SOIL MOISTURE PREDICTION USING SUPPORT VECTOR MACHINES1

ABSTRACT: Herein, a recently developed methodology, Support Vector Machines (SVMs), is presented and applied to the challenge of soil moisture prediction. Support Vector Machines are derived from statistical learning theory and can be used to predict a quantity forward in time based on training that uses past data, hence providing a statistically sound approach to solving inverse problems. The principal strength of SVMs lies in the fact that they employ Structural Risk Minimization (SRM) instead of Empirical Risk Minimization (ERM). The SVMs formulate a quadratic optimization problem that ensures a global optimum, which makes them superior to traditional learning algorithms such as Artificial Neural Networks (ANNs). The resulting model is sparse and not characterized by the “curse of dimensionality.” Soil moisture distribution and variation is helpful in predicting and understanding various hydrologic processes, including weather changes, energy and moisture fluxes, drought, irrigation scheduling, and rainfall/runoff generation. Soil moisture and meteorological data are used to generate SVM predictions for four and seven days ahead. Predictions show good agreement with actual soil moisture measurements. Results from the SVM modeling are compared with predictions obtained from ANN models and show that SVM models performed better for soil moisture forecasting than ANN models.     

Alternative mating tactics in dimorphic males of the harvestman <Emphasis Type="Italic">Longiperna concolor</Emphasis> (Arachnida: Opiliones)

Intense male–male competition for females may drive the evolution of male morphological dimorphism, which is frequently associated with alternative mating tactics. Using modern techniques for the detection of discontinuous allometries, we describe male dimorphism in the Neotropical harvestman Longiperna concolor, the males of which use their elongated, sexually dimorphic legs IV in fights for the possession of territories where females lay eggs. We also tested three predictions related to the existence of alternative mating tactics: (1) if individuals with relatively longer legs IV (majors) are more likely to monopolize access to reproductive resources, they are expected to remain close to stable groups of females more than individuals with relatively shorter legs IV (minors) do; (2) if minors achieve fertilization by moving between territories, they are expected to be less faithful to specific sites; and (3) majors should be observed in aggressive interactions more often. We individually marked all the individuals from a population of Longiperna during the reproductive season and recorded the location of each sighting for males and females as well as the identity of males involved in fights. Majors were more likely to have harems, and large majors were even more likely to do so. Majors were more philopatric and all males involved in fights belonged to this morph. These results strongly suggest that the mating tactic of the majors is based on resource defense whereas that of the minors probably relies on sneaking into the territories of the majors and furtively copulating with females.     

Persistent environmental pollutants in eggs of aplomado falcons from Northern Chihuahua, Mexico, and south Texas, USA

The northern aplomado falcon (Falco femoralis septentrionalis) disappeared from south Texas in the 1940s. Due to great success in the release of captive-reared aplomado falcons in south Texas, there are currently more than 40 established nesting pairs in the region. Addled eggs from aplomado falcons nesting in northern Chihuahua and south Texas were analyzed to determine organochlorine (OC) and inorganic element contaminant burdens and their potential association with egg failures and effects on reproduction. Among the OCs, DDE [1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene] was present at the highest concentrations (range 262-21487 ng/g wet weight) followed by polychlorinated biphenyls (PCBs, range 88-3274 ng/g ww). DDE was greater (P=0.03) in eggs from El Sueco (Chihuahua, Mexico) than in those from Matagorda Island (Texas, USA). DDE concentrations in eggs of aplomado falcons from El Sueco were elevated; however, reproductive success in the two Chihuahuan populations did not seem to be affected by DDE. DDE and metals in potential avian prey of the aplomado falcon from Matagorda Island were very low and below levels in the diet at which some negative effects might be expected. Except for mercury (Hg), metal concentrations in eggs were fairly low and were not different among locations in Chihuahua and south Texas. Hg was somewhat elevated and was greater (P<0.001) in Texas than in the Chihuahua locations. Periodic monitoring of Hg concentrations in addled eggs of aplomado falcons in south Texas is recommended to continue evaluating potential negative effects on their recovery.     

Collaborative Testing of Methods to Measure Air Pollutants II. The Non-Dispersive Infrared Method for Carbon Monoxide

The Methods Standardization Branch of the Environmental Protection Agency, National Environmental Research Center, has undertaken a program to standardize methods used in measuring air pollutants covered by the national primary and secondary air quality standards. This paper presents the results of a collaborative test of the method specified for carbon monoxide.

