1,516
Views
33
CrossRef citations to date
0
Altmetric
Articles

400 predictions: the SEARCH Sea Ice Outlook 2008–2015

&
Pages 274-287 | Received 11 Mar 2016, Accepted 22 Aug 2016, Published online: 26 Sep 2016

Figures & data

Figure 1. Number of SIO contributions by year and method, all months combined, 2008–2015.

Figure 1. Number of SIO contributions by year and method, all months combined, 2008–2015.

Figure 2. SIO June, July and August 2015 predictions for September mean sea ice extent. Box plots indicate the median, IQR and outliers for each distribution.

Figure 2. SIO June, July and August 2015 predictions for September mean sea ice extent. Box plots indicate the median, IQR and outliers for each distribution.

Figure 3. Observed September extent shown with quadratic trend and median of July SIO predictions, 1979–2015.

Figure 3. Observed September extent shown with quadratic trend and median of July SIO predictions, 1979–2015.

Figure 4. Observed September extent compared with median and IQR of July SIO predictions, 2008–2015.

Figure 4. Observed September extent compared with median and IQR of July SIO predictions, 2008–2015.

Figure 5. SIO prediction error (median July SIO minus observed September extent) versus observed change from September the previous year, 2008–2015.

Figure 5. SIO prediction error (median July SIO minus observed September extent) versus observed change from September the previous year, 2008–2015.

Table 1. Comparison of median July SIO predictions with predictions based on linear or quadratic extrapolation (using data from 1979 to the previous year) or persistence (extent same as previous year), over 2008–2015, in millions of km2.

Table 2. Quantile regression coefficients (with robust standard errors), modeling the median absolute prediction error as a function of (0,1) indicators for year, month and method. Base groups are the comparison categories. Regression 2 and 3 apply the new distinction between ice-ocean and ice-ocean-atmosphere modeling; regression 3 includes contributions only from researchers, setting aside those identified as general public.