From Fertile Grounds to Sealed Fields: Assessing and Mapping Soil Ecosystem Services in Forlì’s Urban Landscape (NE Italy)

Abstract

Between 2022 and 2023, the urban soils of Forlì (NE Italy) were surveyed, sampled, analyzed, and mapped over an area of ca. 5700 ha, of which 2820 were sealed. The outcomes of the survey allowed the integration of the existing knowledge about soil and land use with the urban plan and provided the basis to produce a 1:10,000 map of urban soils along with their land capability and an updated 1:50,000 soil map of the municipality. Soil data (textural fractions, pH, organic carbon content) were interpolated over the entire case study area, providing the inputs for locally calibrated pedotransfer functions whose outputs were used to assess a set of seven indicators for the potential supply of soil ecosystem services (SESs): soil biodiversity, buffer capacity, carbon storage, agricultural production, biomass production, water regulation, and water storage. Maps of the seven ecosystem services on a hybrid resolution grid of 25 and 100 m were complemented with an overall urban soil quality map based on the combinations of four different SES indicators. Results show that for several services, hotspots occur not only in the peri-urban agricultural areas but also in unsealed soils within the urban fabric, and that different soils provide high-quality services in diverse constellations depending on the soil characteristics, age and extent of disturbance and degree of sealing.

1. Introduction

	“La terra che fé già la lunga prova
	e di Franceschi sanguinoso mucchio,
	sotto le branche verdi si ritrova”
	Dante Alighieri, Inferno, Canto XXVII

Despite a growing body of scientific literature in the last decade focusing on the role of urban soils as providers of ecosystem services, there is still a generalized lack of implementation of urban soils ecosystem services in the practice of urban planning [1]. If on one side the hyperbolic growth of urban population is acknowledged as one of the signatures of the Anthropocene [2,3], leading planners and policy makers to reframe the role of cities in terms of environmental quality, social justice and sustainable development [4,5], on the other, the potential of urban soils to contribute to urban resilience in facing the challenges posed by the ongoing climate crisis is far from being fully understood and remains untapped [6,7,8]. The assessment of soil-based ecosystem services (SESs) has recently achieved more importance, mostly due to the recognition of the relevance of the regulation services they provide, primarily their capacity to trap greenhouse gases from the atmosphere [9,10]. Nonetheless, as for ecosystem services in the built-up environment, Blanchart et al. [11], in considering the role of soil in urban planning documents, reported that “urban soils are predominantly seen as surface areas to be converted or as a potential threat due to their level of contamination or geotechnical properties”, and in a recent meta-analysis of published literature on mapping urban and peri-urban ecosystem services from more than 200 research papers extracted following the Preferred Reporting Items for Systematic Reviews and Meta-alpha Methods, not a single reference to SESs is mentioned [12]. This is likely to be due to several factors: the inherent complexity and heterogeneity of urban soils [13,14], which requires ad hoc surveys and dedicated human and financial resources, the widespread gaps in soil literacy among city planners and policy makers at the different administrative levels leading to ineffective soil governance [15,16], and the lack of standardized and widely tested approaches in assessing and mapping the ecosystem services of urban soils [17,18,19]. Hyun et al. [20]. proposed an urban soil quality index to evaluate the soil status in various spatial types of urban greenery considering comprehensively ecosystem services and functions of urban soil using ten quantitative soil indicators. More recently, a pedon-based approach called DESTISOL was proposed by Séré et al. [21] and tested in 37 urban soils under various situations and pedoclimates. The architecture of the model is based on 20 physico-chemical–biological soil indicators used to score 15 soil functions based on a detailed set of expert-based decision rules. Among the major research gaps in the assessment of urban soil ecosystem services, there is a lack of universally accepted metrics and indicators for assessing urban soil ecosystem services, making it difficult to compare studies and integrate findings across different regions and contexts. Furthermore, the gap in understanding how to effectively integrate urban soil ecosystem services into urban planning and policy frameworks has limited the adoption of sustainable management practices of urban soils. Addressing such issues in the EU-funded demonstrative LIFE project SOS4Life (Save Our Soils for Life, https://www.sos4life.it/en/project/, accessed 26 March 2025), Calzolari et al. [22] surveyed, assessed and mapped six ecosystem services of urban soils in the city of Carpi (NE Italy), tailoring to the urban context a methodology originally developed for agricultural soils [23]. Among the aims of the project was in fact the definition of a methodology for the detection, evaluation and mapping of ecosystem services provided by urban soils, which was aimed at quantifying ecosystem services and planning actions for their maintenance and improvement, and to identify areas suitable for urban regeneration to compensate the loss of ecosystem services due to land take following urban expansion. Among the activities of the after-LIFE plan, the Municipality of Forlì (NE Italy), a coordinating beneficiary of the SOS4Life project, requested the implementation of the same SOS4Life methodology to integrate the knowledge about urban soils and their ecosystem services within the new General Urban Plan (PUG), as foreseen by the Emilia–Romagna regional law 24/2017 on land planning to reduce and mitigate land take toward the EU target of zero net land take by 2050 [24]. The present work illustrates the outcomes of the urban soils survey carried out in the Forlì urban area between 2022 and 2023, resulting in (i) a map of urban soils at the scale of 1:10,000, (ii) the assessment of seven SESs including habitat for organisms, and (iii) the assessment of the impact of urbanization on SESs supply. Differently from other SES assessments and mapping approaches, the methodological steps followed in this work can be implemented in any urban context if the basic soil properties of urban soils are assessed via an ad hoc survey of urban soils. Furthermore, resorting to standardized indicators for SESs would easily allow a comparison between different urban contexts, and the interpretation of results would be straightforward also for non-soil practitioners. Our results contribute to raising awareness about the role of soil in the built-up environment and provide tools to assess and map urban ecosystem services, highlighting the possibility of integrating soil knowledge into urban planning.

2. Materials and Methods

2.1. Study Area

The study was conducted in the urban and peri-urban area of the municipality of Forlì (228.2 km², 117.430 inhabitants). Forlì (Figure 1; 44°13′21″ N 12°02′27″ E) is located at an average height of 36 m above mean sea level in the plain area of southeastern Emilia–Romagna south of the Po River, a few km away from the foothills of the Tuscan-Romagnolo Apennines to the south, and about 25 km from the Adriatic sea to the east. The urban area, with a history dating back more than 22 centuries, developed over Holocene sediments of the alluvial fans and terraced deposits of three Apennine rivers flowing SW to NE, namely the Montone and the Rabbi to the west and the Ronco to the east. The climate according to the Köppen–Geiger classification is warm temperate and constantly humid with a hot summer (Cfa). For the reference period 1991–2020, an average annual cumulative precipitation of 769 mm was recorded with a minimum average temperature of 3.9 °C (January) and a maximum average temperature of 24.6 °C (July) [25].

