Soil stability

Why we monitor soil stability

The Waikato Regional Council monitors soil stability to get a measure of the quantity of soil in place (stable) and how much has been subjected to erosion and has the potential to generate sediment. It provides information about how particular land use disturbances and natural erosion is affecting the soil resource across the region. Accelerated erosion can lead to the loss of the soil's productive capability, damages to fences, tracks, roads and houses, damage to native habitats or animal habitats, adverse effects on the aesthetic and cultural values associated with land and sediment being generated that could enter waterways.

Measuring the stability of the soil gives us an understanding of its natural stability – how it has been affected by natural erosion processes and an understanding of how land management practices have affected its stability and accelerated erosion. The soil resource is important for farming, forestry, conservation and residential/commercial use so it is important to understand how well the soil resource is remaining in place and what are the factors influencing erosion.

What's happening?

Soil stability is a measure of how well the region’s soil is being kept in place as a resource for farming, forestry, and conservation. In 2017, 49 per cent of the observed sample points contained surfaces that were stable. These surfaces showed no signs of past natural erosion and are completely vegetated (unless the topsoil is disturbed by land use activities such as cultivation and tracking).

Approximately 26 per cent of the observed sample points contained surfaces that were unstable and erosion prone (these included surfaces which showed signs of past erosion but were not currently eroding). About 17 per cent of the sample contained land that had been recently eroded or was eroding at the time of the aerial survey. These areas require careful management and have a limited range of productive land uses.

Extensively disturbed surfaces included areas of land where soil had been removed in whole or in part, re-contoured, or covered by buildings pavements or water. In 2017, 8 per cent of the observed sample points contained surfaces that were extensively disturbed.

The changes in soil stability between 2007 and 2017 are shown in Figure 1. There has been a statistically significant increase in the percentage of sample points within the unstable surfaces (erosion prone) category from 24 per cent in 2007 to 26 per cent in 2017. This increase appears to have been at the expense of stable surfaces which decreased significantly from 52 per cent in 2007 to 49 per cent in 2017. Over the last five years, there has been a significant decrease in unstable eroded and eroding surfaces from 19 per cent in 2012 to 17 per cent in 2017. The percentage of sample points within the extensively disturbed category has not changed significantly over the preceding 5 or 10 years.

Graph - Percentage of sample points that contained stable, unstable and extensively disturbed land surfaces across the Waikato region, 2007 – 2017

Figure 1. Percentage of sample points that contained stable, unstable and extensively disturbed land surfaces across the Waikato region, 2007 – 2017.

How we monitor

The Waikato Regional Council uses aerial photography and a point sample technique to measure soil stability. This consists of one hectare sample areas spread over a 2km grids across the Waikato region.
The monitoring occurs every five years as aerial photography becomes available.
This point sample survey was completed for the first time in 2002 and repeated with aerial photography from 2007, 2012 and 2017. Category definitions were updated in 2007 and consequently only data from the 2007, 2012 and 2017 surveys is presented here.
Point sample analysis is a statistical technique that has long been applied in the natural sciences, to extract sample data from field sites, aerial photographs, or maps. Aerial photographs are visually inspected by someone who has landform and erosion knowledge and skills in photographic interpretation.
Within each sample area, the land stability and any soil disturbance as a result of land use or natural processes are recorded. Soil disturbance is measured by interpreting the one hectare sample area and recording any sign of disturbance. Wherever soil disturbance was recorded within a sample area, a cluster analysis is used to record the percentage of bare ground within each of those sample areas. This determines the percentage of soil bare in the one hectare sample area.
Although the 2 km grid is not spatially random, it does provide a random sample because the underlying landforms, soils and land uses are distributed irregularly across the region. In addition, the 2 by 2 km spacing provides sufficient sample areas to represent the whole of the region.
Burton AS, Taylor A, Hicks DL 2009. Assessing soil stability. In: Land and soil monitoring : a guide for SoE and regional council reporting. Hamilton, Land Management Forum. pp. 89-116
The raw data is stored in GIS layers.
A standard set of codes for photo interpretation has been developed to cover the main land use classes, surface type classes, land use disturbance classes and natural process classes.
A unique reference number for each sample point, from 1 to 6122 was automatically assigned by the GIS program ArcGIS. However because this number may change if the attribute table is re-opened in a different workspace, a locked identification number was assigned by a GIS specialist to each point, enabling their re-location during data sorts and analyses.
Sample point locations were stored as ArcGIS metadata. These are cross referenced to a screen display (shapefile) which shows their locations relative to region-wide orthophotos and map layers e.g. topographic; and a database (attribute table) which contains data recorded for all points. It is duplicated in an Excel spreadsheet which enables data sorting.
An initial data sort was carried out in an Excel spreadsheet, to check for consistency in the use of codes, and correct where necessary in the raw GIS layer. Subsequent generation of summary tables for categories of interest (e.g. soil stability, nature of disturbance) was conducted using custom scripts written in the R programming language.
Statistical functions (scripted in R) were applied to respective summary tables and data frames, enabling calculation of percentages and error margins for each category of interest along with changes over time.
Sort, point count and statistical analysis results for particular topics were stored as four series of summary spreadsheets.
Cluster sampling is used to measure the percentage of bare soil. It entails recording the incidence of bare soil at each of 100 dots set in a 10 x 10 grid within a one hectare area around the sample point. The number of dots with bare soil is recorded as an attribute.
Photo interpretation error is ascertained by randomly selecting 100 points and assessing them in the field against the photo interpretation results. They indicate reliability of photo-interpreted data stored for the points in the sample. For the completed surveys to date photo-interpretation accuracy has typically been in the 85% to 95% range. Most errors are simply confusion of land use etc. with another that is similar.
None relevant to this indicator.
ArcGIS is used to display and store the data from photo interpretation. The data is then exported to Excel and analysed using custom scripts written in the R programming language. The scripts are used to determine point counts, convert totals to percentages of the regional area and summarise the area in each land use (or area with each soil conservation cover as appropriate). Significance tests are also applied, to calculate confidence limits for results and changes over time. The indicator is compiled using data from the point samples.
This indicator is not an actual measure of soil erosion. It is a point sample method that provides regional data on soil stability and uses cluster analysis to provide a measure of bare ground. The point sample analysis technique has been designed to provide statistical data for regions. Region-wide samples are sufficiently large that they can also provide valid data for reasonably large subdivisions within a region. However, data analysis will not be reliable for soil stability/disturbance in an area of land any smaller than 100 km².
No changes are planned for this indicator with respect to the survey design and reporting frequency. However, results from future survey iterations will likely be integrated into a new State of the Environment (SOE) reporting framework (as opposed to the current web-based reporting format).

