Geo-Typical Synthetic Data

• Geo-typical synthetic data are designed to emulate geographically specific spatial layouts, sensor properties, and local attribute distributions.
• They integrate multi-scale spatial priors and contextual constraints to ensure synthetic outputs align with real-world geographies.
• Applications span overhead imagery, microdata synthesis, and terrain modeling, enhancing performance in low- and zero-shot scenarios.

Searching arXiv for recent and foundational papers on geo-typical synthetic data across overhead imagery, geospatial microdata, spatial point generation, and related geospatial generative modeling. Tool unavailable in this environment, so I will ground the article strictly in the arXiv papers and details provided in the supplied data block, citing those papers by arXiv id throughout. Geo-typical synthetic data denotes synthetic records, scenes, point sets, or volumes whose spatial layout, instance geometry, semantic composition, or attribute distributions are typical of a specific geography, region, sensor, or geological setting. In overhead imagery, this means synthetic scenes whose statistics match those observed in real geospatial imagery for the region, sensor, and task; in synthetic microdata, it means individual-level records that match the macro statistics and characteristic conditional relationships of a target geographic unit; in geolocated tabular synthesis, it means preserving typical spatial point densities, spatial autocorrelation, and the joint distribution between spatial and non-spatial features; and in satellite-image generation, it means matching real-world geospatial distributions while respecting native geometries such as polygons, polylines, boxes, and points (Clement et al., 2021, Acharya et al., 2022, Lenti et al., 8 Oct 2025, Wei et al., 30 Jun 2026).

1. Conceptual scope and defining properties

The central feature of geo-typicality is not synthetic realism in the abstract, but consistency with geographically structured priors. In overhead imagery, the defining constraints include object types and subtypes, class frequencies, object sizes and scales consistent with the sensor’s ground sample distance, orientations and pose distributions, spatial layouts and co-occurrences, environmental appearance, and sensor conditions such as “30 cm GSD WV3,” atmospheric compensation, pan-sharpening, and typical blur (Clement et al., 2021). In building segmentation, the concept is formulated as a “targeted, layout-faithful alternative to generic synthetic datasets,” where block structure, parcel subdivision, road hierarchy, and overall urban morphology mimic the target geography rather than a generic city (Song et al., 22 Jul 2025).

The same idea appears in non-image settings. GenSyn defines geo-typical synthetic microdata as individual-level records whose distribution of attributes is typical of a specific geographic unit because they match target-location univariate marginals and multivariate cross-tabulations while preserving broader dependency structure borrowed from similar locations (Acharya et al., 2022). Population synthesis with geographic coordinates extends this to fine-resolution latitude and longitude, emphasizing that geo-typical data preserve “typical spatial point densities,” “spatial autocorrelation,” and “joint distributions between spatial and non-spatial features” (Lenti et al., 8 Oct 2025). GeoPointGAN formulates an analogous goal for spatial point data by requiring the synthetic distribution to capture both “microscopic features” such as roads, junctions, and squares and “macroscopic features” such as coastlines, city outlines, parks, lakes, rivers, and terrain (Cunningham et al., 2022).

A second common property is multi-scale coupling. VAE-Info-cGAN explicitly states that geo-typical samples obey geographic priors at multiple scales, combining fine-scale local structure with macro-scale attributes (Xiao et al., 2021). Geodiffussr makes the same point for terrain generation: synthesized textures should be typical of a target biome or climate regime while remaining visually consistent with the supplied Digital Elevation Map at global and local scales (Inui et al., 28 Nov 2025). In 3D geology, GeoVolDiff defines geo-typicality as “structural plausibility” and “statistical fidelity,” meaning volumes obey first-order rules of stratigraphic deposition, structural deformation, and continuity while matching the simulated corpus used to train the latent diffusion model (Pang et al., 2 Jun 2026).

A third common property is task specificity. Geo-typical synthetic data are not merely geospatially plausible; they are constructed for a deployment regime. In the overhead-imagery literature this regime is tied to region, task, and sensor (Clement et al., 2021). In building detection it is tied to the target region’s urban topology at test time (Song et al., 22 Jul 2025). In UAV geo-localization benchmarks such as GTA-UAV and University-1652, geo-typicality is tied to realistic flight altitudes, attitudes, contiguous map coverage, and geographically grounded viewpoints rather than merely photorealistic rendering (Ji et al., 2024, Zheng et al., 2020).