In May 2023, two separate extreme rainfall events occurred in the area with a cumulative rainfall exceeding 400 mm and locally 600 mm [26]. The local return period of the total accumulated rainfall recorded at the ground according to WMO standards [27] was more than 500 years, and the precipitation resulted in severe flooding in the whole Romagna basins with severe and widespread damage to buildings and infrastructures, which was amplified by the high rate of land take. In 2023, the sealing of the soil surface affected a total of ca. 3832 ha, corresponding to 16.8% of the area of the Forlì municipality (Figure 2), with an increase of 167 ha in the last decade. This share is almost double the regional average, which is equal to 8.9% [28].

The study area corresponds to the continuous built-up area, with an extension of 5691 ha, nearly half of which has urbanized (2645 ha) but with a relevant share of undisturbed agricultural soil, which is mostly in the peri-urban buffer (2771 ha). Urban green areas make up about 420 ha, comprising mainly public and private parks, about 178 ha, followed by urban wastelands, about 144 ha, sport and leisure green, about 66 ha, and roadside and railway side green, about 37 ha [29].

2.2. Available Soil Data and Urban Soil Survey

As most SESs in urban areas are provided by green areas, the semi-quantitative assessment of soil functions and related services focused on gardens and parks, enclosed or peri-urban agricultural areas, sports green areas and flower beds generally larger than 0.1 ha. Private gardens and appurtenances were not considered, having verified their high degree of sealing (paths, parking lots, access ramps to garages, etc.). However, even these soils, although strongly altered, perform some functions to some extent. Differently from most of the cities of the Emilia–Romagna plain, whose urban areas are rather compact, and the peri-urban area is effectively such, in Forlì, enclosed agricultural lands are rather widespread, so the peri-urban area is quite fragmented.

A set of preliminary “urban pedolandscape units” were defined considering the distribution of “natural soils” underlying the town as derived from the soil map of Emilia-Romagna plain at a 1:50,000 scale [30], and the urbanistic structure was defined in terms of (i) age of the different buildings and parts of the town, (ii) typology of the various neighborhoods, and (iii) occurrence of public and private green areas. With the support of these maps, and resorting to multitemporal aerial images, eighty-three urban pedolandscape units were defined. Accordingly, 240 sites were in the urban (N = 160) and peri-urban (N = 80) area (Figure 1b) with the aim of covering the different pedolandscape units identified. Each unit was also characterized in terms of the degree of disturbance the soil was subject to, resorting to a multitemporal aerial photo examination (1954, 1978, 1989, 1994, 1995, 1997, 2003, 2006, 2008, 2011, 2014, 2017, 2018, and 2020) to assess the age and extent of soil disturbance. This was accomplished via an aerial photograph interpretation of stereoscopic images until 2014 and then via assessment of the differences in the Normalized Difference Vegetation Index (NDVI) derived from Sentinel-2 images via the Google Earth Engine as described in [28]. Additionally, 8 soil profiles were dug, described, sampled and analyzed at the end of summer 2023. In total, there were 762 soil observations in the municipal area (3/km²) and 346 in the study area (6/km²), 248 of which were new observations from the ad hoc survey of urban and peri-urban soils and 98 previous ones. The collected samples for the reference depth 0–30 cm were analyzed for texture (sand, silt and clay) limestone content %, pH, organic carbon %, and cation exchange capacity; in total, 592 samples from 172 sites were analyzed. Soil textural fractions (USDA limits) [31] were determined with the pipette method (DM 13/09/1999, Method II.5, II.6) in the case of clay and silt and with the sieve method in the case of sand (DM 13/09/1999, Method II.5); pH was determined in H₂O with a 1:2.5 soil to water ratio (DM 13/09/1999, Method III.1), the percentage of total CaCO₃ was measured with the gas volumetric method (DM 13/09/1999, Method V.1), and organic C % was assessed with an elemental analyzer (DM 13/09/1999, Method VII.1) [32].

Through the integration of the multitemporal aerial photos’ series with the map of the sealed areas available for the whole of Italy at a 10 m resolution [28], the LIDAR digital model with a 2 m resolution, and the soil observations and analyses, the preliminary map of urban pedolandscapes (1:10,000) and the soil map of the entire municipality (1:50,000) were updated and finalized.

2.3. Soil Functions and SESs Assessment

The assessment of the soil contribution to seven urban SESs resorted to indicators based on measured or estimated soil properties assumed as a proxy of soil functions according to the methodology described in Calzolari et al. [23]. The method considers a set of indicators of soil functionality linked to the provision of ecosystem services. The ecosystem services, the underpinning soil functions and the soil data required as input for their calculation are presented in

Table 1. Ecosystem services (ESs), underpinning soil functions, indicators and input data.

Ecosystem Service a	CICES Code 5.1 b	Soil Contribution to ES c	Soil Function	Indicator	Input Data for Calculation	Code
Regulating	2.2.1.1 2.3.3.2	Buffering capacity for nutrients and pollutants: natural attenuation (potential)	Storing filtering and transforming nutrients, substances and water	Cation exchange capacity (CEC) Soil reaction Rooting depth	C org. % Clay % pH Coarse fraction %	BUF
Regulating	2.1.1.2 2.3.3.2	Carbon sequestration (potential)	Carbon pool	Carbon sequestration actual	C org % Bulk density	CST
Provisioning	1.1.1.1	Food provision (potential)	Biomass production	Land capability (LC) map	LCC and integrades	PRO
Provisioning	1.1.1.x 1.1.5.x	Biomass supply (potential)	Biomass production	NDVI average 2015–2020	NDVI (LANDSAT8)	BIOMASS
Regulating	2.2.1.3	Water regulation: Runoff-flood control (potential)	Storing filtering and transforming nutrients, substances and water	Infiltration capacity	Ksat (mm/h) Psie (cm)	WAR
Regulating (Provisioning)	2.2.1.3 (4.2.2.2)	Water regulation: Water storage (potential)	Storing filtering and transforming nutrients, substances and water	Water content at field capacity Presence of water table	Field capacity (−33 kPa)	WAS
Supporting	2.2.2.3	Habitat for soil organisms	Biodiversity pool	Potential habitat for soil organisms	Index QBS-ar Covariates DSM	BIO

^a MEA, 2005 [33]; ^b CICES Haines-Young, R., Potschin, M.B., 2018 [34]; ^c Dominati et al., 2010 [35].

