Geoportal: SDI Architecture, Metrics and Guide

Q: What is the difference between a geoportal and a spatial data infrastructure (SDI)?

A Spatial Data Infrastructure (SDI) is the overarching framework encompassing institutional policies, governance, data agreements, technical standards, and computing infrastructure that coordinates spatial data sharing across organizations. A geoportal is the specific, web-accessible catalog interface and client application that serves as the visible entry point to the SDI, enabling users to search, preview, and download spatial datasets.

Q: What is the difference between OGC CSW and OGC API – Records?

OGC Catalogue Service for the Web (CSW 2.0.2) is a mature, XML/SOAP-based catalog protocol defining standard operations (GetCapabilities, GetRecords, DescribeRecord) to query ISO 19139 metadata. In contrast, OGC API – Records is the modern, cloud-native successor built on RESTful principles, OpenAPI specifications, and JSON/GeoJSON payloads, enabling seamless integration with modern web search engines and browser clients.

Q: How do geoportals harvest metadata from distributed repositories?

Geoportals employ automated harvesters that execute scheduled CSW GetRecords requests or Web Accessible Folder (WAF) crawls against remote server endpoints. The harvester extracts ISO 19139 XML metadata, validates syntax against schema profiles, extracts spatial bounding boxes and keywords, and indexes records into high-performance search databases such as Elasticsearch or PostgreSQL/PostGIS for faceted discovery.

Q: What open-source software packages are used to deploy geoportals?

Common open-source geoportal solutions include GeoNetwork opensource (a mature Java-based catalog supporting ISO 19115 and CSW), pycsw (an OGC-certified lightweight Python CSW server), CKAN with the ckanext-spatial extension (an enterprise open data portal adapted for geospatial discovery), and GeoNode (an integrated Django-based platform combining GeoServer, pycsw, and web mapping interfaces).

A geoportal is a centralized, web-based gateway that enables researchers, geospatial analysts, and data managers to discover, view, query, and access geospatial datasets and cartographic web services across a Spatial Data Infrastructure (SDI). In geographic information science and research data management, geoportals operate as the discovery tier of distributed geospatial architectures, federating metadata compliant with ISO 19115 standards and executing query protocols defined by the Open Geospatial Consortium (OGC), including Catalogue Services for the Web (CSW) and OGC API – Records.

Definition in one paragraph

A geoportal is a web-based catalog application and discovery gateway that provides an authoritative, single point of access to query, evaluate, visualize, and retrieve spatial datasets and services organized within a Spatial Data Infrastructure. Built upon international standards such as ISO 19115-1:2014 for geographic metadata and OGC Catalogue Services, a geoportal harvests and indexes structured metadata from distributed repositories, enabling spatial, temporal, and thematic querying without requiring users to download massive raw GIS archives or maintain local server infrastructure. In academic and institutional workflows, geoportals integrate heterogeneous spatial assets—such as topographic lidar elevation models from OpenTopography, climate datasets following CF Conventions, cadastral surveys, and municipal planning boundaries—bridging enterprise platforms like ArcGIS Hub with cross-domain data catalog profiles such as DCAT-AP.

Formula and mathematical foundations

Evaluating the indexing quality, query precision, and operational reliability of a geoportal relies on three primary quantitative metrics: the Metadata Completeness Index (MCI), spatial bounding box Intersection over Union (IoU), and network service availability.

1. Metadata Completeness Index (MCI): The completeness of a single catalog record across K mandatory ISO 19115 / INSPIRE metadata elements is defined as:

MCI = (1 / K) × ∑_k=1^K m_k × 100%

where K is the total number of mandatory elements evaluated (e.g., K = 10: Resource Title, Abstract, Resource Type, Locator URL, Unique Identifier, Topic Category, Geographic Bounding Box, Temporal Extent, Lineage, and Responsible Party) and m_k ∈ {0, 1} is a binary indicator where m_k = 1 if element k is validly populated and m_k = 0 if missing or null. For a collection of N records, the collection-level completeness is:

MCI_mean = [1 / (N × K)] × ∑_i=1^N ∑_k=1^K m_i,k × 100%

2. Spatial Query Bounding Box Overlap (Intersection over Union, IoU): Spatial retrieval relevance between a query bounding box B_q and a dataset bounding box B_d is measured via the Jaccard spatial overlap:

