Hi Iticha,

Hope I understood your question. In my opinion SMU is better grid because one SMU can be smaller or larger than other SMU of that SMU has different soil properties. In other words, both grid and feature data can be used for digital soil database. Since it is easier to do spatial analysis using grid data than feature/polygon data, scientist prefer to have grid data for analysis and evaluation. In terms of accuracy, I am confident that feature data is better than grid data because we set up a boundary for one type of soil in SMU. However, depending on grid size we need to average the soil characteristics in one grid or need to assume soil type based on major soil type exist in particular grid. Hope I am able to clarify a little bit. Thanks

Dear Dr. Ray,

You raised very important points. Still I want you to elaborate more to make it clear the most critical sentences "it is easier to do spatial analysis using grid data than feature/polygon data, scientist prefer to have grid data for analysis and evaluation".

Besides, what is meant by "depending on grid size we need to average the soil characteristics in one grid or need to assume soil type based on major soil type exist in particular grid". Is soil type defined for each SMU or each grid in grid sampling?

In grid data, one soil type is represented by one SMU that exists here and there on the map. Here, soils under one SMU require more or less similar management measure. But in feature data, as far as I understood from you answer "it is possible to set up more than one soil type within a single SMU each having their own boundary". Here, it is difficult to set one management measure for one SMU because it may contains more than one soil type and thus management measure is recommended for each soil type within SMU. Am I got your point Dr. Ray? Please would you like to extend your points?

Dear all, I don't agree with the statement in the question: "grid sampling is suitable to digital soil mapping and provide better soil information". In many landscapes, especially of glacial, but also alluvial origin, it would be truth only for very dense grids, e.g. 5mx5m, what is unacceptable due to costs. The use of regular grid is easy and "comfotable", because grids are implemented in many softwares, but it creates a risk of omitting of some relatively small, but important areas. Such areas e.g. of about 0.2 ha of sandy soil within large area of loamy soil (or vice versa) may be easily omitted, but they are important for specific site management.

For the purposes of precision agriculture, I would rather combine directed soil sampling, including all areas differing in position in relief, soil colour, vegetation status (feature data, as mentioned by Ram L. Ray). If such areas are large, the grid sampling might be used within them, to confirm their internal homogeneity.

I think it is more reasonable to map soil management units (SMU) rather on the base of crop/management limiting factors (susceptibility to drought, excess of water, erosion etc), than on soil type. For example, within one area we may find some Regosols and Cambisols with similar limiting factors.

Dear Dr. Stepien, you raised interesting points. What do you mean by the sentence "it is more reasonable to map soil management units (SMU) rather on the base of crop/management limiting factors (susceptibility to drought, excess of water, erosion etc), than on soil type?". The soil types that could be mapped as soil mapping units are developed mainly based on soil forming factors such as landform, slope, soil texture, color, soil depth, etc. Even these factors can be used to develop SMU in digital soil mapping following grid sampling and analysis. Even following classification of SMU using those factors such as susceptibility to drought, excess of water, erosion etc, we may find different soil types having relatively similar limitations in such factors.

Dear Birhanu,

it depends on definition of soil type.

I agree, the criteria you mentioned may clearly inform on limiting factors,. For example the areas of higher susceptibility to drought would be indicated by higher elevations and south aspect (in northern hemisphere), sandy or gravelly texture (or other, with underlaid by these material at small depths, e. g. 40 cm, bright colors and small depth. The susceptibility to waterlogging will be indicated by lower positions in releif (depressions), dark colors (organic matter accumulates more in areas of higher humidity) and clayey texture.

However, more taxonomic idea of soil type, understood as a sequence and properties of soil horizons, as in Classification of Polish soils, what more or less corresponds to WRG Reference Groups, not always inform on limiting factors. For example, soils high in organic matter, with mollic horizon - Phaeozems, may be both susceptible to excess of water (Gleyic and Stagnic Phaeozems) or to drought (some others, in higher positions in relief, especially if underlaid by sands. Both groups of Phaeozems may occur on one, the same field.

Regards, Michał

Hi Iticha,

"it is easier to do spatial analysis using grid data than feature/polygon data, scientist prefer to have grid data for analysis and evaluation". I was supporting your statement " Why grid sampling is suitable for digital soil mapping". I also agree with Dr. Stepien's statement. Let me elaborate little more on what I was trying to say by grid versus feature data. I was referring feature/polygon data for SMUs. For spatial analysis (using GIS tool), it is easier to do analysis because most of the other parameters are available in grid/raster format. For examples, digital elevation model (DEM), soil slope, soil moisture, cover management factor, soil erodibility factor etc. Being said that it does not mean we cannot do analysis using SMUs because mostly, soil grid data is developed from SMUs or feature data.

Besides, what is meant by "depending on grid size we need to average the soil characteristics in one grid or need to assume soil type based on major soil type exist in particular grid". Is soil type defined for each SMU or each grid in grid sampling?

I was trying to compare SMU and Grid data. SMUs can be irregular in size/area. For example, soil type 1 can have 1 m^2 of area and soil type 2 can have 1 km^2 of area. However, grid data is either in regular size 5mx5m or 1km x 1km or 5k X 5km. If 1 grid has multiple SMUs, we need to average the soil characteristics or just assume the soil type which has more coverage area.

it is possible to set up more than one soil type within a single SMU each having their own boundary". I was referring SMUs as a feature data. As the attached document said, SMUc can be "Mono-taxa" for a single soil type and "Multi-txa" for two or more main soil types.

https://www.nrcs.usda.gov/Internet/FSE_DOCUMENTS/nrcs144p2_025378.pdf

I hope it helps. Please feel free to bring more discussion points.

Thanks

Regards

Ram Ray

Dear Dr Ray and Stepien,

Your answers are very clear and valuable. From a grid data, we can develop probability map for individual soil properties. Besides, we can develop SMUs from grid samples. In this case, what is the relationship between grid data and feature data?

Dear Iticha,

If you have time to go through these links, that is what I was trying to explain about Grid data and feature data.

Feature Data:

http://desktop.arcgis.com/en/arcmap/10.3/manage-data/geodatabases/feature-class-basics.htm

Grid Data:

http://desktop.arcgis.com/en/arcmap/10.3/manage-data/raster-and-images/esri-grid-format.htm

Thank you

Dear Dr. Ray,

I found the materials very important. Thank you very much.