.

Table 2. Input data and calculations of soil ecosystem service (SESs) indicators.

SES Code	Input Data	Calculation
BUF	CEC (cmolc/kg) depending on OC (%) and clay (%) CEC = 6.332 + 0.404 clay + 1.690 OC (R2 = 0.75) pH Coarse fragments content, sk (%) Average depth of shallow water table, WT (cm)	BUF0-1 = Log CEC (pH; sk)0-1 with pH < 6.5 reduction by 0.25 or 0.5 depending on CEC and by 0.25 for sk > 30% Water Table (WT) depth < 30 cm BUF0-1 = Log CSC (pH; sk)0-1 × WT/30
CST	Organic carbon, OC (%) Bulk density, BD (Mg m−3)	CST0-1 = log [OC × BD × (1-sk)]0-1
PRO	Land capability (LC) classes and intergrades [39]	LCC reclassification (0-1)
BIOMASS	NDVI (Normalized Difference Vegetation Index)	Standardization (0-1) NDVI (average median values 2015–2020)
WAR	Saturated hydraulic conductivity, Ksat (mmh−1) Air entry potential, PSIe (cm)	WAR0-1 = logKsat0-1 − PSIe0-1
WAS	Water content at field capacity (−33 kPa), WCFC (vol/vol) Average depth of water table, WT (cm) sk, coarse fragments (Ø > 2 mm, vol/vol)	WAS0-1 = (WCFC × 1-sk)0-1 if WT > 100 cm, and WAS0-1 = (WCFC × 1-sk) × WT/100 if WT < 100 cm
BIO	Soil Biological Quality index, QBSar [40] Covariate per digital soil mapping	Spatialization of QBSar point data values via DSM (Quantile Random Forest)

reports the calculations applied to derive each indicator, as specified in Calzolari et al. [23], for all of the indicators reported in Table 1, except for BIO and BIOMASS. The former was derived from applying the digital soil mapping (DSM) technique based on machine learning algorithms [36] to the set of Soil Biological Quality index values [37] available at a regional scale, while the latter is based on the value of the Normalized Difference Vegetation Index (NDVI) [38] derived from Landsat8 via Google Earth Engine [39].

The Land Capability Classes (LCCs), being originally assessed for agricultural and forest lands [40,41], were also defined for the urban area based on the urban soil units of the new map on scale 1:10,000. The hydraulic soil properties listed in Table 2, BD (bulk density, Mg/m³), Ksat (saturated hydraulic conductivity, mmh⁻¹), PSIe (air entry potential, cm), WCFC (water content at field capacity, vol/vol) were calculated, resorting to a set of locally calibrated pedotransfer functions (PTFs) [42,43]. The average shallow groundwater depth WT (cm) was derived from Calzolari and Ungaro (2012) [44].

The values of each SES indicator are given as numbers in the range from 0 to 1, resorting to an interval normalization transformation [45]:

Xi_0-1 = (X_i − X_min)/(X_max − X_min)

(1)

where Xi_0-1 is the standardized value [0, 1], Xi is the current value, and X_min and X_max are, respectively, the maximum and minimum of the indicator observed in the considered territory. The maximum observed value is set equal to 1, and the value 0 indicates the relative minimum in the area considered. The results are then strongly influenced by the degree of variability of the measured and estimated soil properties observed at the scale of investigation. By considering specific portions of territory of administrative, planning or management relevance (e.g., province, union of municipalities, municipality), the indicators can be normalized over the area of interest at the scale of investigation. In this case, the indicators have been normalized in the range 0–1 over the entire municipal territory. A synthetic index of soil quality (IQ4) based on the potential supply of ecosystem regulating and provisioning services was eventually calculated as the sum of four indicators: BUF, CST, PRO and WAR. The IQ4 values were then classed 1 to 5 based on the breakdown of the observed distribution, with 1 and 5 indicating the highest and the lowest quality, respectively.

2.4. Mapping Urban Soils Properties and Ecosystem Services

Through the geostatistical processing of point data, followed by the use of locally calibrated PTFs, the maps of soil properties at the basis of the estimation of the SESs indicators were produced for the entire municipal territory. Seven maps were therefore derived, one for each SESs, in raster format with a hybrid resolution, i.e., 25 m for the core urban area and 100 m for the rest of the municipal territory.

The spatialization of the point data over the entire municipal area was carried out using sequential Gaussian simulations (SGSs) [46]. Being based on multiple realizations, SGSs allow for the assessment of uncertainty in predictions, which is crucial for risk-based decision making. Furthermore, SGS realizations maintain the spatial autocorrelation and statistical properties of the data, ensuring realistic representations of soil variability, avoiding the shortcomings of deterministic interpolation methods and the inability of machine learning algorithms to deal with spatial autocorrelation. Although computationally demanding, SGS can easily integrate auxiliary data in the process of simulation: in the case presented in this work, the outcomes were conditioned on the mean values of the urban soil map delineation.

The implementation of the SGSs (N = 25) required data transformation into normal scores, i.e., the normalized residues of the average values calculated for the individual properties (sand, silt, clay, pH and C org) at the level of the soil map delineations at a scale of 1:50,000 for the entire municipal territory and at a scale of 1:10,000 for the urbanized core area. Once each point was associated with the mean value of the delineation in which it falls, the residual value (deviation from the mean value) was calculated by difference. Once normalized via a normal score transform, this was used to estimate and model the experimental semivariograms needed to implement the geostatistical simulations through ordinary kriging on a regular grid of 25 m in the urban area and 100 m in the remaining municipal territory. The mean value of the 25 simulations was then subjected to an inverse transformation to obtain the residual value to which the mean value of the delineation was finally added to obtain the estimate value of each variable. An exception to this procedure is represented by the percentage of coarse fragments, whose distribution is mainly represented by values equal to zero, a situation in which a geostatistical approach is not applicable. For this reason, in the case of the coarse fragments content, the mean value attributed to the delineations of the soil map was used based on the available data (N = 1121). All the geostatistical analyses presented in this paper were carried out with the software Wingslib1.3.1 [47], which works in conjunction with the GSLIB90 executables [48]. All GIS operations and mapping were performed using QGIS v3.22.11 [49].