IoU(B_q, B_d) = Area(B_q ∩ B_d) / Area(B_q ∪ B_d) = Area(B_q ∩ B_d) / [Area(B_q) + Area(B_d) − Area(B_q ∩ B_d)]

where B_q = [x_q,min, y_q,min, x_q,max, y_q,max] and B_d = [x_d,min, y_d,min, x_d,max, y_d,max] denote coordinate extents in decimal degrees (EPSG:4326), and B_int = [max(x_q,min, x_d,min), max(y_q,min, y_d,min), min(x_q,max, x_d,max), min(y_q,max, y_d,max)]. If min > max on either axis, Area(B_q ∩ B_d) = 0.

3. Network Service Availability (A): Under European Commission Regulation (EC) No 976/2009 for INSPIRE discovery services, geoportals must maintain a minimum service availability of 99.0%:

A = [1 − (T_downtime / T_total)] × 100%

where T_downtime is cumulative unscheduled downtime and T_total is total scheduled operating duration (for 30 days, T_total = 30 × 24 × 60 = 43,200 minutes).

Worked example with real numbers and arithmetic shown

An institutional SDI assessment evaluates an operational geoportal catalog holding N = 250 spatial dataset records across K = 10 mandatory ISO 19115-1 elements, tests spatial bounding box retrieval, and evaluates monthly service uptime.

Step 1: Collection Metadata Completeness. The 250 catalog records encompass a total possible score of N × K = 250 × 10 = 2,500 element evaluations. Automated catalog validation through the geoportal CSW endpoint identifies 2,190 valid mandatory elements (∑_i=1²⁵⁰ ∑_k=1¹⁰ m_i,k = 2,190). The mean collection completeness is:

MCI_mean = (2,190 / 2,500) × 100% = 0.8760 × 100% = 87.60%

For a specific record (Record #42, “Metropolitan Land Use Layer”), 9 of the 10 elements are populated (Lineage statement missing):

MCI₄₂ = (9 / 10) × 100% = 90.0%

Step 2: Spatial Query Overlap (IoU). A researcher executes a bounding box search for a municipal study zone B_q = [−74.05°, 40.65°, −73.85°, 40.85°]:

Width_q = −73.85 − (−74.05) = 0.20°; Height_q = 40.85 − 40.65 = 0.20°
Area(B_q) = 0.20 × 0.20 = 0.0400 deg²

Record #42 encompasses a regional extent B_d = [−74.15°, 40.60°, −73.90°, 40.80°]:

Width_d = −73.90 − (−74.15) = 0.25°; Height_d = 40.80 − 40.60 = 0.20°
Area(B_d) = 0.25 × 0.20 = 0.0500 deg²

The intersection bounding box B_int is:

x_int,min = max(−74.05, −74.15) = −74.05°; y_int,min = max(40.65, 40.60) = 40.65°
x_int,max = min(−73.85, −73.90) = −73.90°; y_int,max = min(40.85, 40.80) = 40.80°
Width_int = −73.90 − (−74.05) = 0.15°; Height_int = 40.80 − 40.65 = 0.15°
Area(B_q ∩ B_d) = 0.15 × 0.15 = 0.0225 deg²

The union area and spatial IoU are computed as:

Area(B_q ∪ B_d) = 0.0400 + 0.0500 − 0.0225 = 0.0675 deg²

IoU(B_q, B_d) = 0.0225 / 0.0675 = 1 / 3 ≈ 0.3333 (or 33.33%)

Step 3: Service Availability. Over a 30-day monitoring window (T_total = 43,200 minutes), 152 minutes of unscheduled downtime occurred:

A = [1 − (152 / 43,200)] × 100% = [1 − 0.0035185] × 100% = 99.6481% ≈ 99.65%

Because 99.65% ≥ 99.00%, the geoportal conforms to the legal availability standard of Regulation (EC) No 976/2009.

