AND
Abstract. Using airborne strip-transect monitoring data, maps of bird densities were estimated for several bird species. Because the densities are transformed count data with many genuine zeros, we combined a generalised linear modelling approach with geostatistics for the spatial interpolation. Wat...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Text |
| Language: | English |
| Subjects: | |
| Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.124.2094 http://www.geog.uu.nl/~pebesma/publ/gs_corr.pdf |
| Summary: | Abstract. Using airborne strip-transect monitoring data, maps of bird densities were estimated for several bird species. Because the densities are transformed count data with many genuine zeros, we combined a generalised linear modelling approach with geostatistics for the spatial interpolation. Water depth and distance to coast were used as covariates to model the trend. A generalised estimating equations-like approach was used to estimate the trend, the spatial correlation function and the over-dispersion parameter. The residual variance was taken proportional to the (varying) mean, and non-stationary residual variances and covariances were obtained from known means, a stationary correlogram and the over-dispersion parameter. The results for one species (Fulmarus glacialis) are shown as approximate 95 % prediction intervals of 5 km × 5 km block mean densities. 1. |
|---|