An online air pollution forecasting system using neural networks

In this work, an online air pollution forecasting system for Greater Istanbul Area is developed. The system predicts three air pollution indicator (SO(2), PM(10) and CO) levels for the next three days (+1, +2, and +3 days) using neural networks. AirPolTool, a user-friendly website (http://airpol.fatih.edu.tr), publishes +1, +2, and +3 days predictions of air pollutants updated twice a day. Experiments presented in this paper show that quite accurate predictions of air pollutant indicator levels are possible with a simple neural network. It is shown that further optimizations of the model can be achieved using different input parameters and different experimental setups. Firstly, +1, +2, and +3 days' pollution levels are predicted independently using same training data, then +2 and +3 days are predicted cumulatively using previously days predicted values. Better prediction results are obtained in the cumulative method. Secondly, the size of training data base used in the model is optimized. The best modeling performance with minimum error rate is achieved using 3-15 past days in the training data set. Finally, the effect of the day of week as an input parameter is investigated. Better forecasts with higher accuracy are observed using the day of week as an input parameter.     

Non-timber forest product dependence among the <Emphasis Type="Italic">Jah Hut</Emphasis> subgroup of Peninsular Malaysia’s Orang Asli

Non-timber forest products (NTFP) represent key sources of cash and subsistence income for millions of rural and indigenous peoples living in tropical developing countries throughout the world. The current study investigates the use and significance of NTFP within a sample of Peninsular Malaysia’s Orang Asli (indigenous people). Data collected via household surveys across three sampling phases reveals that more than 75% of the population is actively engaged in NTFP collection. Household responses indicate diversity in both the types and uses of products collected. NTFP collection participation, frequency of collection, and collection reliance are found to be significantly negatively related to village proximity to the market, as well as to income level relative to the Malaysian poverty line. When collection variables are examined by different product categories, relationships with market access and income group are variable. Implications for different approaches to forest conservation and rural development are discussed.     

Vertical movements, behavior, and habitat of bigeye tuna (Thunnus obesus) in the equatorial eastern Pacific Ocean, ascertained from archival tag data

The results presented in this report are based on analyses of 16,721 days of data downloaded from 96 archival tags recovered from bigeye tuna (Thunnus obesus; 54–159 cm in length, 0.97–5.44 years of age) at liberty from 31 to 1,508 days in the equatorial eastern Pacific Ocean. Analyses of daily timed depth and temperature records resulted in the classification of the data into three daily behavior types: characteristic, associative (associated with floating objects), and other. There is a significant positive correlation between the proportion of time fish exhibit characteristic behavior and increasing length, and significant negative correlations between the proportion of time bigeye exhibit associative and other behavior with increasing length. For the smallest (54–80 cm) to largest (100–159 cm) length classes, the vertical habitats utilized when exhibiting non-associative behaviors were 99 and 98% of the time above the thermocline depth (60 m) during the night, at the same average depth of 34 m, and 60 and 72% of the time below the thermocline during the day at average depths of 163 and 183 m, respectively. For the same smallest to largest length classes, when exhibiting associative behavior, the average nighttime and daytime depths were 25 and 21, and 33 and 37 m, respectively. The apparent effects of the environment on the behavior of the fish are discussed.     

Spatial Variability of Pool-Tail Fines in Mountain Gravel-Bed Stream Affects Grid-Count Results1

Bunte, Kristin, John P. Potyondy, Kurt W. Swingle, and Steven R. Abt, 2012. Spatial Variability of Pool-Tail Fines in Mountain Gravel-Bed Stream Affects Grid-Count Results. Journal of the American Water Resources Association (JAWRA) 48(3): 530-545. DOI: 10.1111/j.1752-1688.2011.00629.x Abstract: Fine sediment (<2 and <6 mm) particles underlying a 49-intersection grid placed on a streambed at 25, 50, and 75% of the wetted pool-tail width are commonly counted to assess the status and trend of aquatic ecosystems or to monitor changes in the supply of fines in mountain gravel-bed streams. However, results vary even when crews perform nearly identical procedures. This study hypothesized that spatial variability of pool-tail fines affects grid-count results and that a sampling scheme can be optimized for precision and accuracy. Grid counts taken at seven evenly spaced locations across the wetted width of 10 pool tails in a pool-riffle study stream indicated a bankward fining trend with secondary peaks of fines within the stream center. Sampling locations close to the waterlines harbored more than twice as many fines as central locations. Most of the five grid-count schemes derived from the seven sampled locations produced significantly different results. Compared with sampling at all seven locations, schemes that focus near waterlines overpredicted fines, while those that focus on the center underpredicted them. Variability of fines among pool tails was the highest within a broad band along the waterlines; hence, focusing sampling there yielded the most variable results. The scheme sampling at 25, 50, and 75% of the wetted width had the lowest precision and moderate accuracy. Accuracy and precision of grid-count results can be greatly improved by sampling at seven even-spaced locations across the pool tail.