Gerald Whittaker

A hybrid genetic algorithm for multiobjective problems with activity analysis-based local search

The objective of this research was the development of a method that integrated an activity analysis model of profits from production with a biophysical model, and included the capacity for optimization over multiple objectives. We specified a hybrid genetic algorithm using activity analysis as a local search method, and NSGA-II for calculation of the multiple objective Pareto optimal set. We describe a parallel computing approach to computation of the genetic algorithm, and apply the algorithm to evaluation of an input tax to regulate pollution from agricultural production.

Optimization-based trade-off analysis of biodiesel crop production for managing an agricultural catchment

Political agendas worldwide include increased production of biofuel, which multiplies the trade-offs among conflicting objectives, including food and fodder production, water quantity, water quality, biodiversity, and ecosystem services. Quantification of trade-offs among objectives in bioenergy crop production is most frequently accomplished by a comparison of a limited number of plausible scenarios. Here we analyze biophysical trade-offs among bioenergy crop production based on rape seed, food crop production, water quantity, and water quality in the Parthe catchment in Central Germany. Based on an integrated river basin model (SWAT) and a multi-objective genetic algorithm (NSGA-II), we estimated Pareto optimal frontiers among multiple objectives. Results indicate that the same level of bioenergy crop production can be achieved at different costs with respect to the other objectives. Intermediate rapeseed production does not lead to strong trade-offs with water quality and low flow if a reduction of food and fodder production can be accepted. Compared to solutions focused on maximizing food and fodder yield, solutions with intermediate rapeseed production even improve with respect to water quality and low flow. If rapeseed production is further increased, negative effects on low flow prevail. The major achievement of the optimization approach is the quantification of the functional trade-offs for the feasible range of all objectives. The application of the approach provides the results of what is in effect an infinite number of scenarios. We offer a general methodology that may be used to support recommendations for the best way to achieve certain goals, and to compare the optimal outcomes given different policy preferences. In addition, visualization options of the resulting non-dominated solutions are discussed.

Detection of Overparameterization and Overfitting in an Automatic Calibration of SWAT

Distributed hydrologic models based on small-scale physical processes tend to have a large number of parameters to represent spatial heterogeneity. This characteristic requires the use of a large number of parameters in model calibration. It is a common view that calibration with a large number parameters produces overparameterization and overfitting. Recent work using prior information, spatial information, and constraints on parameters for regularization of the calibration problem has improved model predictions using a few dozen parameters. We demonstrate that the Soil and Water Assessment Tool (SWAT) and the information associated with a SWAT watershed setup provide a regularized problem with many of recently published regularization techniques already utilized in SWAT. Our hypothesis is that the Soil and Water Assessment Tool (SWAT) regularizes the inverse problem so that a stable solution can be obtained for calibration of SWAT using a very large number of parameters, where very large means up to 10,000 calibration parameters. In this study, a two-objective calibration genetic algorithm based on a non-dominated sorting genetic algorithm (NSGA-II) was used to calibrate the Blue River basin in Oklahoma. We introduce the use of intermediate solutions found by the genetic algorithm to test identification of calibration parameters and diagnose model overfitting. Defining identification as the capability of a model to constrain the estimation of parameters, we introduced a method for statistically testing for changes from the initial uniform distribution of each parameter. We found that all 4,198 parameters used to calculate the Blue River SWAT model were identified. Diagnostic comparisons of goodness-of-fit measures for the calibration and validation periods provided strong evidence that the model was not overfitted.

High resolution characterization of soil biological communities by nucleic acid and fatty acid analyses

Fatty acid methyl ester (FAME) and length heterogeneity-polymerase chain reaction (LH-PCR) analyses were used to generate ‘fingerprints’ of FAMEs and eubacterial 16S rDNA sequences characteristic of agricultural soil communities. We hypothesized that pooling data from two methods that characterized different components of soil biological communities would improve the resolution of fingerprints characterizing the effects of contrasting tillage and ground cover practices. By using supervised classifications of FAME and LH-PCR, a discriminant analysis procedure distinguished soils from contrasting tillage and ground cover management and predicted the origin of soil samples. Used independently, FAME provided higher resolution of tillage, ground cover, and field location than LH-PCR, but LH-PCR was effective at identifying field location. Pooling data from both methods did not enhance the predictive power. A comparison of linear discriminant analysis, quadratic discriminant analysis, and nonparametric density estimation demonstrated that minimizing assumptions about data distribution improved the capacity of FAME analysis to resolve differences in soil types. Use of a purely statistical Bayesian method to select a subset of fatty acids (FA’s) as variables in discriminant analyses identified FA’s that represented signature FA’s for specific groups of organisms. Published by Elsevier Science Ltd.

