Using remote sensing tools to assess land use transitions in unsustainable arid agro-ecosystems

Abstract

This research investigates the human impact on land-cover dynamics in arid agro-ecosystems. Our study area was La Costa de Hermosillo (northwestern Mexico), where the unregulated use of water resources has resulted in the abandonment of irrigated agricultural fields and a shift to new economic activities. Using remote sensing and ancillary datasets combined with classification and regression tree (CART) models, we mapped land-cover class distributions over 22 years (1988–2009) to characterize agricultural changes following management decisions. Our land-cover classification maps had an overall accuracy of over 80%. Using these maps, we were able to show the decrease in agriculture from approximately 115,066 to 66,044 ha between 1988 and 2009 and the conversion to alternative economic activities, with aquaculture increasing from 0 to 10,083 ha during the same period. Our analyses also show the temporal–spatial dynamics of land-use management practices, which suggest that implementation of the remote sensing methods developed in this manuscript may contribute to bridging the gap of knowledge between ecological effects and unsustainable management practices and decrease the time required to inform and make policy decisions in arid agro-ecosystems.