t The mapping of forest resources in a study region is approached when the region is partitioned into spatial units by means of a completely data driven, designbased sampling strategy. When auxiliary variables are available for all the units, the prediction of the densities of an interest attribute can be performed by using an assisting model. Under these circumstances, the model residuals are interpolated using the inverse distance weighting interpolator with a data-driven smoothing parameter selection, and the density of the attribute for each unit is obtained by summing prediction and interpolated error. Finally, densities are rescaled to match the traditional total estimate with the sum of mapped values. The uncertainty is accounted for by a bootstrap procedure. A simulation study is performed and a case study is presented.
La mappatura di risorse forestali per le unita spaziali che compongono ` una regione di studio viene affrontata con un approccio data-driven basato sul disegno. Se sono disponibili variabili ausiliarie, la previsione del valore dell’attributo di interesse puo essere effettuata con strategie assistite da modello. I residui del ` modello vengono previsti utilizzando l’interpolatore inverse distance weighting dove il parametro di smorzamento viene selezionato con una procedura data-driven e le densita dell’attributo vengono ottenute sommando le previsioni e gli errori ` interpolati. Inoltre le densita vengono riscalate in modo tale che la stima del to- ` tale ottenuta sommando i valori previsti coincida con quella ottenuta con approcci tradizionali. L’incertezza delle stime e valutata attraverso una procedura bootstrap. ` Uno studio di simulazione viene effettuato e viene presentato un caso di studio.
Franceschi, S., DI BIASE, R.M., Fattorini, L., Marcheselli, M., Pisani, C. (2022). Data-driven design-based mapping of forest resources. In Book of the Short Papers SIS 2022 (pp.1245-1251). Milano : Pearson Italia.
Data-driven design-based mapping of forest resources
Sara Franceschi;Rosa Maria Di Biase;Lorenzo Fattorini;Marzia Marcheselli;Caterina Pisani
2022-01-01
Abstract
t The mapping of forest resources in a study region is approached when the region is partitioned into spatial units by means of a completely data driven, designbased sampling strategy. When auxiliary variables are available for all the units, the prediction of the densities of an interest attribute can be performed by using an assisting model. Under these circumstances, the model residuals are interpolated using the inverse distance weighting interpolator with a data-driven smoothing parameter selection, and the density of the attribute for each unit is obtained by summing prediction and interpolated error. Finally, densities are rescaled to match the traditional total estimate with the sum of mapped values. The uncertainty is accounted for by a bootstrap procedure. A simulation study is performed and a case study is presented.File | Dimensione | Formato | |
---|---|---|---|
Franceschi et al. SIS.pdf
accesso aperto
Tipologia:
PDF editoriale
Licenza:
Creative commons
Dimensione
8.17 MB
Formato
Adobe PDF
|
8.17 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11365/1275294