3. Results

3.1. Urban Soil Map 1:10,000 and Degree of Soil Disturbance

Evidence from the survey allowed in the first place to assess the degree of soil disturbance at each sampling site (Figure 3a), which was summarized in seven different classes as follows:

• Undisturbed “natural” soils found in agricultural areas but also in public parks and uncultivated land;
• Disturbed soils with superficial fill of allochthonous soil material in the topsoil;
• Deep disturbance in place: like 2 but the disturbance is deeper with overturned horizons and buried soil;
• All fill: often it is made of building material mixed with soil with high percentages of coarse elements so that augering and sampling are impeded;
• Disturbed after sampling;
• Urbanized and sealed;
• Natural topsoils reworked: no fill, mixing of Ap with bricks and artifacts. The bricks can be from the Roman age. They occur quite frequently also in agricultural areas.

Furthermore, urban soils are almost always characterized by thin surface layers enriched in organic matter, as they are almost always found under permanent herbaceous cover. The seven classes of soil disturbance were reduced to four (Figure 3b) when applied to the 79 mapping units identified in the urban core area (Figure 4): (i) undisturbed “natural” soils (merging classes 1 and 7); (ii) disturbed soils (merging classes 2 to 5); (iii) urbanized soils at different degree of sealing (class 6); and (iv) urbanized/disturbed mixed class.

Figure 3. Degree of soil disturbance (a) at sampling points and (b) over the surveyed urban core area.

Figure 3. Degree of soil disturbance (a) at sampling points and (b) over the surveyed urban core area.

Figure 4. Soil map (scale 1:10,000) of the urban core area. A description of urban soils mapping unit is provided in the Supplementary Materials (Table S1).

Figure 4. Soil map (scale 1:10,000) of the urban core area. A description of urban soils mapping unit is provided in the Supplementary Materials (Table S1).

Undisturbed soils are the majority (48.4% of the area, 2771.43 ha) and are mostly found in peri-urban and intra-urban agriculture, the latter being quite widespread in Forlì. A small percentage of these soils is also found in urban green areas or in areas included in industrial zones. Disturbed soil was mostly observed within urban and industrial areas; in small percentages, they were also found in the agricultural land (mostly abandoned; the most frequent use is lawn). The percentage of mapped occurrence is about 5% (427 ha), but it is believed that it is probably higher if we consider small gardens, urban flowerbeds, roundabouts, etc. that were not mapped due to their small size.

To identify the extent of the disturbance, we relied on the survey data and the examination of aerial photos (from 1936 to 2020). With a few exceptions, the disturbed urban soils are derived from overturning and carrying over of soils present on site or from nearby locations with a very low occurrence of allochthonous soil material. Urban soils encompass all soils that had artifacts such as bricks, ceramic, cement, plastic, and building materials within the profile. Four soil typologies (Figure 5) were created for the urbanized area (36.02%, 2051.25 ha) based on the degree of soil sealing and risk of flooding: (i) URB1 (134.26 ha, 2.36%): historical center, where the occurrence of unsealed soil is variable but overall quite low; (ii) URB2 (1177.15 ha, 20.67%): residential areas generally characterized by single houses or buildings with small gardens and green areas characterized by the presence of disturbed soils; (iii) URB3 (11.45 ha, 0.20%): urbanized area comparable to URB2 in terms of soil sealing but located in areas at higher risk of flooding (floodplain areas); (iv) URB4 (728.39 ha, 12.79%): industrial areas where the percentage of soil sealing is very high—higher than the URB2 and URB3 units. The unsealed areas were mapped in most cases; they are private green areas such as lawns or small gardens.

Very often, these urbanized areas, even if contiguous, were divided into separate mapping units based on the underlying original soils as derived from the 1:50,000 soil map. Finally, some urbanized areas have larger unsealed surfaces for which some cartographic units have been identified, which are associations of urbanized and interlocked disturbed soils (612.93 ha, 10.78%).

3.2. Maps of Soil Properties

Table 3. Descriptive statistics of selected soil properties (0–30 cm) for the available soil data used for SESs assessment in the municipality of Forlì.

Variable		Agricultural Soils	Undisturbed Soils	Disturbed Soils	Urbanized Soils	Urbanized/ Disturbed Soils	Urban Core Area	Forlì Whole Area
	Num. Obs.	792	195	71	37	26	355	1121
Sand %	Mean	20.64	19.64	20.11	19.16	19.45	19.62	20.35
	Dev. Std.	7.09	9.16	7.47	8.29	10.41	8.66	7.64
	Min	3.87	3.38	6.22	6.22	3.87	3.38	3.38
	Max	46.97	43.38	27.20	35.43	43.38	43.38	46.97
Silt %	Mean	50.58	51.12	52.41	50.36	52.04	51.33	50.82
	Dev. Std.	4.88	3.78	2.71	3.57	5.79	3.81	4.60
	Min	33.65	36.05	47.19	43.76	36.05	36.05	33.65
	Max	62.94	60.36	56.74	56.74	60.36	60.36	62.94
Clay %	Mean	28.78	29.24	27.48	30.48	28.51	29.04	28.83
	Dev. Std.	6.36	8.35	7.64	7.51	8.12	7.97	6.91
	Min	16.41	16.41	21.42	20.24	16.41	16.41	16.41
	Max	43.96	42.78	42.78	42.78	42.78	42.78	43.96
C org. %	Mean	1.03	1.33	1.77	1.50	1.31	1.43	1.15
	Dev. Std.	0.29	0.46	0.45	0.44	0.47	0.48	0.41
	Min	0.46	0.76	1.08	0.81	0.76	0.76	0.46
	Max	2.34	2.21	2.68	2.21	2.21	2.68	2.68
pH	Mean	7.81	7.82	7.71	7.82	7.87	7.80	7.80
	Dev. Std.	0.27	0.18	0.16	0.17	0.18	0.19	0.24
	Min	6.56	7.44	7.48	7.57	7.56	6.99	6.56
	Max	8.29	8.13	8.05	8.05	8.13	8.13	8.29
Skel %	Mean	0.09	0.98	1.59	1.33	0.70	1.08	0.40
	Dev. Std.	0.51	1.58	1.63	1.93	1.12	1.62	1.09
	Min	0.00	0.00	0.00	0.00	0.00	0.00	0.00
	Max	6.97	6.97	6.97	6.97	3.25	6.97	6.97

summarizes the descriptive statistics of the soil properties requested for the assessment of the indicators of potential SESs supply.

