Extending a prototype knowledge and object based image analysis model to coarser spatial resolution imagery: An example from the Missouri River

By:

Links

Abstract

A prototype knowledge- and object-based image analysis model was developed to inventory and map least tern and piping plover habitat on the Missouri River, USA. The model has been used to inventory the state of sandbars annually for 4 segments of the Missouri River since 2006 using QuickBird imagery. Interpretation of the state of sandbars is difficult when images for the segment are acquired at different river stages and different states of vegetation phenology and canopy cover. Concurrent QuickBird and RapidEye images were classified using the model and the spatial correspondence of classes in the land cover and sandbar maps were analysed for the spatial extent of the images and at nest locations for both bird species. Omission and commission errors were low for unvegetated land cover classes used for nesting by both bird species and for land cover types with continuous vegetation cover and water. Errors were larger for land cover classes characterized by a mixture of sand and vegetation. Sandbar classification decisions are made using information on land cover class proportions and disagreement between sandbar classes was resolved using fuzzy membership possibilities. Regression analysis of area for a paired sample of 47 sandbars indicated an average positive bias, 1.15 ha, for RapidEye that did not vary with sandbar size. RapidEye has potential to reduce temporal uncertainty about least tern and piping plover habitat but would not be suitable for mapping sandbar erosion, and characterization of sandbar shapes or vegetation patches at fine spatial resolution.

Study Area

Publication type Conference Paper
Publication Subtype Conference Paper
Title Extending a prototype knowledge and object based image analysis model to coarser spatial resolution imagery: An example from the Missouri River
Year Published 2012
Language English
Publisher Instituto Nacional de Pesquisas Espaciais - INPE
Contributing office(s) Northern Prairie Wildlife Research Center
Description 6 p.
Larger Work Type Book
Larger Work Subtype Conference publication
Larger Work Title Proceedings of the 4th Conference on GEographic Object-Based Image Analysis - GEOBIA 2012
First page 530
Last page 535
Conference Title Conference on Geographic Object-Bsed Image Analysis (GEOBIA) 2012
Conference Location Rio de Janerio, Brazil
Conference Date May 7-9, 2012
Country United States
State Montana, Nebraska, North Dakota, South Dakota
Other Geospatial Missouri River
Online Only (Y/N) N
Additional Online Files (Y/N) N
Google Analytic Metrics Metrics page
Additional publication details