Soil Sampling as a Basis for Fertilizer Application (continued)

SF-990 (Revised) August 1998

Soil Sample Handling

Samples intended for NO₃-N sampling should be stored in ice chests during transport. Moist samples subjected to heat will increase N mineralization and test values will increase during transport/storage. Samples intended for NO₃-N determination should be air-dried immediately after collection to prevent alteration of NO₃-N concentrations due to microbial activity. Spread samples uniformly on clean paper in a dust free area. Another procedure is to transport the samples immediately to a soil testing laboratory in a cold ice chest. Usually, the soil laboratory attaches a drying charge for wet soil samples. Rubber gloves should be used to handle samples intended for chloride analysis to prevent contamination from chloride in perspiration.

Soil samples intended for Zn analysis should not come into contact with any galvanized surface, including the soil sampling tool, bucket, drying container or grinder.

Soil Sample Collecting, Where and How

Where to collect a soil sample and how many samples to collect depends on the sampling goal. Traditionally in North Dakota, the goal has been to provide one soil test level to describe a field. This approach works well in some situations, especially when the test value is low. However, because of the variability of nutrients in the field, one test level from a field may not represent a large part of the field. Some producers, having received a high soil test report, continue to apply the same fertilizer rates as in the past because they lack confidence in the test. Recent research has developed methods to increase the confidence in soil test values while keeping sampling costs low.

Sampling goals can be separated into two categories; determining nutrient levels in Whole Fields, or determining Within Field Values.

Determining whole field nutrient values

Collecting a selectively random sample composite is the traditional North Dakota sampling strategy for determining whole field nutrient values. A field composite sample should consist of at least 20 selectively random soil cores. A field sampled in this manner should give the field mean plus or minus 15% at least 80% of the time (Figure 4). Selectively random sampling means that the field is sampled only in areas which represent most of the field area. Unusual landscape features such as eroded areas, saline or sodic zones and old building lots are not sampled. Also, avoid sampling in dead furrows or back furrows, under old manure or hay piles, sugarbeet tare piles, animal droppings, next to ditches, sloughs and roads, known banded fertilizer locations, and small depressions.

Figure 4. The number of subsamples required for a composite soil sample for NO₃-N with various levels of accuracy for an 80�H precision level. (Adapted from Swenson et al., 1984).

There are often questions about what constitutes a "field." Some samplers collect one composite sample per section or one per quarter-section. Others separate the field into large landscape zones and treat each as a field. Some may divide a quarter into three to four equal sub-fields and sample each individually. Generally, the smaller the area, the more representative of the area the sample values will be. Figure 5 shows two examples of a suggested way to obtain representative samples from fields.

Figure 5.

Using a composite soil sample to direct fertilizer recommendations has several advantages:

• It is relatively inexpensive. Soil sampling is relatively quick, only 20 to 30 cores are needed to represent a field, and only one analysis is required for each field.
• Results are mostly reproducible.
• Results can easily be tracked from year to year.

Composite soil samples, however, have several inherent disadvantages:

• "Unusual areas" not sampled may comprise significant acreage in a field.
• Large portions of the field may be over- or under-fertilized.
• There is a low level of confidence that high soil test values represent most of the field.
• Sometimes it is difficult to distinguish which locations are unusual.

Composite sampling is most representative when within field variability is low. Low within field variability is most common when composite soil test levels are low. A field composite test of 20 lb NO₃-N/acre means that at least 95% of the area sampled contains levels between 10 and 30 lb NO₃-N/acre.

Collecting at least 20 soil cores from a field results in a large amount of soil being collected. In some soils, such as fine sandy loams, the soil may break up easily in a bucket, enabling thorough mixing before a 2/3 pint subsample is obtained for analysis. However, many soils do not break up easily. It may be necessary to take the entire sample out of the field, dry and grind it to obtain a good mixture. The resulting sample, whatever the method of collection and preparation, must represent the 20 core locations to provide the most accurate and reproducible results.

Sampling for within-field nutrient levels

Because of the limitations of composite soil testing, and because of the growing popularity of site-specific farming, different methods of obtaining nutrient values within fields are needed. Sampling for determining within-field nutrient levels can be accomplished through two different methods; grid sampling and directed sampling. Grid sampling reveals fertility patterns through dense systematic sampling, while the directed sampling method assumes there is a predictable and logical reason for fertility patterns to exist and uses this reason to reduce sample number while maintaining high quality information compared to dense grid sampling. Directed sampling has also been called "zone sampling," "smart sampling"and "smart zones."

Grid sampling

Grid samples were first taken in a regular, predictable pattern across the field (Figure 6).

However, the regular grid can easily contain bias because of streaking of fertilizer or manure applications in the past. With GPS technology (Global Positioning Satellite receivers), grid sampling need not be regularly spaced. Irregularly spaced interval positions can reproducibly be located as accurately as regularly spaced grids. Irregular grids, such as the systematic unaligned grid, also provide the opportunity for greater statistical evaluation through a process called "kriging" (pronounced "kreeging"). Many researchers prefer kriging as an estimator of values between actual samples because it carries an estimate of error along with the estimated value. Other estimators such as inverse distance, polynomial and triangulation carry no such estimate of error. Other grid sampling types are random, random stratified, staggered start, and the diamond/triangle/hexagon grid pattern.

Grid sampling can be a good tool for sampling within field nutrient levels if samples are taken densely enough. The accepted grid spacing from recent research, including in North Dakota, is about one sample per acre. This approach, however, is very expensive and time-consuming, and has forced many commercial soil samplers and producers to accept less information about their fields and use a 2.5 acre grid or larger. In North Dakota, even a 2.5 acre grid is considered expensive and prohibitive. A 4-5 acre grid is more commonly used. The 4-5 acre grid has been used to reveal variability in soil test levels, but it may not be very accurate in representing within-field nutrient levels nor does it represent fertility patterns well (Figure 7). The use of a 4-5 acre grid should not be considered a dense systematic grid.

Figure 7. Comparison at Valley City, 1995, of NO₃-N mapping using topography and selected grid spacings.

|  index  |  more   |

SF-990 (Revised) August 1998