overlooked Snowfall on the Hydrological Balance of Apennine Mountain Aquifers in Central Italy

Abstract

The accurate knowledge of groundwater availability and its variations is crucial for sustainable groundwater management; in this framework, the water balance is a useful tool to assess the availability of water resources. Currently, the management authority needs a more precise evaluation of groundwater availability to face the rising freshwater demand. In this work, water balance has been determined for the main aquifers in the central Apennines (Italy)—over 2000 km² wide—and the calculated outflow was compared with springs’ discharge from the data. Inflow data were collected over a 6-year period, from 2018 to 2023, considering both rainfall and snow; the contribution of the snow melting has often been omitted or rarely considered as immediate liquid contribution in the previous works, where usually only liquid inflows from rain have been considered. The snow contribution has been properly evaluated from a recent network of snow gauges and included in the total precipitation for more accurate results. Indeed, for each aquifer, monthly inflow datasets from rain gauges have been interpolated inside the structure using the equations obtained from regression lines and then used for a water balance assessment. An initial comparison of water balances, estimated with and without snow data, demonstrates that neglecting the snow contribution can lead to an underestimation of infiltration values. A comparison between calculated outflows including the snow melt and the measured springs’ discharge has shown a good correspondence for each investigated aquifer.

1. Introduction

The evaluation of snowfall’s contribution to meteorological inputs for the hydrological balance is harder than that of rainfall. However, in Mediterranean and central European regions, its contribution is quantitatively comparable to that of rainfall, particularly in high-altitude mountain areas. Snowfall can also play a significant role in southern climates when at higher elevations, as seen in the Italian central and southern Apennines.

The difficulty in estimating the contribution of snow—both as an inflow and as a measure of snow accumulation on the ground—stems from two main challenges: technical–logistic and hydrological factors.

The technical and logistical challenges can be summarized in three main points: (i) The lack of high-altitude monitoring stations where snow contribution is significant. This is primarily due to logistical issues, making it difficult to install and, more important, maintain and power the necessary equipment. (ii) Technical difficulties in accurately measuring the amount of snow accumulated on the ground using simple equipment. Specifically, pluviometers or snow gauges require electrical resistance to melt the snow they collect, which often presents operational issues due to the need for a power supply and the extreme conditions in which they are placed. (iii) Similar challenges are encountered in the installation and maintenance of instruments used to monitor the thickness of the snow cover [1].

The hydrological challenges arise from the fact that snow measured by rain gauges is treated as liquid precipitation and is thus considered an immediate input to the hydrological budget. Actually, snow can accumulate on the ground, remain as a snowpack throughout the cold season, and only contribute to the hydrological input during the melting in the warmer months. Furthermore, the duration of snow cover varies depending on factors such as altitude, temperature, topography, and amount of snowfall [2].

The difficulties outlined above mean that snowfall contribution generally is not considered if there are no rain gauges in the study area, it is approximated and added to liquid precipitation if functioning precipitation gauges are present in the area or, rarely, it is considered a change in snow cover, limited to areas with snow cover monitoring.

With the advent of remote sensing [3,4] and satellite imagery, the estimation of snow’s contribution to the hydrological balance has improved, but it remains challenging due to the high cost of satellite surveys and the need for ground-based validation. Satellite images alone cannot provide reliable information on snow thickness, and the presence of vegetation complicates accurate assessment.

The central Apennines, despite its relatively southern location, is an area rich in water resources, both surface and groundwater [5,6,7,8,9], due to the presence of extensive carbonate aquifers and high elevations where snow contributes significantly. Recent national legislation in Italy and EU regulations, aimed at rationalizing and optimizing water resource use, emphasize the urgent need to define the hydrological balances of aquifers—particularly considering recent climate changes. These changes affect the hydrological balance both directly, through alterations in the distribution, intensity, and form of precipitation (rainfall and snowfall) [10,11,12], and indirectly, through changes in evapotranspiration rates due to rising temperatures [13,14,15,16].

