Research output per year
Research output per year
Johannes Balling*, Martin Herold, Johannes Reiche
Research output: Contribution to journal › Comment/Letter to the editor › Academic
The authors regret to report an error in the publication, which we identified while working on subsequent studies. We would like to emphasize that this error does not affect the study's setup, results, or the conclusions drawn from our research. The findings of the paper remain valid and unchanged. The error relates to the standard errors reported for the accuracies. Although the accuracies themselves were correctly calculated and remain unaffected, the standard errors—which reflect the confidence interval of the reported accuracies—were erroneously inflated. The inflation of the standard errors resulted from mistakenly hardcoding the variable for the number of samples per stratum in the calculation script, causing a constant value to be applied across all strata instead of adjusting for the varying sample sizes. In reality, the corrected standard errors are smaller, indicating greater statistical confidence in the reported accuracies. We have recalculated the standard errors using the correct sample sizes for each stratum. Incorrect standard errors are primarily shown in the Appendix (Tables Appendix 1 and Appendix 2). There is a single instance in the main text where incorrect standard errors are reported (bold): Original text: The UA and corresponding standard errors for mapped disturbance using speckle filtered backscatter (Site 1: 95.1 ± 5.3 % and Site 2: 98.2 ± 3.4 %) were comparable to results achieved for non– backscatter without speckle filtering (Site 1: 98.6 ± 4.6 % and Site 2: 100 ± 0 %). The PA was higher for speckle filtered backscatter (Site 1: 64.0 ± 12.7 % and Site 2: 81.9 ± 11.5 %) compared to backscatter without speckle filtering (Site 1: 41.3 ± 12.1 % and Site 2: 62.1 ± 14.1 %). Revised text: The UA and corresponding standard errors for mapped disturbance using speckle filtered backscatter (Site 1: 95.1 ± 1.7 % and Site 2: 98.2 ± 1.1 %) were comparable to results achieved for non- backscatter without speckle filtering (Site 1: 98.6 ± 1.4 % and Site 2: 100 ± 0 %). The PA was higher for speckle filtered backscatter (Site 1: 64.0 ± 3.4 % and Site 2: 81.9 ± 3.1 %) compared to backscatter without speckle filtering (Site 1: 41.3 ± 3.6 % and Site 2: 62.1 ± 4.2 %). Corrections for Appendix 1 and Appendix 2 are presented below (corrected standard error values are in bold). [Table presented] Once again, we would like to underscore that these errors do not affect any of the results or conclusions of the study, which remain valid. Nevertheless, we believe it is important to report this to maintain transparency and uphold the high standards of scientific integrity. The authors would like to apologise for any inconvenience caused.
Original language | English |
---|---|
Article number | 104165 |
Pages (from-to) | 104165 |
Number of pages | 3 |
Journal | International Journal of applied Earth Observation and Geoinformation |
Volume | 134 |
Early online date | 19 Sept 2024 |
DOIs |
|
Publication status | Published - 2024 |
Research output: Contribution to journal › Article › Academic › peer-review