Data and trends

Soil disturbance is a measure of the region’s sample points which had soil disturbed due to either natural processes of erosion, deposition, or land use related activities.
In 2017, about 62 per cent of point samples had no soil disturbance and were therefore considered intact and well-vegetated. Soil disturbance (natural, through erosion/deposition, or as a result of land use disturbance) can lead to land degradation and loss of soil. Soil disturbance in the region accounted for 31 per cent of point samples in 2017. The actual measure of bare soil due to disturbance amounted to 2.31 per cent or 57,750 hectares of the region’s area.
About 21 per cent of the regions sample points were disturbed by land use related activities with associated bare soil accounting for 1.40 per cent of the region’s area or 35,000 hectares.
In 2017, the main pressure from land use related activity was from tracking which accounted for 0.95 per cent of total bare soil (Figure 2). There has been no significant change in bare soil related to tracking over the preceding 5- and 10-year periods.
The significant reduction in total bare soil observed over the last 10 years (2007 – 2017) reflects primarily the reduction in bare soil due to cultivation which totalled 0.80 per cent in 2007 compared to 0.23 per cent in 2017. There has also been a significant decline in bare soil exposed by crop harvest related activites (including forestry) from 0.17 per cent in 2007 to 0.06 per cent in 2017.

Figure 2. Change in bare soil by land use related activities, 2007 – 2017. Error bars represent the 95% confidence interval around the mean.
About 10 per cent of the region’s sampling points were disturbed by natural processes of erosion and deposition. Soils affected by natural disturbance can recover over time unless disturbed again.
Associated bare soil was 0.61 per cent or 15,250 hectares, significantly higher than the 0.38 per cent bare soil value observed in 2012. This increase was associated with a significant change from 2012 – 2017 in the surface erosion category, which is made up of sandblows, sheetwash, rock fall, or bare rock and geothermal classes (Figure 3).

Figure 3. Change in bare soil by natural process disturbance, 2007 – 2017. Error bars represent the 95% confidence interval around the mean.
Surfaces where the soil has been partly or completely removed by rural buildings, roads, urban areas, or waterbodies are categorised as extensively disturbed.
In 2017, 7.5 per cent of sample points were extensively disturbed with 4.0 per cent attributed to land use disturbance (i.e., rural buildings, roads, urban areas) and 3.5 per cent to natural disturbance (i.e., water bodies and coastal features). Associated bare soil was 0.29 per cent or 7250 ha. The percentage of sample points with bare soil associated with extensively disturbed areas has not changed significantly over the last 10 years (Figure 4).

Figure 4. Change in bare soil within extensively disturbed land classes, 2007 – 2017. Error bars represent the 95% confidence interval around the mean.
Figure 5 summarises the area of bare soil associated with land use disturbance in conservation and rural land classes for 2007 – 2017.
Dairy pasture accounted for the highest percentage of bare soil in 2017 (0.75 per cent), followed by drystock pasture (0.28 per cent) and forestry (0.17 per cent). Bare soil associated with conservation uses (land use related) remained low for all classes (<0.05 per cent). There were no significant changes in bare soil percentages across land uses classes between 2012 and 2017.

Figure 5. Change in bare soil by conservation and rural land use class. Error bars represent the 95% confidence interval around the mean.