Using municipal shelter data: what the records show, and what residents must confirm

A shelter dataset can help locate recorded facilities or study geographic coverage. It cannot, by itself, tell a resident whether a door is open, a route is accessible or a protected space is currently usable. For immediate instructions, use Home Front Command and the relevant municipality’s current information. This guide explains what two inspected municipal datasets contain and where their usefulness ends.

A correction comes first. Earlier Lions of Zion reports claimed a nationwide expansion of civil-defence data corroborated by the Jerusalem Post. The two source URLs returned 404 in the September 8 check, and the quoted dataset identifier was not found through the government catalog. The expansion claim has been withdrawn. A broken link does not prove that a page never existed, but the evidence retrieved here does not support the assertion. Real municipal datasets do not retroactively validate that alleged report.

Be’er Sheva — useful for mapping recorded points. The city’s government-catalog entry describes shelter locations and codes. It offers spreadsheet, CSV and geographic formats. The inspected CSV DataStore returned 262 rows, with shelter-name/code and latitude/longitude fields. That is a row count in this resource, not a verified count of currently open, accessible shelters. The resource metadata showed July 28, 2026 as its last modification date. The public catalog did not state a fixed refresh schedule; the underlying metadata describes automated updating without a constant interval.

Ma’ale Adumim — useful for reading address associations. The municipal catalog describes a shelter list. Its inspected DataStore returned 528 rows with neighborhood, shelter-address and associated-address fields. The resource’s last-modified timestamp was February 9, 2023; catalog metadata changed in March 2023. Although the metadata describes manual annual updates, that label does not establish that a new file was actually supplied every year. Nor does a row establish the facility’s present access arrangements or distinguish every form of public and private protection.

The two counts should not be compared as readiness scores. Different schemas, geographic coverage, definitions and update histories can produce different row totals. One file emphasizes coordinates and shelter codes; the other includes address associations. Neither inspected schema supplies a live opening-status field, a current inspection certificate or a complete accessibility assessment. A larger number of rows is not evidence that residents are better protected.

For residents, start with the municipality’s current shelter information and confirm the practical details there. Tel Aviv–Yafo’s municipal shelter page, for example, directs inquiries to its 106+ service. Its interactive listing was not fully exposed in the text retrieval used for this review, so this guide does not certify each entry in that list. The IDF’s information directory identifies Home Front Command’s 104 line for protection policy and instructions. The Home Front Command website itself returned an access restriction during this check; no current alert instructions have been copied or inferred from that failed retrieval.

For researchers, preserve the dataset name, publisher, resource format and retrieval date alongside the downloaded file. Record the resource timestamp separately from the date the catalog page changed. Check field definitions, duplicate rows and missing values before counting facilities or combining cities. For geographic analysis, verify the coordinate reference system of the specific export before plotting it. Keep the original data so later corrections can be traced.

Refresh labels need evidence. “Automatic” describes a mechanism, not a guarantee of freshness. “Annual” describes an intended interval, not proof of delivery. The practical check is whether the actual resource changed and whether the publisher explains what changed. A recent catalog timestamp could reflect a description edit rather than a new inspection of every shelter.

Coverage is also limited by what was published. These two resources do not establish a complete national inventory of public shelters, private shared shelters, residential protected rooms or institutional protected spaces. Absence from a dataset is not proof that a location has no protection. Presence is not a substitute for a current condition check. Reporting national coverage would require defined categories, comparable municipal submissions and a documented accounting of missing areas.

The useful distinction is between a reference layer and operational information. Open data supports mapping and scrutiny; current official and municipal channels support decisions about access and instructions. This guide keeps those roles separate and identifies the records actually inspected, rather than promising nationwide completeness or live readiness.