Methods are given and explored for improving verification of cloud ceiling and visibility forecasts. Verification of these forecasts is performed based on data from surface METAR stations, which are sometimes densely located in a particular area and sometimes only sparsely located, whereas the forecasts are gridded over the entire region. If a forecast misses in an area where the stations are densely located then that forecast may be penalized more times than are actually appropriate. Conversely, if a forecast does well in a region with sparsely located stations then it may not be given full credit. Emphasis will be given to the first case because often regions with sparsely located stations are not of interest to those using the forecasts. Spatial statistical techniques will be used to find an "optimal" network design to be used to better score forecasts over densely located regions. Results for a network of 48 monitors in the San Francisco Bay Area suggest that the removal of some stations would provide a more accurate verification of forecasts of visibility. Results for cloud ceiling height data suggested that thinning the network may not be appropriate.