GEOG 6010 Geocomputation

Site Suitability Analysis for Hop Cultivation using Fuzzy Logic

Introduction

A member of the Cannabaceae family, hops are a perennial flowering bine and have been cultivated since the mid 700s AD. (Dodds, 2011)  Arriving in the US in the 1600’s, hops were cultivated to support the brewing industries set up by colonists. (Dodds 2017) Since the 1900’s the majority of hop production has been confined to Oregon and Washington (Kopp 2011, USA Hops).  After prohibition, Idaho began growing hops and the three states have remained the primary producers of hops for the US (USA Hops). 

Hop cultivation requires a specific environment to be productive and financially viable. (Dodds 2017)  There are several important biophysical factors that affect cultivation; hops are susceptible to certain soil drainages, pH, temperature, rainfall, slope, and elevation. (Dodds 2017)  A site suitability analysis for ideal hop locations would require the consideration of these biophysical factors along with eligible land use information. Similar studies of site suitability for burgeoning agricultural markets have proved very useful for future planning (Jones et al. 2014, Daccache et al. 2012, Radiarta et al.  2011, Selim et al. 2018, Butsic et al. 2017) The biophysical traits that will be used in this analysis are: soil, temperature, rainfall, slope, elevation and landuse.

As many of these variables affecting hop cultivation are not obviously discrete categories, fuzzy logic has been employed to create scales of favorable conditions.(Abrahart 2014)  Fuzzy logic has proved a viable approach to handling biophysical parameters for site suitability analysis. (Teh et al. 2011, Chen et al. 2013) Creating a scaled score, based on the input parameters, can provide a broader understanding of potentially suitable locations, rather than a binary output of favorable and unfavorable locations.(Teh et al. 2011)

Methods

Site suitability analysis often requires reconciling different data formats to achieve meaningful results. (Selim et al. 2018)  In this study data is provided in vector and raster format. The area of focus for this analysis is the state of Idaho. Idaho is an emergent area for hop cultivation.  Compared to Washington and Oregon, Idaho is a relatively new market for hop cultivation. Idaho, as of 2015, produces 10% of the hop market, see Figure 1. Oregon produces the next closets amount, approximately 15% nationally.(USA Hops)  The analysis covers areas in Idaho currently under production to new areas that could potentially be put into production.

Soil

Soil data was obtained from the National Resource for Conservation Science.  As soil nutrient levels were not Determining the soils with best drainage involved a comparison of soil classifications from the NRCS database.  Each soil type is provided with a physical description; those fitting the ideal for drainage are classified as a 1. Those soil types not compatible with good drainage were classified as 0, see Figure 2.  

Temperature

Temperature data was provided from the PRISM group in Oregon.  As hop plants benefit from temperatures below 40-43°F, the average temperature was used in this assessment.  Monthly averages were obtained from the PRISM research group for the state of Idaho from 2012 to 2017 for the coldest months of the year, October through March.  Data was provided as rasters, based on interpolation from the PRISM group.  Pixels with average temperatures above 6°C were given a 0, pixels with temperatures below 4.4°C were given a one.  The temperature “grey” area between 6°C and 4.4°C was classified using fuzzy logic. The membership method is continuous, and values between 6-4.4°C are calculated to be between 1 and 0. Figure 3 illustrates the classification outputs.

Rainfall

Rainfall was provided for the state of Idaho by the PRISM group as well.  Rainfall was calculated as an average over a 40 year period, 1981 to 2010, and provided as a raster file.  The unit of the raster file was total annual inches. This file was scaled, so greater rainfall was given a higher weight, and values fell between 0 and 1, see Figure 4.  The intention here was to identify areas needing less irrigation to maintain the hop crop.  

Slope

Steep slopes require more intensive preparation and soil maintenance than gentler slopes. (US Dept. of Agriculture, 1925)  There are three broad ranges of slopes for agriculture. Slopes less than 1.5% are considered ideal for cultivation and require little to no extra preparation to plant or harvest. (Dodds 2017, US Dept. of Agriculture 1925)  Greater than 1.5% but less than 5% are still seen as viable farmland, the plot of land being cultivated may require some grading. (US Dept. of Agriculture 1925) Slopes steeper than 5% are generally not given consideration for farmland, as it requires labor intensive practices to prevent soil erosion and harvest crops with modern machinery (Dodds 2017).   The slope raster was derived from the National Elevation Dataset DEM at a 30m resolution. Slope was assigned a value of 1 if the slope was less than 1.5%. For slope greater than 12%, US Dept. of Agricultural recommends slopes over 12% not be developed for agriculture, a value of 0 is assigned. Between 1.5% and 12% the inverse of the slope is assigned to the pixel, so gentler slopes will be given values closer to 1 and steeper slopes will be calculated with smaller values. These outputs can be seen in Figure 5. 

