Recent workshops brought together several developers, educators and users of software packages extending popular languages for spatial data handling, with a primary focus on R, Python and Julia. Common challenges discussed included handling of spatial or spatio-temporal support, geodetic coordinates, in-memory vector data formats, data cubes, inter-package dependencies, packaging upstream libraries, differences in habits or conventions between the GIS and physical modeling communities, and statistical models. The following set of recommendations have been formulated:

(i) considering software problems across data science language silos helps to understand and standardise analysis approaches, also outside the domain of formal standardisation bodies;
(ii) whether attribute variables have block or point support, and whether they are spatially intensive or extensive has consequences for permitted operations, and hence for software implementing those;
(iii) handling geometries on the sphere rather than on the flat plane requires modifications to the logic of simple features,
(iv) managing communities and fostering diversity is a necessary, on-going effort, and
(v) tools for cross-language development need more attention and support.

This study investigates the added value of operational radar with respect to rain gauges in obtaining high-resolution daily rainfall fields as required in distributed hydrological modelling. To this end we combine data from the Netherlands operational national rain gauge network (330 gauges nation wide) with our own experimental network (30 gauges within 225 km2). Based on 74 selected rainfall events (March-October 2004) the spatial variability of daily rainfall is investigated at three spatial extents: small (225 km 2), medium (10,000 km2) and large (82,875 km2). From this analysis it is shown that semivariograms show no clear dependence on season. Predictions of point rainfall are performed for all three extents using three different geostatistical methods: (i) ordinary kriging (OK; rain gauge data only), (ii) kriging with external drift (KED) and (iii) ordinary colocated cokriging (OCCK), the latter two using both rain gauge data and range-corrected daily radar composites - a standard operational radar product from the Royal Netherlands Meteorological Institute (KNMI). Focus here is on automatic prediction. For small extent rain gauge data alone perform better than radar while for larger extents with lower gauge densities radar performs overall better than rain gauge data alone (OK). Methods using both radar and rain gauge data (KED and OCCK) prove to be more accurate than using either rain gauge data alone (OK) or radar, in particular for larger extents. The added value of radar is positively related to the correlation between radar and rain gauge data. Using a pooled semivariogram is almost as good as using event-based semivariograms, which is convenient if the prediction is to be automated. An interesting result is that the pooled semivariograms perform better in terms of estimating the prediction error (kriging variance) especially for the small- and medium extent, where the number of data points to estimate semivariograms is small and event-based semivariograms are rather unstable.

This study investigates the added value of operational radar with respect to rain gauges in obtaining high-resolution daily rainfall fields as required in distributed hydrological modeling. To this end data from the Netherlands operational national rain gauge network (330 gauges nationwide) is combined with an experimental network (30 gauges within 225 km²). Based on 74 selected rainfall events (March-October 2004) the spatial variability of daily rainfall is investigated at three spatial extents: small (225 km²), medium (10 000 km²), and large (82 875 km²). From this analysis it is shown that semivariograms show no clear dependence on season. Predictions of point rainfall are performed for all three extents using three different geostatistical methods: (i) ordinary kriging (OK; rain gauge data only), (ii) kriging with external drift (KED), and (iii) ordinary collocated cokriging (OCCK), with the latter two using both rain gauge data and range-corrected daily radar composites - a standard operational radar product from the Royal Netherlands Meteorological Institute (KNMI). The focus here is on automatic prediction. For the small extent, rain gauge data alone perform better than radar, while for larger extents with lower gauge densities, radar performs overall better than rain gauge data alone (OK). Methods using both radar and rain gauge data (KED and OCCK) prove to be more accurate than using either rain gauge data alone (OK) or radar, in particular, for larger extents. The added value of radar is positively related to the correlation between radar and rain gauge data. Using a pooled semivariogram is almost as good as using event-based semivariograms, which is convenient if the prediction is to be automated. An interesting result is that the pooled semivariograms perform better in terms of estimating the prediction error (kriging variance) especially for the small and medium extent, where the number of data points to estimate semivariograms is small and event-based semivariograms are rather unstable.