From the values reported in Table 3, it is concluded that compared to agricultural soils (N = 792), the soils of the urban area (N = 355) are characterized by a significantly lower sand content (p = 0.1) as well as significantly higher silt, skeleton and organic C contents (p = 0.05). Likewise, although not statistically significant, higher pH values are found in highly urbanized areas. Analyzing the data in terms of type of disturbance within the urban area, it can be observed that the parameters considered are generally characterized by greater variability (higher standard deviation values), especially in the case of urbanized and urbanized/disturbed soils, as can be seen in the box and whisker plots shown in Figure 6.

The description of eight soil profiles in the urban green areas (

Table 4. Site description and classification of eight urban soil profiles in Forlì.

Site	Site Description			Soil	Classification WRB 2014
Public park	Permanent meadows, not irrigated Natural soil with surface fill			BEL2	Hypereutric Cambisols (Transportic)
Public park	Permanent meadows, not irrigated Surface fill down to 120 cm			DRA	Calcaric Regosols (Prototechnic, Transportic)
Public park	Permanent meadows, not irrigated Surface fill 0–60 cm over in situ alluvial deposits			RTFy	Hypereutric Cambisols (Transportic)
Public park	Permanent meadows, not irrigated Surface fill down to 120 cm			RES	Calcaric Regosols (Prototechnic, Transportic)
Public park	Meadows, old trees, irrigated Surface fill 0–55 cm over in situ alluvial deposits			TEG3	Hypereutric Cambisols (Transportic)
Public park	Sealed with trees is small allotments Planned de-sealing area			TEG3	Hypereutric Cambisols (Transportic)
Public park	Permanent meadows, not irrigated Surface fill down to 150 cm			DRA	Calcaric Regosols (Prototechnic, Transportic)
Public park	Permanent meadows, not irrigated Surface fill down to 190 cm			SBG1	Urbic Ekranic Technosols (Calcaric Epitechnoskeletic)

) was relevant for urban soils mapping purposes, as their field evidence and analyses provided the information necessary to highlight the relationships between in situ soil materials and exogenous material, either soil or of anthropic nature. In this way, it became possible to assess the actual nature and extent of anthropic soil disturbance as well as its effects on the provision of ecosystem services. Analytical data from the soil profiles are provided in the Supplementary Materials (Table S2).

The procedure for estimating the indicators of ecosystem services was preceded by the estimation of the necessary chemical–physical parameters. To this end, the point values measured in the sampling sites were interpolated via geostatistical conditional simulations over the study area to reconstruct the continuous spatial distribution of the parameters reported in Table 3, which was followed by the application of a set of PTFs calibrated on regional data sets from Emilia–Romagna.

The parameters of the semivariograms models (Figure S1), which were used for the spatial interpolation of the normalized residuals of the three textural soil fractions, pH and organic C content, are shown in

Table 5. Semivariogram model parameters for selected soil properties. For all properties, the parameters refer to nested spherical models. The spherical semivariogram model can be written as γ(h) = C₀ + Σ_ni=1 C_i (1.5 h/r_i − 0.5h³/ri³), for h ≤ r_i, where h is the distance (m), C₀ is the nugget, C_i (i = 1, …, n) is the sill of the i nested structure, and r_i is its spatial range (m).

Variable	Nugget C0	Model	Sill C1	Range a1 (m)	Sill C2	Range a2 (m)
Sand %	0.30	Sph. + Sph.	0.60	600	0.10	10,000
Silt %	0.25	Sph. + Sph.	0.63	700	0.14	12,000
Clay %	0.24	Sph. + Sph.	0.57	450	0.18	6700
Organic C %	0.42	Sph. + Sph.	0.46	1058	0.19	5353
pH	0.37	Sph. + Sph.	0.40	996	0.37	2872

. For all variables, a double spherical model with nuggets provided the best interpolation of the experimental semivariograms. The normalized residuals of the considered soil properties have a share of spatially uncorrelated variance (nugget, C₀) ranging between 24 and 42%. The two spatially structured components, C₁ and C₂, accounted for similar shares of the residual variance only in the case of pH, while for the textural fractions and organic C content, the short-range component explained most of the spatially structure variance. The resulting maps are shown in Figure 7a–e; in the case of coarse fragments content, Figure 7f reports the average values of the typological soil units.

At the nodes of a 25 m regular grid over the urban core area, and of a 100 m regular grid for the rest of the municipality, a set of locally calibrated pedotransfer functions estimated the values of the soil properties requested to calculate the indicators for the selected soil ecosystem services. These PTF estimated properties, depicted in Figure 8, were bulk density (BD, Mg m⁻³), which provided the necessary input to estimate the soil C stock (Mg/ha, for the reference 0–30 cm depth), saturated water conductivity (Ksat, mm h⁻¹), air entry point (PSIe, cm), water content at field capacity (WCFC, vol./vol.), and cation exchange capacity (cmol/kg).

3.3. Urban Soils Ecosystem Services: Assessment and Mapping

The maps of the seven indicators of soil ecosystem services considered are shown in Figure 9a–g. All indicators were rescaled on local variability (minimum and maximum values) so to have values between zero and one. The raster maps have a hybrid resolution, which is equal to 25 m in the urban area specifically surveyed and 100 m in the predominantly agricultural areas in the rest of the municipal territory. In the case of PRO, the indicator was based on the 0–1 standardization of the LC classes detected on the new LC map; both map and LC standardized values are provided in the Supplementary Materials in Figure S1 and Table S3, respectively.

The values of the indicators displayed in the maps in Figure 9 were eventually used to characterize the potential ecosystem services provision by soils considering the typological units of the urban soils map (N = 79, Figure 4) and as a function of the degree of disturbance to which the soils of the urbanized area surveyed have been subjected (N = 4, Figure 3b). The descriptive statistics of each indicator for the different cartographic units of the urban soil map can be used to verify which ecosystem services are in synergy with each other and which are antagonistic. Limiting ourselves to considering the first 30 units by territorial extension, which cover a total of 92% of the urbanized area (2546 ha) and considering only the soil surfaces free from infrastructures and buildings, in terms of average values of each indicator, we obtained the figures shown in

Table 6. Average values of ecosystem services indicators and soil quality index IQ4 for the mapping units of the urban soil map.