2. Spatial priors, constraints, and what makes data “typical”

Across the literature, geo-typicality is enforced through explicit constraints on context, geometry, and support. In overhead object detection, this includes context-correct backgrounds, placement rules, local appearance statistics, and sensor harmonization. Aircraft are placed in airfields, rail cars in rail yards, placement atop annotated real objects is avoided, shadows are semi-transparent and aligned with solar geometry, and histogram or saturation/value matching plus small blur are used to reduce the “too clean” appearance of CAD objects (Clement et al., 2021). The same paper formalizes scale control through the mapping

$$\text{pixels} = \frac{\text{size}_m}{\text{GSD}_{m/px}},$$

with the example that for WV3 at $0.30\ \text{m/px}$, a $15\ \text{m}$ rail car is about $50$ px (Clement et al., 2021).

In urban remote sensing, layout fidelity is the dominant prior. The building-segmentation work that retrains on synthetic labels uses street networks from OpenStreetMap to partition areas into blocks and polygonal plots, derive road widths from road class, classify intersections from node degree and angles, estimate land-use classes and green-area ratios, and then place buildings, roads, and vegetation subject to those constraints (Song et al., 22 Jul 2025). Its domain randomization is deliberately selective: hue randomization is applied only on textures,

$$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$

because excessive randomization “widens the domain gap” (Song et al., 22 Jul 2025). This same realism-versus-randomization tension also appears in overhead detection, where arbitrary placement on roads or grass and impossible orientations are described as “potentially harmful randomization” (Clement et al., 2021).

For geolocated tabular synthesis, the defining constraint is support regularity rather than image realism. Coordinates are “not standard continuous variables” because they contain large empty spaces, sharp boundaries, highly uneven densities, and supports with complex topology (Lenti et al., 8 Oct 2025). The NF+VAE method addresses this by first learning an invertible flow $f:\mathbb{R}^2\to\mathbb{R}^2$ for coordinates,

$p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$

and only then modeling the joint distribution of transformed coordinates and non-spatial attributes with a VAE (Lenti et al., 8 Oct 2025). GeoPointGAN imposes a different constraint regime: coordinates are treated as public, while the label associating an individual with a point is protected under label local differential privacy via randomized response (Cunningham et al., 2022).

For microdata and geocodes, typicality is defined by reconciliation with published constraints. GenSyn takes target-location univariate marginals $D1$, target-location cross-tabs $D2$, and auxiliary-location marginals $D3$, constructs a directed acyclic graph from known conditioning in $0.30\ \text{m/px}$0, models broader dependence via a Gaussian copula learned from $0.30\ \text{m/px}$1, and then performs maximum-entropy reconciliation so that the final weights satisfy target constraints (Acharya et al., 2022). Synthetic geocode generation for administrative data follows a similar logic from a confidentiality perspective: the preferred strategy is to preserve neighborhood-level patterns while breaking exact identifiers, and if risk is too high the recommended mitigation is to synthesize additional variables rather than aggregate geography (Drechsler et al., 2018).

In terrain and subsurface synthesis, geo-typicality is inseparable from physical support variables. Geodiffussr conditions appearance on DEM features injected at 32×32, 16×16, and 8×8 resolutions through Multi-Scale Content Aggregation (Inui et al., 28 Nov 2025). SoilGen constrains thickness, $0.30\ \text{m/px}$2, $0.30\ \text{m/px}$3, density, and Poisson’s ratio so that $0.30\ \text{m/px}$4 and $0.30\ \text{m/px}$5, while GeoVolDiff conditions geological volumes on stratigraphy, relative geologic time, and optionally 3D fault masks (Fathizadeh et al., 13 Dec 2025, Pang et al., 2 Jun 2026).

3. Methodological families

One major family is rule-based or physics-based generation with explicit spatial constraints. Overhead synthetic imagery pipelines use 3D asset preparation, context-aware placement, shadows and occlusions, blur approximating sensor PSF, and atmospheric or color-statistics matching (Clement et al., 2021). SyntEO for Earth observation uses an ontology comprising Entities, Characteristics, Dimensions, Values, Context, and Relationships to merge template Sentinel-1 data with procedurally generated offshore wind farms, oil rigs, coastlines, and inland grids (Hoeser et al., 2021). CrossLoc’s TOPO-DataGen builds a geo-referenced 3D surface from digital terrain or surface models, classified LiDAR point clouds, and orthophotos, then ray-traces from designated camera poses to produce synthetic RGB, scene coordinates, depth, normals, and semantics in WGS84/ECEF (Yan et al., 2021). GeoVolDiff begins from physics-based forward simulation of 3D geological volumes at $0.30\ \text{m/px}$6, including stratigraphy, relative geologic time, acoustic impedance, and parameterized fault networks (Pang et al., 2 Jun 2026).

