The maps were prepared using soil information from 23,273 soils samples, collected from 56 districts covering seven provinces. These soil properties are combined with a stack of 168 remote sensing-based soil covariates (SRTM DEM derivatives, climatic images, vegetation index etc.). Later the spatial predictions were generated using a machine learning method and the random forest.

Please refer to the following journal article for the detailed procedure: Hengl T, Mendes de Jesus J, Heuvelink GBM, Ruiperez Gonzalez M, Kilibarda M, Blagotić A, et al. (2017) SoilGrids250m: Global gridded soil information based on machine learning. PLoS ONE 12(2): e0169748. doi:10.1371/journal. pone.0169748

The current version map predicts the following soil properties:

• Soil texture (sand, silt and clay)
• Soil pH
• Organic matter
• Total nitrogen
• Available phosphorus
• Available potassium
• Zinc
• Boron

These are different environmental factors that affect soil properties. Covariate layers are generated using satellite images (raster data). Some of the important factors are topographic data, vegetation, precipitation, temperature, soil parent material, land cover type, landform classes.

Spatial resolution of the maps created is 250 m.

The soil profile data were collected from various government projects including National Land Use Project, Irrigation and Water Resources Management Project, Central Agriculture Laboratory (previously Soil Management Directorate), and Nepal Agricultural Research Council (NARC).

DSM uses advanced computational algorithms that use both soil sample data and environmental variables to generate maps. It does not only use spatial autocorrelation as the means to interpolate the data, but also considers soil forming factors. Also, the process can be automated so that newer versions of the map can be developed faster once new soil samples are collected. As soil properties are combined with environmental variables, a smaller number of soil samples would be enough to generate DSM compared to conventional soil map.

Researchers have done different studies and compared results of DSM with conventional methods. One of the studies showed that the random forests algorithm consistently performed better than the linear regression algorithm. The researchers found Root Mean Squared Error (RMSE) across soil properties and depths decreased on an average of 15–75% . Similarly another study also showed that the DSM method was more accurate than the deterministic pedological approach.

You may use a point or polygon tool on the soil map. You can either click on the farm land or draw a polygon over your farm. The output (soil properties) will be generated in a tabular form that can be downloaded. Caution: The values generated are based on the prediction results, so please consult experts for their interpretation and taking decisions about future farm operation or perform detailed soil analysis for taking further decisions.

Yes, you can select the area of your interest using a polygon function provided in the left top corner of the website. Here are some basic instructions:

• Go to https://soil.narc.gov.np. Go to the Map section and open the soil map.
• The map will be updated. Select the zoom button under tool and drag the map to zoom an area.
• On the left side of the screen on the navigation panel, select the polygon tool to select the area. Draw a rectangle or triangle shape around your interested area on the map and start getting the information.

· APIs

The data and information are made available through APIs, so that people can access them easily on web, phone apps or using programming languages.

· Direct download

The users can directly download data in raster format after registration.

· Interactive tools

The users can use interactive tools for data query. Users can draw a point or polygon over a web map and get different information on soil properties. Later those data can be exported as CSV file format.

• It can identify a domain with similar soil properties, identify soil fertility status (sufficiency and deficiency of plant nutrients).
• Farmers, policy makers, extension workers, agri-input retailers and natural resource managers can use it as a decision support tool as it provides easy access to location-specific information on soil properties (spatial distribution) and plant nutrients, and helps to estimate fertilizer requirements.
• It can be used as input data in models to extrapolate domain specific fertilizer recommendations and to estimate the total amount of fertilizers required for a particular crop and season.
• It helps to identify areas with deficient plant nutrients and provide site specific fertilizer formulations (i.e., determining nutrient grades of blended fertilizers)
• It helps to design the program on soil fertility and fertilizer management including correcting soil acidity, management of organic matter, addressing micronutrient deficiency etc.

To acknowledge the organizations that have provided data or products, Krishiprabidhi requests that data users include a citation to all data supplied through Krishiprabidhi in output products, websites, and publications. Overall, data are provided to the users under the general terms and condition:

• Share: Copy and redistribute the material in any medium or format.
• Attribution: You must give appropriate credit, provide a link to the license, and indicate if changes were made.
• Non-commercial: You cannot use the material for commercial purposes.
• Non-derivative: If you remix, transform or build upon the material, you cannot distribute the modified material.

Yes, soil properties change over the season and it is recommended to update the map every five years.

Please write to us using the Contact us page of our web portal.

To improve the soil map of Nepal, please consider contributing soil data or you have been involved in data collection, please let us know. Organizations/companies that contributed data are acknowledged as contributing organizations/companies on our website and we will be happy to integrate it. Please Contact us to share information about your dataset, and we’ll be in touch.