Interpretation and reading guidance

Geoportal evaluation metrics guide data administrators and researchers in benchmarking repository health, data usability, and harvesting eligibility:

Metric	Threshold Value	Operational Classification	Interpretation and Action
Metadata Completeness (MCI)	≥ 90.0%	High Conformity / Production Ready	Records are fully discoverable and ready for federated OGC CSW harvesting without remediation.
Metadata Completeness (MCI)	75.0% – 89.9%	Moderate Deficit / Audit Warning	Missing secondary elements (such as lineage or temporal bounds); requires remediation prior to international federation.
Metadata Completeness (MCI)	< 75.0%	Non-Conforming / Harvest Blocked	Critical metadata omitted; records should be suppressed from public CSW endpoints until corrected.
Spatial Overlap (IoU)	≥ 0.700	High Spatial Congruence	Dataset boundary tightly aligns with the target study area; minimal extraneous features downloaded.
Spatial Overlap (IoU)	0.250 – 0.699	Partial Spatial Coverage	Dataset encloses or overlaps study area; spatial subsetting or clipping required prior to analysis.
Spatial Overlap (IoU)	< 0.250	Peripheral Boundary Overlap	Incidental corner contact; high processing overhead for negligible spatial overlap.
Service Availability (A)	≥ 99.00%	INSPIRE Compliant	Satisfies legal discovery service criteria; stable for automated research pipelines and harvesting.
Service Availability (A)	< 99.00%	Service Level Deficit	Violates regulatory availability threshold; automated harvesters risk timeout and synchronization failures.

Assumptions and common mistakes

Rigorous geoportal utilization requires avoiding several widespread technical and methodological errors:

Assuming Coordinate Reference Systems Align Automatically: Spatial bounding box queries return false positives or empty results when queries in WGS84 (EPSG:4326) are executed against catalog databases indexed in projected coordinate reference systems (e.g., UTM or State Plane) without explicit datum transformations.
Equating Metadata Presence with Data Availability: An indexed ISO 19115 metadata record in a geoportal does not guarantee that underlying Web Map Service (WMS) or Web Feature Service (WFS) endpoints are active. Complete audits must verify that resource locators return HTTP 200 responses with valid capabilities XML documents.
Mishandling Antimeridian and Polar Bounding Boxes: Bounding boxes crossing the 180th meridian or enclosing polar regions generate inverted or invalid bounding boxes if software naively assumes x_min ≤ x_max without antimeridian normalization.
Omitting Temporal Extent Constraints: Querying spatial datasets solely by geographic coordinates often returns obsolete historical survey revisions. Geoportal filters must combine spatial predicates with ISO 8601 temporal bounds to ensure contemporary data retrieval.
Conflating Metadata Profiles: Mixing ISO 19115:2003, ISO 19115-1:2014, and US FGDC CSDGM schema models leads to XML validation failures during federated CSW harvesting.

How to report it (APA example)

When documenting geoportal evaluations and spatial catalog audits in academic publications, present methodology and findings following APA 7th guidelines:

“A comprehensive metadata completeness and service performance audit of the regional spatial data infrastructure (SDI) geoportal evaluated N = 250 spatial dataset records across K = 10 mandatory metadata elements defined by ISO 19115-1:2014 and Commission Regulation (EC) No 1205/2008. The catalog exhibited a mean Metadata Completeness Index of 87.60% (SD = 8.42%). Core identification elements (Resource Title, Unique Identifier, and Geographic Bounding Box) demonstrated 100.0% conformity (n = 250), whereas Lineage statements showed the highest omission rate (34.0% non-compliant, n = 85). Spatial retrieval efficiency evaluated across 50 municipal bounding box queries yielded a mean Intersection over Union (IoU) of 0.482 (SD = 0.164). Network service monitoring over a continuous 30-day period (T = 43,200 min) recorded 152 minutes of unscheduled downtime, corresponding to a service availability of 99.65%, satisfying the minimum 99.00% operational requirement mandated by Commission Regulation (EC) No 976/2009.”

How to calculate it in Excel, R or Python

The following reproducible scripts demonstrate how to compute the Metadata Completeness Index (MCI) and spatial bounding box Intersection over Union (IoU) across standard software environments.