Mapping Units	Soil Disturbance	Area, ha	Area Share	BIO	BIOMASS	BUF	CST	PRO	WAR	WAS	IQ4
BEL1	undisturbed	468.8	0.17	0.60	0.60	0.52	0.70	0.86	0.35	0.59	2.424
BEL1/LAM1	undisturbed	274.9	0.10	0.69	0.68	0.52	0.75	0.71	0.40	0.58	2.381
MDC2	undisturbed	156.1	0.06	0.54	0.60	0.70	0.73	0.86	0.27	0.67	2.561
URB2	urbanized/sealed	130.1	0.05	0.34	0.36	0.59	0.78	0.00	0.50	0.61	1.862
SMB1	undisturbed	128.6	0.05	0.59	0.56	0.53	0.66	1.00	0.41	0.58	2.598
GRZ1/BOR1	undisturbed	116.6	0.04	0.46	0.59	0.44	0.77	0.65	0.41	0.44	2.273
RTFy	undisturbed	116.2	0.04	0.61	0.57	0.87	0.75	0.86	0.11	0.81	2.588
REM1/CTL4	undisturbed	94.6	0.03	0.68	0.62	0.61	0.65	0.86	0.41	0.61	2.532
TEG2	undisturbed	93.3	0.03	0.64	0.64	0.53	0.73	0.86	0.33	0.60	2.438
TEG1	undisturbed	86.3	0.03	0.62	0.61	0.46	0.72	1.00	0.41	0.55	2.587
BGT2	undisturbed	84.8	0.03	0.61	0.58	0.68	0.54	0.86	0.31	0.63	2.391
CTL4	undisturbed	81.9	0.03	0.51	0.59	0.67	0.70	0.86	0.31	0.65	2.532
RNV1	undisturbed	79.1	0.03	0.61	0.63	0.72	0.70	0.86	0.30	0.67	2.579
RTF1	undisturbed	71.4	0.03	0.56	0.60	0.73	0.77	0.86	0.23	0.71	2.591
URB4	urbanized/sealed	70.5	0.03	0.39	0.38	0.62	0.70	0.00	0.38	0.62	1.689
LBA1	undisturbed	68.7	0.02	0.54	0.56	0.81	0.70	0.71	0.18	0.75	2.395
PTR2	undisturbed	59.9	0.02	0.59	0.63	0.37	0.60	1.00	0.60	0.50	2.569
TEG2/SGR2	undisturbed	40.3	0.01	0.73	0.78	0.60	0.79	0.92	0.31	0.62	2.622
PRD1	undisturbed	32.2	0.01	0.51	0.52	0.58	0.69	0.86	0.34	0.62	2.471
BEL2	disturbed	31.6	0.01	0.49	0.57	0.49	0.72	0.86	0.68	0.56	2.746
URB4-RES	urb./disturbed	31.5	0.01	0.38	0.44	0.65	0.62	0.39	0.20	0.65	1.865
RNV0	undisturbed	30.8	0.01	0.49	0.61	0.88	0.76	0.79	0.17	0.77	2.604
DRA	disturbed	28.8	0.01	0.35	0.65	0.56	0.71	0.86	0.50	0.60	2.621
MDC1	undisturbed	26.9	0.01	0.70	0.68	0.89	0.90	0.71	0.32	0.76	2.816
PRD1/LBA1	undisturbed	25.3	0.01	0.55	0.61	0.71	0.66	0.79	0.31	0.66	2.477
RTF1/MAN1	undisturbed	25.1	0.01	0.39	0.62	0.72	0.74	0.79	0.25	0.69	2.501
SAD1	undisturbed	24.1	0.01	0.58	0.63	0.61	0.45	0.71	0.29	0.63	2.059
CTL4/MDC2	undisturbed	23.8	0.01	0.50	0.56	0.64	0.77	0.86	0.32	0.61	2.582
SBG0	disturbed	22.2	0.01	0.25	0.54	0.47	0.65	0.57	0.77	0.47	2.458
PRD1/PRD3	undisturbed	21.6	0.01	0.81	0.63	0.62	0.74	0.86	0.27	0.65	2.487

and graphically summarized by the radar chart in Figure 10.

From the values in Table 6 and the trend of the values in Figure 10, we can appreciate the severe drop in the average values of the PRO, BIO, and BIOMASS indicators in urbanized and urbanized/disturbed soils. The decrease was on average more contained for WAS and BUF, while in the case of WAR in the URB1 unit, an average value higher than 0.8 was observed. On the other hand, the average of CST remained high with values > 0.7, which was close to 0.8 in the URB2 and URB1 units.

Table 7. Pearson correlation coefficients between the mean values of the ecosystem services indicators of the mapping units of the urban soil map. Statistically significant correlations are in bold (p = 0.05).

Indicator	Mean	Std. Dev.	BIO	BIOMASS	BUF	CST	PRO	WAR	WAS
BIO	0.484	0.180	1.000	0.506	0.427	0.373	0.437	−0.102	0.421
BIOMASS	0.519	0.156	0.506	1.000	0.328	0.351	0.762	0.218	0.330
BUF	0.580	0.152	0.427	0.328	1.000	0.711	0.277	−0.342	0.890
CST	0.680	0.156	0.373	0.351	0.711	1.000	0.258	0.089	0.838
PRO	0.649	0.281	0.437	0.762	0.277	0.258	1.000	−0.001	0.305
WAR	0.395	0.183	−0.102	0.218	−0.342	0.089	−0.001	1.000	−0.217
WAS	0.587	0.137	0.421	0.330	0.890	0.838	0.305	−0.217	1.000

reports the correlation coefficients between the average values of the ecosystem services indicators of the mapping units of the urban soil map. In most cases, the services were in synergy with each other, except for WAR and WAS, and WAR and BUF.