Elevation

Hop cones are quite susceptible to bruising when exposed to strong winds, which can diminish the quality and yield of a hop farm. (Dodds 2017)  Traditional locations for farms are in sheltered valleys or lowers areas with wind breaks. (Dodds 2017) The elevation data was provided by the National Elevation Dataset DEM at a 30m resolution.  Idaho has a mountainous terrain, with many isolated ridges and hills that would potentially be exposed to wind. Canyon floors are not ideal farming locations, transportation can be difficult as well as finding contiguous land parcels to develop.  To identify these specific settings, the elevation data set was categorized to exclude the most extreme values. The raster was first recalculated to assign the inverse of the elevation to give pixels with lower elevation a higher value than pixels at higher elevations.  A manual inspection of these pixel values, and corresponding land features, identified a range between 0.4 and 0.06 as adequately eliminating areas of exposed high elevation and subsequent canyon floors between these peaks. Two rasters were created with separate ranges within this range to assign pixels with fringe values lower values between 0 and 1.  One raster was calculated to assign values between 0.45 and 0.06 as 1 and values outside this range to zero. The second raster was recalculated to assign pixels between 0.4 and 0.07 a 1 and pixels outside this range a 0. The two rasters were averaged to generate a raster landscape with three pixel values between 0 and 1, as shown in Figure 6.

Landuse

Zoning and land available for farming is an important consideration for this analysis.  The National Agricultural Statistic Service maintains a datafile with classifications of land use in the US, including land currently under cultivation, highly developed land, water surfaces, forest, land that is perennially ice and fallow land.  The dataset is available as a raster file with land use categories as pixel values. This raster was recalculated to assign all pixel values not suitable to farming a zero and all other values a 1, see Figure 7. This is the only variable in the site suitability analysis that is not classified using fuzzy logic as it is intended to be a mask to focus the analysis.

Site Suitability Analysis

The resulting rasters from each variable above were scaled 0 to 1 during the creation of fuzzy memberships.  These rasters were then averaged to produce a final raster with a score between 0 and 1 for each pixel, as shown in Figure 8.

Results

The final output raster resulted in a broad range of values.  All variables were scaled 1 to 0, with 1 being acceptable to cultivation and 0 unacceptable.  There are several distinct regions that emerge as the most suitable for hop cultivation, the lower elevation band running east to west in the southern portion of the state, as well as an area in the northwest region of the state.  While much of the state appears moderately acceptable to hop production, these areas have a much higher score. When comparing this output to the Cropland layer provided by the NASS, it appears as much of these areas are also primary agricultural regions in the state.  

It was anticipated that there would be overlap with current hop production and parcels considered optimal for cultivation.  As evidenced in the existing hop farms map, see Figure 9, current production is indeed in an ideal location. The cluster featured in the map is the largest and densest region of hop production in Idaho.  Much of the area is in the upper 10% of suitable areas for cultivation.

Conclusions

The application of fuzzy logic to biophysical variables related to hop cultivation has shown promise in aiding the hop industry.  The site suitability analysis has resulted in understanding of other regions in Idaho where the cultivation of hops could be successful.  Current cultivation is located primarily in the southern region, with the bulk of farms located around Boise in western Idaho. There are many opportunities for farmers to expand into the northern region of Idaho nearer to Couer D’Alene.  As this region is physically similar to the southern region in all but temperature and average rainfall, moving production into the northern part of Idaho may prove beneficial for hop yields. Growing in colder climates may not have as long of growing season, however the threat of downy mildew and other warm weather pests is significantly reduced.

It is of some considerable interest to note how a very large amount of current production is occurring in areas ideal for hop cultivation.  It was expected, as Idaho is a burgeoning hop market with an established agricultural presence, that a larger portion than was indicated would be in less optimal locations.  That this is not the case indicates a more sophisticated level of understanding of the requirements for hop cultivation, and suggests that further study on this subject should perhaps include assessments from local farmers.

For more information, read the full report here.

MSGIS Program Skills

• GIS Analysis

• Spatial Data and Algorithms

• GIS Workflow

• Model Building

• Cartography and Graphic Design

• Spatial Analysis

• Structured Query Language

• Communication Skills