Local significance studies have been recently published, typically focused on individual aquifers, and have aimed to assess the contribution of snowfall. Lorenzi et al. [17] focusing on the Gran Sasso aquifer, highlighted the role of snowmelt in aquifer recharge and examined the impact of temperature increases on the distribution of winter snowfall. Lorenzi et al. [18], again studying the Gran Sasso aquifer, evaluated the influence of snow cover on recharge using satellite imagery and natural isotope analysis of spring waters. Scozzafava and Tallini [19], again in relation to Gran Sasso, proposed modifications to Thornthwaite’s method to consider various aspects including snow melting. In Petitta et al. [20], the authors established a correlation between isotopic recharge rates and average snow cover in several central Apennine aquifers, using isotopic data from spring waters and satellite images. Nanni and Rusi [21] identified the correlation between snowpack thickness, its melt, and variations in flow and chemical parameters in the Majella’s base aquifer springs. Chiaudani et al. [22], using daily data, developed statistical relationships between snow cover thickness, flow rates, and the chemical parameters of the Verde Spring in the Majella aquifer. Di Curzio et al. [23] attempted to use meteorological radar data to estimate meteoric inflows, including snowfall, for the Majella aquifer, while Di Curzio et al. [24] applied similar methods to the central Apennine ridge along the Adriatic Sea, with preliminary results.

This study evaluates the hydrological balance of the central Apennines’ main aquifers, considering both rain and snow contributions from direct measurement. The objective is to quantify the role of snowmelt in recharging aquifers and its impact on the overall water balance on a regional scale using snow gauge measurements instead of indirect methods as in Lorenzi et al. [17,18] or in the other works previously mentioned.

The infiltration calculations, incorporating snowmelt data from a network of snow gauges (2018–2023), were compared with the previous literature, which typically excluded snow contributions. The study interpolated precipitation data using altitude-precipitation regression to include high-altitude areas lacking rain gauges. The method was applied to six carbonate aquifers (covering ~2000 km²), which are crucial for drinking water, hydroelectric, and irrigation purposes. This is the first attempt to objectively integrate snowfall into hydrological assessments for such a large groundwater resource area.

2. Materials and Methods

2.1. Study Area and Hydrogeological Overview

The study focuses on aquifers in the central Apennines which represent the main recharge area for the regional aquifers (Figure 1).

According to the data, these structures are highly permeable for fractures and karst processes with infiltration between 700 and 900 mm/year considering an average 1000 mm/year precipitation [25].

2.2. Hydrological Balance and Data Elaboration

The hydrological balance was calculated, for each aquifer, to estimate effective infiltration and to compare it with the literature; the equation used is

$$I=I_R(P-E)$$

(1)

where P stands for precipitation, including both rainfall and snow (if possible), E is evapotranspiration, I is the infiltration, and I_R indicates the Potential Infiltration Index (PIC).

This is the first attempt to incorporate snowfall into hydrological calculations for a large area with significant groundwater resources, based on objective data rather than indirect estimates. The underlying hypothesis assumes that the snowpack is not treated as solid water subject to evapotranspiration but rather as melted water. In this framework, despite low temperatures, it undergoes the same hydrological processes as liquid precipitation. Rainfall, temperature and snowfall databases are from the Hydrographic Service of the Abruzzo Region; for each aquifer, the rain gauges that recorded rainfall and/or temperature data from January 2018 to December 2023 were selected and monthly cumulated, so that a 6-year dataset is available, as suggested in the current Italian guidelines [26].

In addition to the rain gauges, eight stations of snow cover measurement (see

Table 1. Features of snow cover measuring stations.

Station Name	Altitude (m a.s.l.*)	Mean Annual Value (mm)
Campotosto Diga	1313	215
Caramanico Terme	805	92
L’Aquila, Campo Imperatore	2094	256
Ovindoli	1375	150
Pretoro, Passo Lanciano	1310	256
Roccaraso	1231	140
Scanno, Passo Godi	1542	212

Note: a.s.l.*: above sea level.

and Figure 1) were included to enhance the accuracy of inflow evaluations. These stations are equipped with a sonic radar, which records the return signal emitted by it and allows the thickness of the snow cover to be measured. They are located far from urban areas and vegetated areas to avoid topographic influence.

The snow data, from each station, were evaluated monthly over the six reference years. To ensure the accuracy of the recorded data, they were compared with the corresponding temperature data, enhancing reliability and filtering out any unreliable data. Snow data were only included if they fell within the temperature range of 0 °C to 3 °C, which typically represents the climatic conditions conducive to snowfall [27].

