Nowadays massive computational capabilities of online cloud-based platforms offer dramatic possibilities in the field of geosciences, including the analysis of remote sensing imageries for earth surface processes observation and modelling. The work presents an application of Google’s novel Earth Engine (GEE) platform for the analysis of land cover evolution in the last 30 years for the area of Valles Crucenos, Santa Cruz Department, Bolivia. The region is characterized by the expansion of the neighbouring Santa Cruz de la Sierra, the 2nd fastest growing city in Latin America, that is causing both a migration of rural population to the city and an increase in the fluxes of resources from the peri-urban area, inducing heavy changes in the land cover of the whole department. Land cover maps for the years 1986-87, 1996-97, 2006-07 and 2016-17 were analysed, each one realised with a supervised classification of a composite of two year of Landsat Tier 1 (T1) satellite images. Tier 1 was used to guarantee formal geometric and radiometric quality criteria. All the Landsat T1 scenes used to produce the composite twoyears stacks were selected through a cloud score ranking under a threshold of 10% of image cover. The whole process required no data download, with a computational time of 100 seconds on GEE servers for each supervised classification. CART supervised classification process was applied to Landsat T1 spectral bands (from ~0.45 to ~2.3 μm) on over 3600 training points for 7 land use classes: forest, dry forest, agricultural, shrub-bare land, grasslandpasture, grass-shrubland and urban. CART algorithm showed good performance accuracy levels, in a range between 0.86 and 0.90 in term of K. While the prediction made with land cover change models for the area showed an expected deforestation trend, land cover analysis revealed a decrease of agricultural areas and an increase of forest cover in the last 10 years. Results can be explained as an effect of the progressive abandon of agricultural areas, caused by the urbanisation of rural population in Santa Cruz de la Sierra and in minor urban areas of Valles Crucenos. Highlighting these dynamics can provide support to land use planning and future environmental and socio-economic policies for the area. Moreover, the presented methodology can be simply and quickly extended to similar areas of the whole Latin America, especially where land cover change and deforestation data are not present or are available with a scattered coverage.

Back to the future of land cover and sustainability. Historical land cover analysis in remote Bolivian areas with Google Earth Engine / Cristiano Foderi, Giulio Castelli, Elena Bresci, Fabio Salbitano. - ELETTRONICO. - (2018), pp. 36-36. ((Intervento presentato al convegno AIT2018 ITALIAN SOCIETY OF REMOTE SENSING IX CONFERENCE.

Back to the future of land cover and sustainability. Historical land cover analysis in remote Bolivian areas with Google Earth Engine

Cristiano Foderi;Giulio Castelli;Elena Bresci;Fabio Salbitano
2018

Abstract

Nowadays massive computational capabilities of online cloud-based platforms offer dramatic possibilities in the field of geosciences, including the analysis of remote sensing imageries for earth surface processes observation and modelling. The work presents an application of Google’s novel Earth Engine (GEE) platform for the analysis of land cover evolution in the last 30 years for the area of Valles Crucenos, Santa Cruz Department, Bolivia. The region is characterized by the expansion of the neighbouring Santa Cruz de la Sierra, the 2nd fastest growing city in Latin America, that is causing both a migration of rural population to the city and an increase in the fluxes of resources from the peri-urban area, inducing heavy changes in the land cover of the whole department. Land cover maps for the years 1986-87, 1996-97, 2006-07 and 2016-17 were analysed, each one realised with a supervised classification of a composite of two year of Landsat Tier 1 (T1) satellite images. Tier 1 was used to guarantee formal geometric and radiometric quality criteria. All the Landsat T1 scenes used to produce the composite twoyears stacks were selected through a cloud score ranking under a threshold of 10% of image cover. The whole process required no data download, with a computational time of 100 seconds on GEE servers for each supervised classification. CART supervised classification process was applied to Landsat T1 spectral bands (from ~0.45 to ~2.3 μm) on over 3600 training points for 7 land use classes: forest, dry forest, agricultural, shrub-bare land, grasslandpasture, grass-shrubland and urban. CART algorithm showed good performance accuracy levels, in a range between 0.86 and 0.90 in term of K. While the prediction made with land cover change models for the area showed an expected deforestation trend, land cover analysis revealed a decrease of agricultural areas and an increase of forest cover in the last 10 years. Results can be explained as an effect of the progressive abandon of agricultural areas, caused by the urbanisation of rural population in Santa Cruz de la Sierra and in minor urban areas of Valles Crucenos. Highlighting these dynamics can provide support to land use planning and future environmental and socio-economic policies for the area. Moreover, the presented methodology can be simply and quickly extended to similar areas of the whole Latin America, especially where land cover change and deforestation data are not present or are available with a scattered coverage.
ABSTRACT BOOK AIT2018 ITALIAN SOCIETY OF REMOTE SENSING IX CONFERENCE
AIT2018 ITALIAN SOCIETY OF REMOTE SENSING IX CONFERENCE
Cristiano Foderi, Giulio Castelli, Elena Bresci, Fabio Salbitano
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2158/1142370
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact