|Title||Rescue and renewal of legacy soil resource inventories in Iran as an input to digital soil mapping|
|Author(s)||Rasaei, Zahra; Rossiter, David G.; Farshad, Abbas|
|Source||Geoderma Regional 21 (2020). - ISSN 2352-0094|
|Department(s)||ISRIC - World Soil Information|
|Publication type||Refereed Article in a scientific journal|
|Keyword(s)||Alfisols - Aridisols - Conventional soil mapping - Cornell adequacy criteria - Entisols - Inceptisols - Legacy soil inventories - Mollisols - Vertisols - Weighted overall accuracy - Weighted tau index|
Maps of soil properties and classes are key inputs to land management, especially with the increasing awareness of ecosystem services provided by soils. Due to limited resources for new soil surveys, legacy soil inventories are often the major public source of soil data, and this is the case in Iran. One line of previous work has presented methods for data archaeology, rescue and renewal, and another line has shown the value of legacy surveys as covariates for digital soil mapping (DSM). The present study therefore aimed to integrate these two streams, adding another step of adequacy evaluation of the rescued surveys according to the Cornell guidelines. The study area is a 10,480 km2 region located at the border of Isfahan and Chaharmahal-va-Bakhtiari provinces, Iran, covered by three legacy studies at the scale of 1: 50,000. The legacy maps were georeferenced and geocorrected, after which the Cornell adequacy guidelines were used to assess the quality of soil unit separation as evaluated at four levels of Soil Taxonomy. Evaluation was by weighted accuracy and the Tau coefficient, based on forty-one legacy soil profiles from a correlation study. The weighted accuracy of the map and the Tau index at all classification levels were respectively greater than 70% and 50%, which can be considered as satisfactory separation of soil units. Multinomial logistic regression (MLR) predictive models of soil classes with and without the legacy map as covariate were fit. The predictive accuracy of the models was improved at all taxonomic levels when the renewed legacy soil map was included as a covariate. We conclude that the selected legacy soil maps are of reasonable quality and can be used as reliable and useful inputs to DSM, and we propose that these procedures be widely applied.