Improvement of spatial interpolation accuracy of daily maximum air temperature in urban areas using a stacking ensemble technique

ABSTRACT

The reliable and robust monitoring of air temperature distribution is essential for urban thermal environmental analysis. In this study, a stacking ensemble model consisting of multi-linear regression (MLR), support vector regression (SVR), and random forest (RF) optimized by the SVR is proposed to interpolate the daily maximum air temperature (T_max) during summertime in a mega urban area. A total of 10 geographic variables, including the clear-sky averaged land surface temperature and the normalized difference vegetation index, were used as input variables. The stacking model was compared to Cokriging, three individual data-driven methods, and a simple average ensemble model, all through leave-one-station-out cross validation. The stacking model showed the best performance by improving the generalizability of the individual models and mitigating the sensitivity to the extreme daily T_max. This study demonstrates that the stacking ensemble method can improve the accuracy of spatial interpolation of environmental variables in various research fields.     

A SPATIAL ANALYSIS OF AGRICULTURAL INTENSITY IN A BASOTHO VILLAGE OF SOUTHERN AFRICA

The effect of distance upon agricultural production should be as apparent among primitive subsistence agriculturalists as in market-oriented societies. This analysis of yield variations among the Basotho of southern Africa attempts to determine the frictional effect of distance. The results show that distance does not explain variations in yield even when other physical factors are held constant. Perhaps such variables as household characteristics, not included in the analysis, would provide a better explanation.     

Chronology of late wisconsinan glaciation in middle North America

We propose a chronology of late Wisconsinan glacial fluctuations in middle North America, from Alberta to Wisconsin, based on radiocarbon dates derived solely from wood. Previous chronologies of the southwestern margin of the North American Continental Ice Sheet have depended to a considerable degree on radiocarbon dates from fine-grained organic sediment. This material is commonly contaminated with older carbon, resulting in chronologic confusion. By using only dates from wood, much of the confusion disappears. However, because of the scarcity of wood dates, only four of the sixteen identified fluctuations are accurately dated: an advance into Iowa about 14,000 to 13,500 BP, an advance into South Dakota and Iowa about 12,300 BP, an advance into the Lake Michigan basin about 11,700 BP, and an advance into the Lake Superior basin about 9900 BP. In addition, the beginning of late Wisconsinan glaciation, before 20,000 BP, is fairly well documented. None of the fluctuations in the western part of the region are accurately dated.     

Horizontal and vertical distribution of PCBs and chlorinated pesticides in sediments from Masan Bay,Korea

Horizontal and vertical distributions of organochlorine compounds (OCs) were determined in sediments from Masan Bay. The concentrations of polychlorinated biphenyls (PCBs), dichlorodiphenyltrichloroethanes (DDTs), HCB, hexachlorocyclohexanes (HCHs) and chlordane related compounds (CHLs) in sediments were in the range of 1.24-41.4, 0.28-89.2, 0.02-0.59, nd-1.03, and nd-2.56 ng/g, respectively. The spatial distribution of OCs showed a negative gradient from the inner of the bay to outer part of the bay, indicating that the source of OCs was probably located inside the bay. Compositional pattern of PCB congeners showed a relatively high concentration of high-chlorinated congeners in the inner part of the bay and a relatively low concentration of low-chlorinated congeners in the outer part. In sediment core from Masan Bay maximum concentrations of PCBs and DDTs are observed in the subsurface samples and correspond to an age of early 1980s and late 1960s. The concentration profiles of PCBs and DDTs in sediments of Masan Bay appear to correspond to use of PCBs and DDTs in Korea.     