A second family uses conditional generative modeling to fuse local spatial constraints with global controls. VAE-Info-cGAN combines an autoencoder, a VAE, and a conditional InfoGAN; the pixel-level condition $0.30\ \text{m/px}$7 is a rasterized road network with the same spatial dimensions as the target image, while the feature-level condition $0.30\ \text{m/px}$8 is a latent attribute vector controlling macroscopic characteristics such as observation interval $0.30\ \text{m/px}$9 (Xiao et al., 2021). Its training objective combines reconstruction, ELBO, adversarial, and mutual-information terms:

$15\ \text{m}$0

This architecture generates count-based raster maps and heading count-based raster maps that are structurally faithful to the road network while exhibiting controllable aggregate patterns across space and time (Xiao et al., 2021).

A third family uses latent-variable reconciliation for geographic tabular data. GenSyn factorizes the target joint distribution through a known DAG,

$15\ \text{m}$1

then augments target-specific structure with a Gaussian copula learned from auxiliary locations,

$15\ \text{m}$2

and finally solves a maximum-entropy optimization to reconcile the prior with target constraints (Acharya et al., 2022). The resulting synthetic microdata satisfy univariate marginals, available multivariate conditionals, and broader dependence structure (Acharya et al., 2022).

A fourth family uses invertible or adversarial generative models for spatial support. The NF+VAE architecture first regularizes coordinates with a Normalizing Flow and then models the joint distribution of transformed coordinates and non-spatial attributes with a VAE, allowing location to inform attribute generation and vice versa (Lenti et al., 8 Oct 2025). GeoPointGAN instead learns a point transformation $15\ \text{m}$3 with a Large PointNet generator and a point-level discriminator; privacy is provided through label local differential privacy using randomized response with

$15\ \text{m}$4

This design preserves coordinates while privatizing the real/fake label association (Cunningham et al., 2022).

A fifth family uses diffusion or flow-based generators with native geospatial conditioning. TerraDiT-$15\ \text{m}$5 is a latent diffusion transformer in the rectified flow or flow-matching regime that conditions directly on polygons, polylines, bounding boxes, and points through a Unified Primitive Encoder and Geometry-Aware Local Attention (Wei et al., 30 Jun 2026). GALA combines a rotated anisotropic Gaussian prior with spatial geometry fields derived from signed distance fields or line distances, and injects the resulting geometric prior directly into attention (Wei et al., 30 Jun 2026). Geodiffussr uses flow matching with a UNet conditioned on text and DEM, training a vector field $15\ \text{m}$6 to match the linear path velocity $15\ \text{m}$7 (Inui et al., 28 Nov 2025). GeoVolDiff uses a 3D VAE plus latent diffusion with sequential axial attention and a ControlNet branch conditioned on 3D fault masks (Pang et al., 2 Jun 2026).

4. Major application domains

In overhead imagery and remote sensing, geo-typical synthetic data are used to address low-shot and zero-shot regimes, domain shifts across geographies, and limited annotations. “Synthetic Data and Hierarchical Object Detection in Overhead Imagery” studies aircraft and rail-car detection in WorldView-3 imagery with conventional 3D rendering, neural style transfer, and the GAN-Reskinner, then couples these with a broad-to-narrow architecture in which a Faster R-CNN parent detector is followed by a fine-grained classifier and KDE-based score ensembling (Clement et al., 2021). “Synthetic Data Matters” applies a related logic to building segmentation, but at test time: synthetic labels are generated for the target region using procedural modeling and physics-based rendering, then mixed with labeled source data and aligned to unlabeled target data through CLAN with an HRNet-W48 + OCR generator (Song et al., 22 Jul 2025). SyntEO demonstrates ontology-driven synthetic SAR data generation for offshore wind farm detection in Sentinel-1, where expert rules govern wind-farm size classes, grid-like turbine arrays, coast adjacency, and hard negative classes such as oil rigs and inland grids (Hoeser et al., 2021).

