Sensitivity of Landsat/IKONOS accuracy comparison to errors in photointerpreted reference data and variations in test point sets

We evaluated the sensitivity of a comparison between classified Landsat and IKONOS images to the reference data used to make the comparison. Reference data were produced using two different procedures and by multiple individuals performing identical procedures. We also screened potential test points according to the homogeneity of their surrounding neighbourhoods. Landsat performed relatively better than IKONOS regardless of the reference data used, but the individual performance of each classified image varied significantly according to the reference dataset against which it was scored. The assessed ability of Landsat and IKONOS to recognize certain land‐cover classes was extremely sensitive to the reference data. Differences in individuals' interpretation habits outweighed the effects of the reference data production method, and the test points' neighbourhood homogeneity did not consistently bias overall accuracy. These results support careful documentation of uncertainty associated with reference data, including overall reliability and issues regarding particular land‐cover/use classes.

[1]  J. Scepan,et al.  Thematic validation of high-resolution Global Land-Cover Data sets , 1999 .

[2]  R. Lunetta,et al.  Remote sensing and Geographic Information System data integration: error sources and research issues , 1991 .

[3]  David L. Verbyla,et al.  Optimistic bias in classification accuracy assessment , 1996 .

[4]  Huang Youcai,et al.  Building the Estimation Model of Digitizing Error , 1997 .

[5]  Simon Jones,et al.  Where's the fire? Quantifying uncertainty in a wildfire threat model , 2004 .

[6]  K. Thapa,et al.  Accuracy of spatial data used in geographic information systems , 1992 .

[7]  Q. Qin,et al.  Satellite remote sensing of groundwater: Quantitative modelling and uncertainty reduction using 6S atmospheric simulations , 2004 .

[8]  J. Wickham,et al.  Thematic accuracy of the 1992 National Land-Cover Data for the eastern United States: Statistical methodology and regional results , 2003 .

[9]  D. Roberts,et al.  Sources of error in accuracy assessment of thematic land-cover maps in the Brazilian Amazon , 2004 .

[10]  Alfred Stein,et al.  Spatial variability in classification accuracy of agricultural crops in the Dutch national land-cover database , 2004, Int. J. Geogr. Inf. Sci..

[11]  J. Wickham,et al.  Effects of landscape characteristics on land-cover class accuracy , 2003 .

[12]  C. Ricotta,et al.  Evaluating the degree of fuzziness of thematic maps with a generalized entropy function: A methodological outlook , 2002 .

[13]  Russell G. Congalton,et al.  Assessing the accuracy of remotely sensed data : principles and practices , 1998 .