The test involved analysis of CO in air samples (in cylinders) by participating laboratories. Three concentrations, covering the range of the method which is, 0 to 58 mg/m³, were analyzed dry and humidified on each of three days by 15 collaborators. The method of analysis, nondispersive infrared spectrometry (NDIR), involved an NDIR instrument in combination with different procedures for eliminating water vapor interference. A statistical analysis of the data obtained produced the following results: 1. The checking limit for duplicates (replication error) is 0.5 mg/m³.

2. The repeatability (variation within a laboratory) is 1.6 mg/m³.

3. The reproducibility (variation between laboratories) varies nonlinearly with concentration; i.e., a minimum of 2.3 mg/m³ at a concentration of 20 mg/m³ and ranges as high as 4.3 mg/m³ in the concentration range of 0 to 58 mg/m³.

4. The reproducibility at the level of the national primary ambient air quality standard, 10 mg/m³-8-hour average, is 2.5 mg/m³ or 25%.

5. The minimum detectable sensitivity is estimated to be 0.3 mg/m³.

6. Compensation for water vapor interference is satisfactorily accomplished using drying agents and refrigeration methods. The use of narrow-band optical filters alone may not provide adequate compensation.

7. The accuracy obtained depends upon the availability of reliable calibration gases. Based on the results of this study, the method produces results that average 2.5% high.

Future papers will contain test results for methods to measure other air pollutants.     

Collaborative Testing of Methods to Measure Air Pollutants

A collaborative test was conducted to determine the precision of the chemiluminescent method which has been specified for measuring ozone, to determine photochemical oxidant. Ten laboratories participated in a test involving the analysis of an urban atmosphere containing a photochemical smog mixture. Ozone generators were used to increase the O₃ level over that naturally occurring, in order to cover an adequate range of concentrations. The range tested was 0 to 0.5 ppm.

A statistical analysis of the data obtained was used to derive equations for within laboratory and between laboratory standard deviations. In order to evaluate sampling data, these equations can be used in various statistical procedures to estimate repeatability, reproducibility, lower detectable limit, and other measures that establish the precision of the method.

Using specific definitions for repeatability and reproducibility, the following approximate estimates were obtained in the range of zero to 0.5 ppm:

Repeatability—0.01 to 0.04 ppm (varies with concentration, linear).

Reproducibility—0.01 to 0.09 ppm (varies with concentration, non-linear).

The lower detectable limit depends on instrumental and other variables, and cannot be specified precisely. Under typical assumptions, this limit can be estimated at between 0.006 and 0.009 ppm.     

Life after the layoff: getting a job worth keeping

The competitive environment of business today makes corporate layoffs an organizational reality, and losing one's job can be a highly stressful experience. We propose and test a model that places objective underemployment and subjective underemployment in a causal sequence between organizational actions and employees' restoration of equilibrium by obtaining jobs worth keeping. We longitudinally examine relationships between layoff fairness, workers' stress symptoms and appraisal, and subsequent employment outcomes among 149 laid‐off technical employees over the course of one year. Structural equation model results support seven of nine hypothesized paths, and demonstrate discriminant validity between and mediational properties of objective and subjective underemployment. Findings also reveal the important role that employees' perceptions and subjective assessments play in successfully returning to pre‐job loss equilibrium following displacement. Copyright © 2008 John Wiley & Sons, Ltd.     

On the effects of the dispersant Corexit 9500© during the degradation process of n-alkanes and PAHs in marine sediments

In many coastal countries, oil spill contingency plans include several alternatives for removal of the spilled oil from the ocean. Frequently, these plans include dispersants. Because this process applies chemical substances that may add toxicity to oil that already contains toxic compounds, it is, at times, a controversial method to fight oil pollution. Additionally, local conditions may result in particular complications. We investigated the possible effects of the dispersant Corexit 9500© under conditions similar to those of subtropical oceans. We used fuel oil #6+ diesel as the test mixture. Under certain conditions, at least part of the dispersed oil may reach the sediment, particularly if the dispersant is applied in coastal waters. Nine experimental units were used in this experiment. Similar conditions of water temperature, salinity, air fluxes into the experimental units, and hydrocarbon concentrations in sediments were used. Two treatments and one control, each one with three replicates, were carried out. We concentrated our investigation on sediment, although measurements of water were also taken. Our results suggest that once the oil has penetrated the sediment, no significant differences exist between oil that contains dispersant and oil without dispersant. Noticeable degradation of aliphatic hydrocarbons occurred mainly in the low molecular weight aliphatic hydrocarbons and not in the others. Apparently, degradation of aromatics was easier than that of alkanes. However, some differences were noticed for the degradation of PAHs in the sediment, suggesting a faster degradation under particular conditions in aerobic environments such as under this experiment.