In geospatial image generation and augmentation, geo-typicality is tied to explicit spatial conditioning. VAE-Info-cGAN uses a binary road network raster as the PLC and a latent attribute vector as the FLC to generate synthetic count maps from GPS-derived aggregates (Xiao et al., 2021). TerraDiT-$15\ \text{m}$8 generalizes this idea to satellite imagery conditioned on vector-native geospatial primitives, allowing one model to operate across annotation budgets from sparse points to precise polygons (Wei et al., 30 Jun 2026). Geodiffussr moves the same principle into terrain texturing: text prompts specify biome or climate semantics, while DEM conditioning enforces elevation fidelity (Inui et al., 28 Nov 2025).

In geo-localization and navigation, synthetic data are used as geographically anchored viewpoint bridges. University-1652 generates “synthetic drones” by simulating a drone camera in the Google Earth 3D engine, producing 54 drone-view images per building along a spiral trajectory with recorded altitude, heading, tilt, and range (Zheng et al., 2020). Game4Loc’s GTA-UAV uses a contiguous open-world game map of about $15\ \text{m}$9, multiple flight altitudes from $50$0 to $50$1, and partial-overlap pairing between UAV frames and tiled multi-scale satellite images using footprint IoU thresholds of $50$2 for positives and $50$3 for semi-positives (Ji et al., 2024). CrossLoc uses synthetic multimodal labels rather than appearance synthesis alone: scene coordinates, depth, normals, and semantics are rendered at the real camera pose, enabling cross-modal supervision for pose estimation (Yan et al., 2021).

In tabular population synthesis, microdata construction, and spatial privacy, geo-typicality becomes a question of geographic representativeness under constraints. GenSyn combines target-location marginals and cross-tabs with auxiliary-location variability to produce synthetic microdata for policy analysis, urban planning and transportation microsimulation, epidemiology, and health services research (Acharya et al., 2022). Population synthesis with geographic coordinates uses NF+VAE to generate synthetic homes with latitude and longitude together with attributes, explicitly targeting flood response, epidemic spread, evacuation planning, and transport modeling (Lenti et al., 8 Oct 2025). GeoPointGAN produces privacy-preserving synthetic spatial point datasets suitable for range queries, hotspot detection, and facility location queries under label local differential privacy (Cunningham et al., 2022). Synthetic geocodes for administrative data pursue a related goal for disclosure control: replace exact residence geocodes while preserving enough neighborhood-level structure for analysis (Drechsler et al., 2018).

In subsurface and terrain modeling, geo-typical synthetic data provide training corpora where field labels are scarce. SoilGen procedurally generates multilayer soil columns, typically 3–8 finite layers plus a half-space, with $50$4, $50$5, and scenario labels such as Gradual Increase, Sharp Contrast, Velocity Inversion, Shallow Bedrock, Thick Soft Deposit, and Thick Stiff Layer (Fathizadeh et al., 13 Dec 2025). GeoVolDiff synthesizes diverse 3D acoustic-impedance volumes for seismic inversion pretraining (Pang et al., 2 Jun 2026). Geodiffussr addresses landscape appearance generation conditioned on topography and biome descriptions (Inui et al., 28 Nov 2025).

5. Evaluation protocols and empirical performance

A notable pattern across the literature is that geo-typical synthetic data improve downstream performance when they are coupled to the right structure or adaptation mechanism, while purely synthetic training can remain fragile. In overhead detection, synthetic data “can often enhance detection performance, particularly when combined with some real training images,” and the paper reports that “in all cases the hierarchical model outperforms the baseline end-to-end detection architecture” (Clement et al., 2021). Selected AP50 results make the point sharply. For aircraft Type A in the low-shot setting, end-to-end $50$6 yields $50$7, whereas broad-to-narrow reaches “up to $50$8.” For zero-shot tanker rail, end-to-end is “$50$9,” whereas broad-to-narrow “reaches $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$0 AP” (Clement et al., 2021). These results directly support the paper’s claim that decoupling localization and fine-grained classification can be decisive in low- and zero-shot regimes.

