Improving the accuracy of livestock distribution estimates through spatial interpolation

摘要: Animal distribution maps serve many purposes such as estimating transmission risk of zoonotic pathogens to both animals and humans. The reliability usability is highly dependent on the quality input data. However, decisions how perform livestock surveys are often based previous work without considering possible consequences. A better understanding impact using different sample designs processing steps accuracy estimates was acquired through iterative experiments detailed survey. importance size, design aggregation demonstrated spatial interpolation presented a potential way improve cattle number estimates. As expected, results show that an increasing size increased precision but these improvements were mainly seen when initial relatively low (e.g. median relative error decrease 0.04% per sampled parish for sizes below 500 parishes). For higher sizes, added value further samples declined rapidly 0.01% above parishes. When two-stage stratified applied yield more evenly distributed samples, levels densities stabilised at lower compared one-stage sampling. Aggregating resulting yielded significantly accurate because averaging under- over-estimates aggregating from subcounty district level, P <0.009 2,077 parishes samples). During aggregation, area-weighted mean values assigned administrative unit levels. this step preceded by fill in missing non-sampled areas, improved remarkably. This counts especially spatially even <0.001 170 sampling level). Whether same observations apply scale should be investigated.