Excel Calculation

For a dataset record with 10 mandatory metadata elements recorded as binary flags (1 = present, 0 = absent) in cells B2:K2:

=AVERAGE(B2:K2) * 100

For spatial bounding box Intersection over Union with a query box in A2:D2 (x_q,min, y_q,min, x_q,max, y_q,max) and a dataset box in E2:H2 (x_d,min, y_d,min, x_d,max, y_d,max):

Intersection Width:  =MAX(0, MIN(C2, G2) - MAX(A2, E2))
Intersection Height: =MAX(0, MIN(D2, H2) - MAX(B2, F2))
Intersection Area:   =I2 * J2
Union Area:          =((C2 - A2) * (D2 - B2)) + ((G2 - E2) * (H2 - F2)) - K2
Spatial IoU:         =K2 / L2

R Calculation

# Compute Metadata Completeness Index (MCI) and Spatial IoU in R
compute_mci <- function(record, mandatory_fields) {
  present <- sum(!is.na(record[mandatory_fields]) & record[mandatory_fields] != "")
  (present / length(mandatory_fields)) * 100
}

compute_bbox_iou <- function(box_q, box_d) {
  # Format: c(min_x, min_y, max_x, max_y)
  int_w <- max(0, min(box_q[3], box_d[3]) - max(box_q[1], box_d[1]))
  int_h <- max(0, min(box_q[4], box_d[4]) - max(box_q[2], box_d[2]))
  area_int <- int_w * int_h
  
  area_q <- (box_q[3] - box_q[1]) * (box_q[4] - box_q[2])
  area_d <- (box_d[3] - box_d[1]) * (box_d[4] - box_d[2])
  area_union <- area_q + area_d - area_int
  
  if (area_union > 0) area_int / area_union else 0
}

# Example verification
fields <- c("title", "abstract", "type", "locator", "identifier",
            "topic", "bbox", "temporal", "lineage", "contact")
sample_rec <- list(title = "Metropolitan Land Use", abstract = "Parcels", type = "dataset",
                   locator = "https://geoportal.org/d1", identifier = "urn:uuid:1",
                   topic = "planningCadastre", bbox = c(-74.15, 40.60, -73.90, 40.80),
                   temporal = "2024", lineage = "", contact = "[email protected]")

mci_score <- compute_mci(sample_rec, fields)
iou_score <- compute_bbox_iou(c(-74.05, 40.65, -73.85, 40.85), sample_rec$bbox)
cat(sprintf("MCI: %.1f%% | Spatial IoU: %.4fn", mci_score, iou_score))

Python Calculation

# Compute Metadata Completeness Index (MCI) and Spatial IoU in Python
def compute_mci(metadata_dict, mandatory_fields):
    """Calculates percentage completeness across mandatory metadata fields."""
    present = sum(1 for f in mandatory_fields if metadata_dict.get(f))
    return (present / len(mandatory_fields)) * 100.0

def compute_bbox_iou(box_q, box_d):
    """Calculates Intersection over Union for bounding boxes: [min_x, min_y, max_x, max_y]."""
    int_w = max(0.0, min(box_q[2], box_d[2]) - max(box_q[0], box_d[0]))
    int_h = max(0.0, min(box_q[3], box_d[3]) - max(box_q[1], box_d[1]))
    area_int = int_w * int_h
    
    area_q = (box_q[2] - box_q[0]) * (box_q[3] - box_q[1])
    area_d = (box_d[2] - box_d[0]) * (box_d[3] - box_d[1])
    area_union = area_q + area_d - area_int
    
    return area_int / area_union if area_union > 0 else 0.0

fields = ["title", "abstract", "type", "locator", "identifier",
          "topic", "bbox", "temporal", "lineage", "contact"]
record_42 = {
    "title": "Metropolitan Land Use and Parcel Boundary Layer",
    "abstract": "High-resolution parcel-level cadastral survey.",
    "type": "dataset",
    "locator": "https://geoportal.example.org/layers/parcels",
    "identifier": "urn:uuid:7c9e6679-7425-40de-944b-e07fc1f90ae7",
    "topic": "planningCadastre",
    "bbox": [-74.15, 40.60, -73.90, 40.80],
    "temporal": "2024-01-01/2024-12-31",
    "lineage": None,
    "contact": "[email protected]"
}

mci = compute_mci(record_42, fields)
iou = compute_bbox_iou([-74.05, 40.65, -73.85, 40.85], record_42["bbox"])
print(f"MCI: {mci:.1f}% | Spatial IoU: {iou:.4f}")

Frequently Asked Questions

What is the difference between a geoportal and a spatial data infrastructure (SDI)?