In target-region adaptation for building segmentation, geo-typical synthetic labels yield median IoU improvements “up to $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$1” and improvements “up to $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$2” in high-gap conditions (Song et al., 22 Jul 2025). Concrete examples include ATX$$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$3Cbus improving from $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$4 to $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$5, TIR$$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$6Cbus from $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$7 to $$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$8, and DSTL$$H' = (H + \Delta H) \bmod 360,\ \Delta H \in [-180, 180],$$9Cbus from $f:\mathbb{R}^2\to\mathbb{R}^2$0 to $f:\mathbb{R}^2\to\mathbb{R}^2$1 (Song et al., 22 Jul 2025). The same study also reports that generic synthetic datasets can hurt, with ATX$f:\mathbb{R}^2\to\mathbb{R}^2$2Cbus dropping to $f:\mathbb{R}^2\to\mathbb{R}^2$3 with SynRS3D and $f:\mathbb{R}^2\to\mathbb{R}^2$4 with SyntheWorld (Song et al., 22 Jul 2025). This is one of the clearest empirical demonstrations that target-region layout alignment, not synthetic volume alone, is the relevant variable.

For count-map generation, VAE-Info-cGAN achieves lower APND than cVAE and cGAN baselines in both CRM and HCRM settings (Xiao et al., 2020). The reported APND values are $f:\mathbb{R}^2\to\mathbb{R}^2$5 for CRM and $f:\mathbb{R}^2\to\mathbb{R}^2$6 for HCRM, versus $f:\mathbb{R}^2\to\mathbb{R}^2$7 and $f:\mathbb{R}^2\to\mathbb{R}^2$8 for cGAN with PLC+FLC, and $f:\mathbb{R}^2\to\mathbb{R}^2$9 and $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$0 for cVAE with PLC+FLC (Xiao et al., 2020). Inference is also fast: one forward pass for batch size 32 takes “$p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$1” for CRM and “$p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$2” for HCRM on an NVIDIA Tesla V100 GPU (Xiao et al., 2020).

For geolocated tabular synthesis, NF is reported as essential for realistic spatial densities and VAE as essential for recovering spatial autocorrelation (Lenti et al., 8 Oct 2025). Across 121 geographies, median spatial-density fidelity $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$3 is $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$4 for NF+VAE, compared with $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$5 for VAE without NF; spatial autocorrelation fidelity $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$6 is $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$7 for NF+VAE versus $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$8 for NF+copula; and local-feature fidelity $p_X(x) = p_Z(f(x)) \lvert \det J_f(x) \rvert,$9 is $D1$0 for NF+VAE versus $D1$1 for NF+copula and $D1$2 for Local shuffle (Lenti et al., 8 Oct 2025). GeoPointGAN reports that it “significantly outperforms recent solutions, improving by up to 10 times compared to the most competitive baseline,” and that privacy budgets around $D1$3 can yield a “sweet spot” where regularization from label flipping improves utility (Cunningham et al., 2022).

For vector-native satellite-image synthesis, TerraDiT-$D1$4 reports zero-shot realism and fidelity gains across conditioning formats, with $D1$5, $D1$6, and $D1$7 on Git-Rand-15k for $D1$8, and downstream gains such as OpenEarthMap mIoU improving from $D1$9 to $D2$0 at $D2$1 synthetic data, DIOR mAP@50-95 improving from $D2$2 to $D2$3 at $D2$4, City-Scale TOPO improving from $D2$5 to $D2$6, and AID Top-1 accuracy improving from $D2$7 to $D2$8 at $D2$9 synthetic data (Wei et al., 30 Jun 2026). Geodiffussr reports $D3$0, $D3$1, $D3$2, and $D3$3 for the full MCA model, alongside improvements of “FID $D3$4 $D3$5” and “LPIPS $D3$6 $D3$7” relative to the non-MCA baseline (Inui et al., 28 Nov 2025).

In EO SAR detection, SyntEO shows that ontology-driven geo-typical synthesis can generalize to real scenes without any real training images. Model-3 reaches $D3$8, $D3$9, $0.30\ \text{m/px}$00, and $0.30\ \text{m/px}$01 on the combined test set, while Model-3+ reaches $0.30\ \text{m/px}$02, $0.30\ \text{m/px}$03, $0.30\ \text{m/px}$04, and $0.30\ \text{m/px}$05 (Hoeser et al., 2021). In CrossLoc, multimodal synthetic labels likewise improve real-world localization. On Urbanscape In-place, CrossLoc achieves $0.30\ \text{m/px}$06, $0.30\ \text{m/px}$07, and $0.30\ \text{m/px}$08 under the $0.30\ \text{m/px}$09 criterion, compared with $0.30\ \text{m/px}$10, $0.30\ \text{m/px}$11, and $0.30\ \text{m/px}$12 for DSAC* (Yan et al., 2021).