In Figure 2, the snow cover thickness data are presented as daily cumulative values for an entire month (February 2020), along with, as an example, the first week of the following month (March 2020, separated by the blue line). As can be observed, the processing of these data is quite complex. First, negative values (indicated by the blue circle in Figure 2), which resulted from clear measurement errors, were discarded. Subsequently, for calculating the monthly cumulative snow height, only the increments in snow thickness (green segments) were considered. The red segments, on the other hand, represent snowmelt and were excluded from the cumulative calculation, as they do not reflect an increase in snow thickness.

Lastly, the data were monthly cumulated and then added to the corresponding monthly rainfall using the relationship according to which 1 cm = 1 mm of rain [17].

The spatialization for each aquifer was executed using monthly regression lines that correlate rainfall with altitude (Figure 3). These equations were integrated with Digital Elevation Model (DEM) maps within the QGIS (Version 3.28.15) environment, resulting in monthly rainfall maps with a resolution of 10 m × 10 m.

This process has been repeated for temperature and only rainfall data.

For each station, evapotranspiration was calculated both monthly and yearly; Turc and Turc modified methods [30] were applied for annual evapotranspiration (${\mathrm{E}\mathrm{T}}_{\mathrm{r}}$), which was defined as

$${\mathrm{E}\mathrm{T}}_{\mathrm{r}}={\displaystyle \frac{P}{\sqrt{\left(0.9+\frac{P^2}{L^2}\right)}}}$$

(2)

where L is the evaporative potential of the atmosphere (300 + 25T + 0.05T³) and T is the mean yearly temperature of air (°C).

The Turc modified is based on Equation (2), but it considers L as (300 + 25T_P + 0.05T_P³), with $T_P=\sum {\displaystyle \frac{P_i}{T_i}}$, where P_i and T_i are the rainfall and air temperature values related to the ith month, respectively.

The Thornthwaite and Mather equation [31] provides a monthly evaluation of evapotranspiration (${\mathrm{E}\mathrm{T}}_{\mathrm{p}\mathrm{i}}$) using

$${\mathrm{E}\mathrm{T}}_{\mathrm{p}\mathrm{i}}=\mathrm{K}\left[1.6{\left({\displaystyle \frac{10{\mathrm{T}}_{\mathrm{i}}}{\mathrm{I}}}\right)}^{\mathrm{a}}\right]$$

(3)

where $\mathrm{K}={\displaystyle \frac{\mathrm{n}\mathrm{o}.\mathrm{o}\mathrm{f} \mathrm{d}\mathrm{a}\mathrm{y}\mathrm{l}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t} \mathrm{h}\mathrm{o}\mathrm{u}\mathrm{r}\mathrm{s}}{\frac{1}{2}\mathrm{n}\mathrm{o}.\mathrm{o}\mathrm{f} \mathrm{h}\mathrm{o}\mathrm{u}\mathrm{r}\mathrm{s} \mathrm{i}\mathrm{n} \mathrm{a} \mathrm{d}\mathrm{a}\mathrm{y}}}$ is a corrective coefficient for the latitude; T_i is the air temperature related to the ith month; (in °C) and $a=0.49239+1792\times {10}^{-5}\mathrm{I}-771·{10}^{-7}{\mathrm{I}}^2+675\times {10}^{-9}{\mathrm{I}}^3$ is the exponent of Equation (3), which is based on the yearly heat index $I=\sum _{\mathrm{i}=1}^{12}{\left({\displaystyle \frac{{\mathrm{T}}_{\mathrm{i}}}{8}}\right)}^{1.514}$.

As with the rainfall and temperature data, annual and monthly evapotranspiration have been correlated with altitude and used for evapotranspiration maps.

Monthly and yearly infiltration maps were obtained using Equation (1), first obtaining total outflow maps and then infiltration ones through Potential Infiltration Coefficients defined by Celico [32].

3. Results and Discussion

The method described was applied to various aquifers, resulting in 14 maps for each hydrological balance term (except for precipitation and temperature, which have 13 maps). Of these, 12 maps represent monthly data, while 2 maps describe annual data. The water balance was estimated using two different approaches for inflows: initially considering only rainfall as the inflow data and subsequently including snow data with rainfall.