Hydrogeochemical characterization of major factors affecting the quality of shallow groundwater in the coastal area at Kimje in South Korea

This study has evaluated the main hydrogeochemical characteristics that affect the quality of shallow groundwater in the coastal area at Kimje in South Korea. In this area, the chemical composition of groundwater is mostly classified into Na⁺?Cl^? and Ca²⁺+Mg²⁺?NO₃^-+Cl^? types. These types are affected by three major factors: seawater intrusion, fertilizers and redox processes. The Na⁺?Cl^? type, which is generally localized in the coastal area, shows typical characteristics of groundwater affected not only by modern seawater but also by old seawater residing in the reclaimed land. In contrast, the Ca²⁺+Mg²⁺?NO₃^-+Cl^? type, which is usually found inland, is very contaminated by fertilizers. As groundwater flows from the inland area to the coast, the redox condition changes from an oxic condition to a suboxic/anoxic condition. In addition, the reclamation has caused a large amount of underlying organic matter. These effects provide optimum conditions for the occurrence of redox processes in the coastal area.     

Evaluation of remedial alternatives for a petroleum contaminated unconfined aquifer with fluctuating groundwater level

This study evaluated remedial alternatives for a petroleum-contaminated site where an unconfined aquifer composed of a sandy layer of about 3–3.5 m thickness is covered by alluvial deposits and reclaimed soil of about 1.5 m thickness. Precambrian gneiss, of low permeability, lies below the sandy layer. The shallow water table is about 3 m below the surface, but shows high fluctuations of up to 1.5 m in response to precipitation events. The unsaturated soil near the water table and the groundwater are highly contaminated with petroleum hydrocarbons, especially toluene, which have apparently leaked from storage tanks. Selection of the remedial alternatives required consideration of the relevant laws on soil and groundwater conservation in Korea, the results of risk analysis and the hydrogeological conditions. The contaminated area has been divided into zones in which different remediation goals are set based on risk analysis and the degree of natural attenuation. It is estimated that the clean-up goal can be achieved in two years by the combined use of a trench drain and well point pumping to collect the contaminated groundwater for treatment, and a dual air injection system for the contaminated soil.     

Climate Change and Winter Wheat Management: A Modelling Scenario for South-Eastern England

Crop models are useful tools for assessing the impact of climate change on crop production. The dynamic crop-growth model, CERES-Wheat is used to examine crop management responses, including yield, under six climate change scenarios for the years 2025 and 2050 on the Estate of Imperial College at Wye, Kent, U.K. Sensitivity analysis shows a dry matter yield decrease in response to increases in temperature alone. CERES-Wheat was then constrained to assess the crop performance under water-limited production scenarios with different soils, and the results show that crop grain yield actually increases, largely due to CO₂ fertilisation leading to increased rates of photosynthesis. Different management practices (planting dates and nitrogen application) were applied to find the best adaptation strategies. In general, `early' sowing (10th September) had the highest simulated yield, and `late' sowing (10th November) the lowest. For the soils tested, the highest and sustained crop production was obtained from Hamble soils (silt loam) compared with either the Fyfield (sandy) or Denchworth (clay). Adding nitrogen and other fertilisers would likely be necessary to take full advantage of the CO₂ fertilisationeffect and to compensate, in some cases, for yield losses caused by climate change where water shortage becomes serious.     

Biomonitoring of coastal pollution status using protozoan communities with a modified PFU method

Structural and functional parameters of protozoan communities were assessed as indicators of water quality in Korean coastal waters in the summer of 2000. A modified polyurethane foam unit (PFU) method, named the bottled PFU (BPFU) system, was used in order to carry out the bioassessment. Both parameters suggested that biomonitoring using the BPFU system was more effective than the conventional PFU method in offshore areas. The species number collected by the BPFU system generally decreased as pollution intensity increased at three main stations and was greater than that collected using the PFU method (paired t-test, t = 4.83, p < 0.0001). The Margalef diversity index coincided well with the water conditions. The diversity index values calculated from the BPFU system were also significantly higher than those from the PFU method (paired t-test, t = 5.37, p < 0.0001). Furthermore, the functional parameters, i.e. S(eq),G and T90%, correlated with the pollution status and could thus clearly discriminate the different classes of water quality.