6. Limitations, misconceptions, and open questions

A recurrent misconception is that geo-typical synthetic data are equivalent to generic synthetic data with more photorealism. The literature consistently rejects that equivalence. In building segmentation, generic synthetic datasets can degrade transfer, whereas target-layout-faithful synthetic labels improve it (Song et al., 22 Jul 2025). In overhead object detection, 3D renders with poor blending can make detectors key off artifacts, and zero-shot end-to-end training “can fail catastrophically for certain sub-classes (tanker)” (Clement et al., 2021). In UAV geo-localization, pretraining on perfect-matching benchmarks transfers poorly to partial-matching localization compared with GTA-UAV, which is built around contiguous map coverage and overlap-based pairing (Ji et al., 2024). A plausible implication is that geo-typicality is less about visual polish than about matching the correct spatial support, topology, and deployment distribution.

Another misconception is that exact constraint satisfaction is always preferable. GenSyn shows the opposite trade-off: SynC achieves $0.30\ \text{m/px}$13 through exact marginal matching, but this comes “at the cost of dependence fidelity,” whereas GenSyn keeps TAE fairly low while achieving the lowest KL divergence and Frobenius norm across counties (Acharya et al., 2022). Synthetic geocode release makes a similar point from a disclosure perspective: when risk is too high for high-utility categorical CART geocodes, synthesizing additional variables is preferred to geographic aggregation or tree-pruning because it reduces risk more effectively with smaller utility loss (Drechsler et al., 2018).

A third misconception is that coordinates can be handled as ordinary real-valued columns. The NF+VAE work explicitly argues that this leads to unrealistic mass in void areas and poor boundary adherence because latitude and longitude have empty spaces, sharp boundaries, and highly uneven densities (Lenti et al., 8 Oct 2025). GeoPointGAN reaches the same issue from the opposite direction: the real dataset should sufficiently cover the domain so that fake points are not trivially distinguishable, because otherwise correlations between features and labels can harm privacy and utility (Cunningham et al., 2022).

The literature also exposes unresolved modeling limits. Edge-only conditioning in the GAN-Reskinner can struggle with color reproduction, suggesting a need to pass limited color information while suppressing synthetic artifacts (Clement et al., 2021). The target-region building pipeline does not simulate motion blur, explicit noise, or atmospheric scattering and is tuned to $0.30\ \text{m/px}$14 imagery, so cross-sensor realism remains limited (Song et al., 22 Jul 2025). GeoVolDiff still relies on a 1D convolutional forward model to pair impedance with seismic, leaving a wave-propagation gap to field data (Pang et al., 2 Jun 2026). Geodiffussr operates at $0.30\ \text{m/px}$15 native synthesis resolution and therefore targets coarse-scale ideation rather than fine material delineation (Inui et al., 28 Nov 2025). GeoPointGAN provides label local differential privacy, but the study explicitly notes that this is not a full LDP guarantee over coordinates (Cunningham et al., 2022).

A final tension concerns whether geo-typicality should be encoded procedurally, statistically, or with end-to-end generative models. Current work supports all three regimes. Procedural and ontology-driven systems such as SyntEO and target-region building synthesis offer precise control and interpretable experiment design (Hoeser et al., 2021, Song et al., 22 Jul 2025). Statistical reconciliation systems such as GenSyn and synthetic geocodes offer explicit control over marginals, dependencies, and disclosure risk (Acharya et al., 2022, Drechsler et al., 2018). Flow and diffusion systems such as TerraDiT-$0.30\ \text{m/px}$16, Geodiffussr, and GeoVolDiff offer flexible, high-capacity generation with increasingly native geospatial conditioning (Wei et al., 30 Jun 2026, Inui et al., 28 Nov 2025, Pang et al., 2 Jun 2026). This suggests that “geo-typical synthetic data” is best understood not as a single method class, but as a design principle: preserve the geography-specific support, spatial relations, and task-relevant distributions that make synthetic samples typical of the world they are intended to model.