Impact of land use patterns and agricultural practices on water quality in the Calapooia River Basin of western Oregon

Agricultural practices, including tillage, fertilization, and residue management, can affect surface runoff, soil erosion, and nutrient cycling. These processes, in turn, may adversely affect (1) quality of aquatic resources as habitat for amphibians, fish, and invertebrates, (2) costs of treating surface and ground water to meet drinking water standards, and (3) large-scale biogeochemistry. This study characterized the surface water sources of nitrogen (N) (total, nitrate [NO3−], ammonium [NH4+], and dissolved organic N) and sediment active within 40 subbasins of the Calapooia River Basin in western Oregon in monthly samples over three cropping years. The subbasins included both independent and nested drainages, with wide ranges in tree cover, agricultural practices, slopes, and soils. Sediment and N form concentrations were tested against weather and agricultural practice variables. Subbasin land use ranged from 96% forest to 100% agriculture. Average slopes varied from 1.3% to 18.9%, and surface water quality ranged from 0.5 to 43 mg L−1 (ppm) total N maxima and 29 to 249 mg L−1 suspended sediment maxima. Total N during the winter was positively related to percentage landcover of seven common agricultural crops (nongrass seed summer annuals, established seed crops of perennial ryegrass [Lolium perenne L.], tall fescue [Schedonorus phoenix {Scop.} Holub], orchardgrass [Dactylis glomerata L.], clover [Trifolium spp.], and newly planted stands of perennial ryegrass and clover) and negatively related to cover by trees and one seed crop, Italian (annual) ryegrass (Lolium multiflorum). Results for NO3− and total N were highly similar. Sediment concentrations were most strongly related to rainfall totals during periods of 4 and 14 days prior to sampling, with smaller effects of soil disturbance. Fourier analysis of total N over time identified four prominent groups of subbasins: those with (1) low, (2) medium, and (3) high impacts of N (up to 2, 8, and 21 mg L−1, respectively) and a strong cyclical signal peaking in December and (4) those with very high impact of N (up to 43 mg L−1) and a weak time series signal. Preponderance of N in streams draining agriculturally dominated subbasins was in the form of the NO3− ion, implying mineralization of N that had been incorporated within plant tissue following its initial application in the spring as urea-based fertilizer. Since mineralization is driven by seasonal rainfall and temperature patterns, changes in agronomic practices designed to reduce prompt runoff of fertilizer are unlikely to achieve to more than ~24% reduction in N export to streams.

Restricting Pesticide Use: The Impact on Profitability by Farm Size

A sample of 226 cash grain farms in the Lake States-Corn Belt region are analyzed to estimate the impact of restricting pesticide use on profits. These 226 farms are classified into small, medium, and large farms according to their sale revenues. The results suggest the existence of pest management practices that could substantially reduce pesticide use without incurring economic losses. The reductions in profits associated with gradual reductions in pesticide expenditure appear to increase with farm size.

Conservation Practices in Western Oregon Perennial Grass Seed Systems

Rapid changes in practices used to produce perennial grass seed crops in the U.S. Pacific Northwest region and shortened lengths of time that perennial grass seed fields remain in production have increased the need for additional rotation crops that are adapted to the poorly drained soils found in western Oregon. This research was conducted at three sites to determine ways to manage meadowfoam (Limnanthes alba Hartw. ex Benth.) as a component in perennial grass seed rotation systems. Experiments were conducted in 1997, 1998, and 2001 to investigate combinations of spring-applied herbicide and N fertilizer and times of applications, direct-seeded and conventional tillage establishment methods, and previous crop residue management on meadowfoam seed yield, seed oil concentration, and oil yield. No spring-applied fertilizer or herbicide produced responses for all yield components as great as or greater than any other treatment combination. Direct-seeded meadowfoam yielded more oil than the conventional establishment treatment. There was no effect of residue management amounts from grass seed grown in the previous rotation sequence on meadowfoam production; however, maximal residue management, especially if used in combination with direct-seeded meadowfoam, should reduce annual soil erosion. Meadowfoam is suited to low-input production and is adapted to the use of conservation practices including direct seeding and maximal residue management in perennial grass seed systems.

GIS Analysis of Spatial Clustering and Temporal Change in Weeds of Grass Seed Crops

