When the Mask Meets the Parcel: Fit for Purpose? Mask Sensitivity and Temporal Uncertainty in Linking Parcel-Level Soil Analyses to Sentinel-2 across the Menemen Plain, Türkiye
Linking soil-test records to satellite spectra requires suitable observations at compatible spatial and temporal support. This poster presents an observation-side audit for the Menemen Plain, lower Gediz alluvium, Türkiye, using Sentinel-2 L2A observations, cadastral parcel geometries and an EnMAP screening comparison. No laboratory results or soil-test archive data were used. The analysis covers 4,766 agricultural cadastral parcels and 758 Sentinel-2 acquisition dates from 2 January 2016 through 25 October 2026. Mask A requires SCL = 5, NDVI < 0.25 and NDMI < 0; Mask B additionally requires NBR2 < 0.075. At 20 m, eligibility requires at least four pixels in a 10 m inward-buffered parcel core. A qualifying date has at least 80% of that fixed core passing the mask, with cloudy pixels retained in the denominator. Only 2,870 parcels are geometrically eligible. Mask B yields at least one qualifying date for 2,210 parcels and recurrent observations in at least nine of eleven years for 946 parcels. Among parcels smaller than 0.5 ha, only 331 of 2,354 (14%) ever qualify. The median distance to the nearest qualifying observation is 90 days for Mask B (n = 2,210), compared with 37 days for Mask A (n = 2,504); these are conditional statistics for different parcel groups. Tightening the NBR2 threshold to 0.05 reduces the respective counts to 1,856 parcels with any qualifying date and 606 recurrent parcels. Across 651 EnMAP candidate pixels, 366 (56%) pass Sentinel-2 Mask B on both nearby observation dates. Median continuum-removed band depth near 2.2 micrometres differs between the Mask-B and A-only groups, illustrating sensitivity to surface screening. The results show how parcel geometry, screening and season affect opportunities for soil–satellite linkage. Agreement between two processing workflows is not field validation. Thresholds were not calibrated locally, and the EnMAP comparison does not establish mineral or nutrient concentrations.
Authors
- Tahsin Ünal
- Burcu Koçoğlu (ORCID: https://orcid.org/0009-0007-9092-7836)
Institutions
- RISE Research Institutes of Sweden (SE)
- Ministry of Agriculture and Forestry (TR)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-30
- DOI
- https://doi.org/10.5281/zenodo.23816144
- Primary Topic
- Remote Sensing in Agriculture
- Type
- article
- Field-Weighted Citation Impact
- 0.00