Quantitative and qualitative assessment of the accuracy of neurointerpolated annual mean precipitation and temperature surfaces for China

This study quantitatively and qualitatively compares new annual mean precipitation and temperature surfaces for China produced using neurointerpolation (Bryan and Adams, in press) with Gestalt climate maps (Editorial Committee of the Physical Geography of China, 1995) at a variety of scales. The new mean precipitation and temperature maps display good broad scale correspondence with the Gestalt maps. However, correspondence decreases at increasingly finer scales. The new climate maps capture substantial fine-scale spatial variation seemingly driven by topographic influences and continentality that is not present on the Gestalt maps. The precipitation map however, lacks important subregional scale variation in several areas. It is concluded that whilst climate maps may quite readily be produced by neurointerpolation, it is still imperative to validate the new maps with Gestalt maps which consider a more complex set of indicators than climate station and topographic data alone.