Methodological proposal to assess plastic greenhouses land cover change from the combination of archival aerial orthoimages and Landsat data

This work outlines a methodological proposal to assess plastic covered greenhouses (PCG) land cover change from the combination of archival aerial orthoimages and Landsat data. In this way, landscape spatial metrics were semi-automatically derived to be used in the analysis of the spatial arrangement of PCG areas. The experimental process consisted of two main phases: (i) mapping PCG through a semi-automatic object-based image analysis (OBIA) approach relying on segmentation plus non-parametric supervised classification; (ii) processing the PCG classified objects to yield different landscape spatial metrics. The case study has focused on two high density PCG sites located in south-eastern Spain. To analyse PCG land cover evolution, each study site was composed of three multi-temporal remote sensed datasets formed by the fusion of orthoimages (O) derived from archival aerial photography and temporally corresponding Landsat images (L). In terms of PCG mapping performance, the best results were obtained when using O + L datasets as complementary data to be used in a data fusion process. In addition, a new feature called “Greenhouse Detection Index” has been successfully developed and tested, yielding excellent results at the mapping phase. Finally, the semi-automatically extracted PCG land cover metrics, though depicting some variability, have reproduced the behaviour and temporal trend of the manually obtained ones (manual digitalisation) reasonably well. These results can be translated to an exponential reduction of time and cost for analysing long-term PCG land cover change.