What was highlighted at the mapping units level became even more evident when observing the trend of the weighted average values as a function of the type of disturbance that characterizes each mapping unit (Figure 11). The BIO indicator was on average significantly higher in the “undisturbed” soils of peri-urban agricultural areas with a decreasing trend as the degree of disturbance increased. In the case of the BIOMASS indicator, the average value of disturbed soils was close to that of undisturbed soils and was significantly higher than that of urbanized and urbanized/disturbed soils. As for PRO, the average values of the mapping units characterized by undisturbed agricultural soils and disturbed soils were both higher than 0.8, and they were significantly higher than the average values characterizing the units with intermediate typologies urbanized/disturbed (0.377 ± 0.054) and obviously of the completely urbanized ones where the PRO value was null. The average values of the BUF, CST and WAS indicators were not significantly affected by the degree of disturbance and turned out to be very close to each other. It is interesting to note that in the case of CST, the soils of the green areas of the urbanized fabric had the highest average value (0.708 ± 0.098), which was slightly higher than that estimated for the agricultural soils of the peri-urban area (0.700 ± 0.031). The BUF and WAS indicators were on average higher in undisturbed agricultural soils and in urbanized soils. Finally, in the case of WAR, the indicator showed an average and significantly higher value in the mapping units characterized by the presence of disturbed soils (0.531 ± 0.088) and urbanized soils (0.441 ± 0.329). In this last case, however, a high variability was observed.

4. Discussion

The findings of this work need to be framed in the context of the urbanization dynamics which characterized the territory of Emilia–Romagna since the last two decades with a significant impact on the provision of soil-based ecosystem services. In 2023, the general figure in terms of soil sealing for the whole region reports an average value of 8.91%, which is equivalent to 2005.5 km² [28]; the same figures in 2006 were 8.36% and 1880.7 km² with an average loss of soil equal to 734 ha per year. This is indeed quite high a figure, as Emilia–Romagna is the third Italian region for land take compared to the national average (7.16%). However, this does not give a realistic idea of the level of urbanization of part of the plain areas of the region, as it obviously also accounts for all the hilly and mountainous areas, which sum up to nearly half of the region and are very sparsely inhabited. When considering the data at the municipal level, in fact, a much more worrying image emerges for the inhabited centers of the plains, especially for those located along the major NW–SE road axes with values often above 15%, as in the case of Forlì (16.5%, 3832 ha) [28]. Such a continuous loss of agricultural soil was not coupled with a positive demographic trend or with significant emigration from rural areas toward the city; rather, it is associated with changes associated with recent needs for new industrial settlements, large storage spaces for goods and increasingly widespread infrastructures. Considering the whole region, between 2003 and 2020, sealed areas grew constantly and significantly: urbanized areas, which include residences, commercial and industrial infrastructures and everything classified by the Corine Land Cover Classification (CLC) with code 1 increased by 10.4% (+25,984.2 ha). It must be noted that the most significant increases were recorded in some specific classes such as goods sorting plants (+126.4%), landfills and deposits of quarries, mines and industries (+69.6%), networks for the distribution and production of energy (+44.9%) and railway networks (+25.7%) [50]. In this scenario of constantly increasing soil sealing, the implementation of the Regional Law n. 24/2017 “Regional discipline on the protection and use of the land”, aiming to zero net land take by 2050, is still limited and so far, it has missed its goals. This on the one hand depended on the possibility for the municipalities to postpone for three years the approval of their new General Urban Plans (PUGs), which would have introduced a limit of 3% on additional land take by 2050. On the other hand, it was determined by the exclusion from the 3% ceiling, established by the law, of public or public interest interventions, of the expansion of already established business activities and of new production facilities of significant regional and/or state interest. Nevertheless, even if the land take did not reverse its growing trend, the application of the Regional Law by the municipalities which implemented it through the approval of their PUGs spared from land take 15.274 ha of agricultural soil out of 21.922 which were planned for being built compliant with the old urban plans.

Similarly, Forlì increasingly lost an additional 127.7 ha of agricultural soil between 2017 and 2023 with a peak of 35 ha in 2023, further reducing the stock of natural capital represented by soils and impacting the potential supply of ecosystem services. It was estimated that between 1997 and 2016, Forlì lost due to land take approximately 570 ha of agricultural soils, corresponding to ca. EUR 194 million of average agricultural value and the capacity to produce approximately 370,000 tons of wheat per year. Of these soils, 78% were high-quality soils: deep alluvial soils with excellent chemical–physical fertility characteristics. In terms of land capability class (LCC), approximately 21% of the LCC I soils were lost, and more than 34% of LCC II soils were lost, suggesting the best soils are those consumed the most. In the same time span, the capacity of storing approximately 3.8 million m³ of water was lost along with 319,000 tons of soil organic carbon, which was equivalent to 1,171,730 tons of CO₂ being released back into the atmosphere [51]. Furthermore, the recent and repeated severe flood events of 2023 and 2024 following unprecedented rainfall events highlighted the inadequacy of the prevention and mitigation strategies put in place so far and the absolute necessity to prevent further land take. To this goal, the municipality of Forlì in 2024 adopted the SES maps presented in this work as an integral part of its PUG in the attempt to reach two results: preserving the best-quality soils as identified by the IQ4 index from additional sealing and using the index to guide compensation measurements in all circumstances. Sealing is not avoidable through the identification of candidate areas for de-sealing whose quality and extent could compensate the further soil loss.

The methodology presented in this work, initially tested in the municipality of Carpi [22], was further refined for its application to the municipality of Forlì by adopting a finer estimation grid with a hybrid resolution to tackle the major inherent complexity of the Forlì urban soils and the higher spatial variation in the natural soils underlying the urban fabric. The SES assessment approach, although providing outputs in terms of standardized indicators scores and maps, is based on quantitatively assessed soil functions based on measured or PTF derived soil data. For example, in the case of the regulating ES indicators WAS and CST, it was calculated that in the 420 ha of urban green areas, the first 30 cm of soil in the green areas of the municipality of Forlì have the potential to store 17,707 m³ of water and have sequestered 17,409 t of soil carbon, which was equivalent to 63,890 t CO₂ removed from the atmosphere. The approach, making it possible to post-process the results with reference to different spatial domains, allowed also verifying that the SESs potential supply in green urban areas depends on soil functional recovery after anthropic disturbance, highlighting that urban green spaces may possess very different functionality levels in terms of the potential supply of specific ecosystem services, as already pointed out in other studies in different urban contexts [18,52,53,54,55,56], and that soil quality assessment should be one of the core elements of designing urban green infrastructure [57,58,59].