Abstract Ten years of Oregon Seed Certification Service (OSCS) preharvest field inspections converted from a nonspatial database to a geographic information system (GIS) were analyzed for patterns in spatial distribution of occurrence and severity of the 36 most common weeds of grass seed crops. This was done under the assumptions that those patterns would be primarily consequences of interactions among farming practices, soil properties, and biological traits of the weeds, and that improved understanding of the interactions would benefit the grass seed industry. Kriging, Ripleys K-function, and both Morans I spatial autocorrelation and Getis-Ord General G high/low clustering using the multiple fixed distance band option all produced roughly similar classifications of weeds possessing strongest and weakest spatial clustering patterns. When Morans I and General G analyses of maximum weed severity observed within individual fields over the life of stands were conducted using the inverse distance weighting option, however, results were highly sensitive to the presence of a small number of overlapping fields in the 10-yr record. Addition of any offset in the range from 6 to 6,437 m to measured distances between field centroids in inverse distance weighting matrices removed this sensitivity, and produced results closely matching those for the multiple fixed distance band method. Clustering was significant for maximum severity within fields over the 10-yr period for all 43 weeds and in 78% of single-year analyses. The remaining 22% of single-year cases showed random rather than dispersed distribution patterns. In decreasing order, weeds with strongest inverse-distance spatial autocorrelation were German velvetgrass, field bindweed, roughstalk bluegrass, annual bluegrass, orchardgrass, common velvetgrass, Italian ryegrass, Agrostis spp., and perennial ryegrass. Of these nine weeds, distance for peak spatial autocorrelation ranged from 2 km for Agrostis spp. to 34 km for common velvetgrass. Weeds with stronger spatial autocorrelation had greater range between distance of peak spatial autocorrelation and maximum range of significance. Z-scores for General G high/low clustering were substantially lower than corresponding values for Morans I spatial autocorrelation, although the same two weeds (German velvetgrass and field bindweed) showed strongest clustering using both measures. Simultaneous patterns in Morans I and General G implied that management practices relatively ineffective in controlling weeds usually played a greater role in causing weeds to cluster than highly effective practices, although both types of practices impacted Italian ryegrass distribution. Distance of peak high/low clustering among perennial weeds was smallest (1 to 3 km) for Canada thistle, field bindweed, Agrostis spp., and western wildcucumber, likely indicating that these weeds occurred in patchy infestations extending across neighboring fields. Although both wild carrot and field bindweed doubled in average severity over the period from 1994 to 2003, wild carrot was the only weed clearly undergoing an increase in spatial autocorrelation. Soil chemical and physical properties and dummy variables for soil type and crop explained small but significant portions of total variance in redundancy and canonical correspondence analysis of weed occurrence and severity. Fitch-Morgoliash tree diagrams and Redundancy Analysis (RDA) and Canonical Correspondence Analysis (CCA) ordinations revealed substantial differences among soil types in weed occurrence and severity. Gi* local hot-spot clustering combined with feature class to raster conversion protected grower expectations of confidentiality while describing dominant spatial features of weed distribution patterns in maps released to the public. Nomenclature: Annual bluegrass, Poa annua L. POAAN; Canada thistle, Cirsium arvense (L.) Scop. CIRAR; common velvetgrass, Holcus lanatus L. HOLLA; field bindweed, Convolvulus arvensis L. CONAR; German velvetgrass, Holcus mollis L. HOLMO; Italian ryegrass, Lolium multiflorum Lam. LOLMU; orchardgrass, Dactylis glomerata L. DACGL; perennial ryegrass, Lolium perenne L. LOLPE; roughstalk bluegrass, Poa trivialis L. POATR; western wildcucumber, Marah oreganus (T. & G.) T. J. Howell ECNOR; wild carrot, Daucus carota L. DAUCA.

Multivariate applications of the ash in regression

A simple algorithm for estimating the regression function over the United States is introduced. The approach allows for data obtained from a complicated sampling design, as well as for the inclusion of a few additional covariates. The regression estimates are obtained from an associated probability density estimate, namely the averaged shifted histogram. The algorithm has proven especially successful over a large mesh, say 300 by 200 nodes, in a data rich setting, even on a 486 computer running Splus. We currently run much higher resolution meshes on a Pentium. Commonly available alternative codes including kriging failed to produce useful estimates in this setting.

Remote sensing classification of grass seed cropping practices in western Oregon

Our primary objective was extending knowledge of major crop rotations and stand establishment conditions present in 4800 grass seed fields surveyed over three years in western Oregon to the entire Willamette Valley through classification of multiband Landsat images and multi-temporal Moderate Resolution Imaging Spectroradiometer (MODIS) 16-day composite Normalized Difference Vegetation Index (NDVI). Mismatch in resolution between MODIS and Landsat data was resolved by edging of training and test validation areas using 3 by 3 neighbourhood tests for class uniformity, resampling of MODIS data to 50-m resolution followed by 3 by 3 neighbourhood smoothing to artificially enhance resolution, and resampling to 30 m for stacking data in groups of up to 64, 55 and 81 bands in 2004–2005, 2005–2006 and 2006–2007. Imposing several object-based rules raised final classification accuracies to 84.7, 77.1 and 87.6% for 16 categories of cropping practices in 2005, 2006 and 2007. Total grass seed area was under-predicted by 3.9, 5.4 and 1.8% compared to yearly Cooperative Extension Service estimates, with Italian ryegrass overestimated by an average of 8.4% and perennial ryegrass, orchardgrass and tall fescue underestimated by 10.4, 3.3 and 2.1%. Knowledge of field disturbance patterns will be crucial in future landscape-level analyses of relationships among ecosystem services.