In Figure 4, an example of the 12 monthly rainfall and snow maps for the Majella aquifer (n. 3 in Figure 1) is shown. These maps, along with the evapotranspiration maps, were used to calculate the total outflow by difference.

In Figure 5, the Monte Morrone (n. 2 in Figure 1) monthly and annual outflow maps are shown. As expected, the outflow values are lower during the dry period (from June to September) than during the wet one; this trend was observed for each aquifer.

In Figure 6, the example of infiltration maps for Monte Marsicano (n. 6 in Figure 1) is reported; as well as for outflow maps, the infiltration values are lower during the dry period. Even in infiltration terms, this trend was observed for each aquifer.

In

Table 2. Infiltration values for the main aquifers. Turc* stands for Turc modified method; Thornth stands for Thornthwaite method.

Majella Infiltration
	Rainfall and Snow Inflow							Only Rainfall Inflow
	mm				L/s			mm				L/s
	Min	Max	Mean	Std. Dev	Min	Max	Mean	Min	Max	Mean	Std. Dev	Min	Max	Mean
Thornth			1090	494			9903			844	357			7671
Turc	19	2279	1069	523	173	20,701	9710	28	1713	822	385	254	15,560	7467
Turc*	25	2346	1107	534	227	21,310	10,055	33	1752	846	391	300	15,914	7685
Monte Morrone Infiltration
Thornth			780	382			2637			716	339			2242
Turc	47	1346	634	322	159	4555	2145	48	1209	579	284	162	4090	1959
Turc*	54	1402	668	331	181	4741	2261	54	1236	602	286	186	4181	2036
Monte Marsicano Infiltration
Thornth			1016	274			7535			893	218			6623
Turc	86	1620	973	284	637	12,009	7215	85	1374	855	230	630	10,108	6340
Turc*	92	1686	1019	293	683	12,502	7556	91	1416	889	234	675	10,500	6592
Gran Sasso Infiltration
Thornth			457	140			18,197			391	93			15,558
Turc	35	997	409	154	1405	39,683	16,270	44	743	342	105	1751	29,564	13,608
Turc*	39	1027	427	157	1560	40,845	16,994	48	772	361	109	1910	30,718	14,364
Genzana–Greco Infiltration
Thornth			688	121			6036			610	82			5354
Turc	110	937	630	122	965	8224	5530	110	765	556	83	965	6714	4880
Turc*	117	978	662	126	1027	8584	5810	116	787	579	83	1018	6908	5082
Monte Porrara Infiltration
Thornth			490	124			1447			460	110			1360
Turc	76	765	443	131	226	2226	1309	71	679	403	113	210	2007	1191
Turc*	81	787	462	132	239	2326	1367	76	701	423	115	225	2072	1250

, the annual infiltration is reported from the maps’ statistical elaboration: maximum, minimum, and mean values were carried out for each method. The results are shown for both water balances, the one estimated using rainfall and snow as inflows, and the one calculated only with rainfall data.

As shown, infiltration estimates based on snow and rainfall measurements are higher compared to those derived from rainfall alone.

Table 3. Percentage variations between mean infiltration values.

Majella	Variation
Thornthwaite	29%
Turc	30%
Turc*	31%
Monte Morrone	Variation
Thornthwaite	14%
Turc	14%
Turc*	15%
Monte Marsicano	Variation
Thornthwaite	9%
Turc	10%
Turc*	11%
Gran Sasso	Variation
Thornthwaite	17%
Turc	20%
Turc*	18%
Genzana–Greco	Variation
Thornthwaite	13%
Turc	13%
Turc*	14%
Monte Porrara	Variation
Thornthwaite	7%
Turc	10%
Turc*	9%

Turc* stands for Turc modified.

shows the percentage variations between mean infiltration values for each aquifer. In accordance with the distribution of high-altitude sub-plain areas, notably, the highest percentage variation is observed for the Majella aquifer characterized by the presence of summit plateaus and karst forms at altitudes ranging from 1600 to 2900 m above sea level, while the lowest occurs for Monte Porrara, characterized by the presence of a single crest without summit plateaus.

These results underscore the critical role of snow cover in inflow calculations; clearly, excluding snow cover data can lead to a significant underestimation of effective infiltration values.

