Public water in data

1The sample

The dataset merges points imported from OpenStreetMap with community reports. Every analysis on water status considers verified observations only: photo-backed reports approved one by one by the moderators. The study perimeter is Italy; international coverage gets its own closing section.

Origin of the mapped points: 45,296 from OpenStreetMap, 1,117 added directly by the community.

2Methodology

1. Collection: app users report the state of a fountain with a photo. Each report records water status, position and date.
2. Verification: every report goes through manual moderation. Only confirmed reports become observations and enter the analyses; rejected and pending ones are excluded.
3. Unit of analysis: the verified observation, classified into three states — drinkable water, no water, non-drinkable water.
4. Seasons: the meteorological convention is used (winter = December-February, spring = March-May, summer = June-August, autumn = September-November).
5. Geographic perimeter: the analyses cover fountains geolocated in Italy only. Points abroad are summarised in a dedicated section, with no detailed analysis.

3Seasonality of water status

Observations from all years are aggregated by meteorological season. The percentage composition by status reveals whether fountain reliability follows the seasonal cycle, for instance winter anti-freeze shutdowns or summer water crises.

In the sample, the share of “no water” observations peaks in Winter (33.0% of the seasonal observations).

The same signal, read month by month with all years pooled, pinpoints the peaks the four seasonal blocks smooth out. Months without observations are left blank.

4Year-over-year evolution

The same composition, read year by year, tracks the trajectory of the network over time. Variations reflect both the real state of the fountains and the growth of the community: recent years rest on a much broader base of observations.

5Status dynamics over time

For fountains with at least two verified observations, consecutive observation pairs are analysed (66 pairs in the sample). The transition matrix estimates the probability that a fountain found in one state is found in the same state, or a different one, at the next observation.

Reconfirmation decays with time: among fountains found drinkable, the curve shows the share found drinkable again at the next visit, as a function of the interval between the two observations.

In the observed recoveries, a median of 145 days passes between a failure observation and the first drinkable one. The value also reflects how often the community revisits, not just repair times: read it as an upper bound.

6Freshness of the observations

For every observed fountain we measure the age of its last verified observation. The distribution quantifies how recent the published data is: an observation describes the fountain at the time of the visit, and ages with it.

Median age of the last observation: 98 days. 3.7% of the observed fountains have not received a verified observation in over two years: this is where new reports are most valuable.

7Community activity

The monthly series of verified observations measures the pace of data collection. Downstream, the moderation funnel documents the verification process across all received reports, fountains and public toilets included.

Confirmation rate over decided reports: 99.6%.

8Territorial distribution

Grouping the regions into the three ISTAT macro areas (South and Islands merged) makes the territorial comparison readable at a glance. The composition is computed on the verified network of each area.

Italian regions ranked by mapped fountains. The drinkable share is computed on fountains with a verified status only, so regions with few verifications express less stable estimates.

The coverage is concentrated: the top 10 municipalities hold 19.1% of the mapped italian fountains, with a median of 1 fountains per municipality. The Gini index of the distribution is 0.502 (0 = evenly spread, 1 = maximum concentration).

9The rest of the world

Outside Italy the coverage is thinner and verifications are few: the numbers below describe the extent of the map, without the detailed analyses reserved for the italian perimeter.

Countries with the most mapped fountains: Spain (11,506), France (8,060), Switzerland (4,716), Austria (2,448), Germany (1,989), Hungary (1,663).

10Community ratings

Beyond water status, observations may include a 1 to 5 rating. The distribution of verified ratings measures the perceived quality of the fountains.

Average rating: 4.31 out of 5, computed over 338 verified ratings.

11Limitations

• The sample is not random; observed fountains are the ones users visit, with a likely over-representation of urban and touristic areas.
• Observation density grows with the community; year-over-year comparisons also reflect this growth, not only the state of the network.
• The recorded status is the one at visit time; between two observations the real status may change several times without leaving a trace.
• Moderator verification assesses the credibility of the report; it does not certify chemical analyses of the water.