To characterize and rank the areas of potential urban development in terms of loss of potential ecosystem service supply in case of future urbanization, a soil quality index IQ4 was calculated considering four ecosystem services selected with the local stakeholders, namely BUF, CST, PRO and WAR. The definition of indices of urban soil quality demands a balance between comprehensiveness and feasibility: the selection should from one side encompass a sufficient range of soil processes relevant to the environmental quality of the urban area and on the other be realistic in terms of data inputs required for their assessment. According to Lehmann and Stahr [60,61], a multi-level approach based on four different dimensions of soil multifunctionality, i.e., hazard protection, production, environmental quality and cultural heritage, should be integrated in the planning process. Quite surprisingly though, the required information would be derived from available data, i.e., geological maps, hydrological maps, historical information, and information on building ground, as soil mapping was considered only necessary for calibration. Tresch et al. [62] pointed out that understanding soil quality in urban ecosystems would need multi-indicator frameworks to capture the complexity of soil characteristics and the influencing factors in space and time. In their study on the urban soil quality in Zurich, Switzerland, they resorted to a multivariate soil quality assessment considering a set of 44 soil quality indicators (17 biological, 9 chemical, 10 physical and 8 addressing SOM) measured at 170 sites in 85 city gardens along a gradient of urban density. Although providing a deep insight into the processes determining the quality and the function of garden soils in the urban ecosystems, such a complex multivariate framework could hardly become part of the planning process. A quantitative assessment of soil quality was implemented by Mamehpour et al. [63] comparing two models under different urbanization scenarios in West Azerbaijan Province, northwestern Iran, based on a set of twenty-four soil attributes. Those included a combination of fertility, salinity, and sodicity attributes and heavy metals (0–50 cm) and were tested for 14 soil profiles of urban and non-urban fields, for which the two approaches returned equal rankings. More recently, Séré et al. [64] selected and tested soil quality indicators for 109 different urban soils located in seven cities of western Europe and under various land uses. Nine soil functions and sub-functions (SOM cycle, nutrient storage, nutrient retention capacity, nutrient recycling, vegetation support, water retention, water infiltration and biodiversity) were ranked in four classes from 0 to 3, resorting to 29 indicators derived from laboratory measurements of soil properties, profile descriptions and site observations, with the lowest scores for sealed soils and soils located in urban brownfields, whereas the highest were found for soils located in city parks or urban agriculture. Although based on an extensive set of soil data and encompassing a wider range of soil functions, the implementation of the scoring approach requires a solid soil science expertise, time and substantial budgets, making its integration in urban planning difficult.

In the case of Forlì, we combined an indicator-based approach with a spatially explicit geostatistical approach which allowed for incorporating spatial variability to analyze and visualize an overall soil quality index (IQ4) across urban landscapes and trying to address urban-planning specific challenges. In doing so, intermediate outcomes maps of basic (textural fractions, pH, organic C) and PTFs-derived soil properties were produced for the 0–30 cm reference layer. The selected SES indicators cover specific dimensions of soil multifunctionality which are deemed strategic by the Forlì municipality for urban and peri-urban environmental quality: hazard mitigation (WAR), food production (PRO), climate regulation (CST) and pollution attenuation (BUF). In agreement with the Regional Soil Survey staff, the selection of the indicators to be included in the quality index was also guided by the criteria of lower correlation between the four indicators in order to reduce the effect of existing synergies and trade-offs between services, as highlighted in Table 7, so as not to overemphasize the effect of existing service bundles [65]. Given the data requirement for assessing the four single indicators, the final IQ4 score relies on a total of nineteen different soil properties. The average soil quality score for the whole urban area is equal to 2.42 (±0.12): 44% of the urban soil units have an average score lower than the average, encompassing 32% of the urban core area (908.3 ha). The soil units with the lowest quality score are those characterized by the presence of urbanized (IQ4 avg. 1.88, std. dev. 0.21) and highly disturbed soils (IQ4 avg. 2.06, std. dev. 0.10), whose score are in all cases the lowest for PRO and in some units also for WAR. Undisturbed and disturbed soils have very similar IQ4 scores, equal, respectively to 2.53 and 2.51, but in the case of the disturbed soils, the observed variability, expressed in terms of standard deviation is almost double, i.e., 0.15 vs. 0.08, respectively.

In our approach, as suggested by Suleymanov et al. [66], we supported the geostatistical interpolation of soil properties using the information provided by the urban soil map whose units account for the degree of disturbance and differentiate into industrial, residential, recreational and other areas within the urban fabric. In this way, the inherent heterogeneity of the urban soils and of their properties was explicitly accounted for in the evaluation of the ecosystem services provided by urban soils and in the assessment of urban soil quality.

5. Conclusions

This work presented an approach to foster the integration of urban soils and their potential ecosystem services supply into the practice of urban planning, allowing simultaneously the quantification of soil-based ecosystem services loss due to urbanization. Spatially explicit knowledge of the properties of urban and peri-urban soil allowed the implementation of an ecosystem service assessment framework originally designed for agricultural soils. This required on one hand the identification of urban soil typological units characterized by different degrees of anthropic disturbance and on the other the spatial estimation and mapping of soil properties measured at sampling points. Both pieces of information were provided by an ad hoc survey of urban and peri-urban soils which complemented the existing soil information for the agricultural soils surrounding the urban area. Once integrated into the general urban plan, the classed soil quality index allowed the identification of high-quality soil areas whose urbanization should be avoided or limited, providing at the same time the basis for compensating for further soil loss by de-sealing urbanized areas comparable in terms of overall soil quality or in the provision of targeted ecosystem services. Results showed that notwithstanding a substantial degree of anthropic disturbance, the supply of regulating services can still be of good quality also in urbanized areas and then relevant if not crucial in terms of extreme events mitigation.

To date, 166 municipalities out of 330 in Emilia–Romagna requested from the regional soil services the information on the ecosystem services provided by agricultural and forest soil in the areas of their administrative competence to integrate them into the general urban plans. Such information resulted from the implementation of the SES assessment framework presented in this work to the whole region. Further research, though, is required soon to assess the effectiveness of the integration of such information in the PUGs to verify if and to which extent it provided stakeholders and planners with useful knowledge to preserve, or eventually restore, the multifunctionality of urban soils in terms of ecosystem service supply and possibly trigger further ad hoc soil surveys in the main urban areas of the region.