Another impact of accounting for snow cover in the inflow term can be observed in the regression lines of rainfall vs. altitude. As shown in the example in Figure 7 (left), when snow cover is not considered, the inflows at higher altitudes are underestimated, leading to an inversion of the regression line. On the other hand, when considering snowfall, the values at the higher stations increase their value, inducing a normal relationship between altitude and precipitation Figure 7 (right). This trend has been observed in several aquifers during the winter and fall months.

In order to evaluate the significance of the infiltration values obtained using both the contributions of rain and snow, these values were compared with the outflow ones from the literature calculated directly from the flow measurements of the main springs fed by the hydrogeological body.

The comparison should not be interpreted at a quantitative level but only as an estimate due to the following characteristics of the balances from the literature:

• The reference periods are different from those of the present study, and some are referred to periods of the last century with thermometric and pluviometric regimes, even if averaged over many years, that are different from those of the present study.
• The duration of the periods analyzed varies from a few dozen years to a few years,
• The number of analyzed springs varies from author to author.
• In many balances, only the basal springs with flow rates higher than tens of L/s are considered, while those, even if numerous, with flow rates lower than tens of L/s are excluded from the calculation.

In

Table 4. Infiltration values calculated for the main aquifers compared with springs’ outflow from literature. Turc* stands for Turc modified method.

Majella
Data Source	Mean infiltration (L/s)	Mean infiltration (mm)
Thornthwaite	9903	1089
Turc	9710	1069
Turc*	10,055	1107
Data Source	Discharge (L/s)	Discharge (mm)
Abruzzo Region [36]	8773	966
Monte Morrone
Data Source	Infiltration (L/s)	Mean infiltration (mm)
Thornthwaite	2637	780
Turc	2145	634
Turc*	2261	668
Data Source	Discharge (L/s)	Discharge (mm)
Conese et al. [37]	1475	436
Monte Marsicano
Data Source	Infiltration (L/s)	Mean infiltration (mm)
Thornthwaite	7535	1016
Turc	7215	973
Turc*	7556	1019
Data Source	Discharge (L/s)	Discharge (mm)
Boni & Ruisi [38]	7900	1065
Monte Porrara
Data Source	Infiltration (L/s)	Mean infiltration (mm)
Thornthwaite	1447	490
Turc	1309	443
Turc*	1367	462
Data Source	Discharge (L/s)	Discharge (mm)
Abruzzo Region [36]	1460	494

, the infiltration values from the present study are compared with the spring outflow values from literature. For each water body, the most complete bibliographic reference, among all the known ones, in terms of accuracy of the study and number of springs analyzed was chosen. For some water bodies, such as Gran Sasso–Sirente, the comparison was not performed due to the considerable uncertainties in identifying the extension of the water body. In fact, some authors include the secondary water body of Monte Sirente in the calculation [5,33,34], others did not [19,35].

4. Conclusions

The contribution of snow to the recharge of the aquifers of the central Apennines had never been verified with direct measurements of the snowpack over a large area. The estimates and calculations performed so far referred to single aquifers, or to recharge areas of single springs. The calculation methods referred to indirect principles based on satellite images and sometimes on evaluations from isotopic hydrology. In this work, the contribution of snow was evaluated by applying geostatistical techniques to snowpack thickness data collected from seven monitoring stations over a period of six consecutive years. Monthly snowpack data were combined with rainfall data to calculate the hydrological balance of the aquifers. The comparison between calculations that included snow contribution and those based solely on geostatistical interpolation of rainfall data revealed that snowmelt contributes between 10% and 30% of the total infiltration, a portion that would otherwise have been overlooked. The result suggests an important reflection on the extent of infiltration and consequent recharge due to snow, considering that in recent years the distribution and thickness of snow have decreased. Considering the observed percentages of infiltration from snow, it is recommended to incorporate snow thickness as a key parameter in future hydrological balance estimates, particularly when modeling scenarios involving changes in snow cover extent and distribution. The main limitation of an ideal study is the lack of monitoring stations in the area. Therefore, it would be advisable to expand the network of measurement stations to account for variations in snow depth due to topography. This can be achieved by increasing the number of stations and strategically distributing them based on elevation.