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.     

Detecting patterns of climate change in long-term forecasts of marine environmental parameters

ABSTRACT

Forecasting environmental parameters in the distant future requires complex modelling and large computational resources. Due to the sensitivity and complexity of forecast models, long-term parameter forecasts (e.g. up to 2100) are uncommon and only produced by a few organisations, in heterogeneous formats and based on different assumptions of greenhouse gases emissions. However, data mining techniques can be used to coerce the data to a uniform time and spatial representation, which facilitates their use in many applications. In this paper, streams of big data coming from AquaMaps and NASA collections of 126 long-term forecasts of nine types of environmental parameters are processed through a cloud computing platform in order to (i) standardise and harmonise the data representations, (ii) produce intermediate scenarios and new informative parameters, and (iii) align all sets on a common time and spatial resolution. Time series cross-correlation applied to these aligned datasets reveals patterns of climate change and similarities between parameter trends in 10 marine areas. Our results highlight that (i) the Mediterranean Sea may have a standalone ‘response’ to climate change with respect to other areas, (ii) the Poles are most representative of global forecasted change, and (iii) the trends are generally alarming for most oceans.     

The 3-hour-interval prediction of ground-level temperature in South Korea using dynamic linear models

The 3-hour-interval prediction of ground-level temperature from 00 h out to 45 h in South Korea(38 stations) is performed using the DLM (dynamic linear model) in order to eliminate the systematic error of numerical model forecasts. Numerical model forecasts and observations are used as input values of the DLM. According to the comparison of the DLM forecasts to the KFM (Kalman filter model) forecasts with RMSE and bias, the DLM is useful to improve the accuracy of prediction.     

Meteorological monitoring of an active volcano: Implications for eruption prediction

Rainfall data collected on and around the Soufriere Hills Volcano, Montserrat between 1998 and 2003 were analysed to assess the impact on primary volcanic activity, defined here as pyroclastic flows, dome collapses, and explosions. Fifteen such rainfall-triggered events were identified. If greater than 20 mm of rain fell on a particular day, the probability of a dome collapse occurring on that day increased by a factor of 6.3% to 9.2%, compared to a randomly chosen day. Similarly, the probability of observing pyroclastic flows and explosions on a day with > 20 mm of rainfall increased by factors of 2.6 and 5.4, respectively. These statistically significant links increased as the rainfall threshold was increased. Seventy percent of these rainfall-induced dome collapse episodes occurred on the same calendar day (most within a few hours) as the onset of intense rainfall, but an extra 3 occurred one or two calendar days later. The state of the volcano was important, with the rainfall–volcanic activity link being strongest during periods of unstable dome growth and weakest during periods of no dome growth or after a recent major collapse.Over 50% of the heavy rain days were associated with large-scale weather systems that can potentially be forecast up to a few days ahead. However, the remaining heavy rain days were associated with small-scale, essentially unpredictable weather systems. There was significant variability in the amount of rainfall recorded by different rain gauges, reflecting topographic variations around the volcano but also the inherent small-scale variability within an individual weather system. Hence, any monitoring/warning program is recommended to use a network, rather than just a single gauge. The seasonal cycle in rainfall was pronounced, with nearly all the heavy rain days occurring in the May–December wet season. Hence, the dome was at its most vulnerable at the beginning of the wet season after a period of uninterrupted growth. Interannual variability in rainfall was related to tropical Pacific and Atlantic sea surface temperature anomalies, and holds out the prospect of some limited skill in volcanic hazard forecasts at even longer lead times.     

潞西盆地浅层地下水脆弱性预测分析

运用DRASTIC方法，评价潞西盆地浅层地下水脆弱性，结合区内人类工程活动因子。对潞西盆地远期地下水脆弱性进行预测分析，成果可为区内土地规划、备用水源地的选择、垃圾填埋场的选址及实现水资源可持续利用提供合理依据。     

基于改进的Elman神经网络的中长期径流预报

径流中长期预报长期以来一直都是人们关注的热点研究问题。现行的径流预报方法很多,传统的有时间序列法,多元回归分析法等,这些方法虽然简单易用,但是如果预报对象提供的样本容量偏小或者因子选择不够合理,都会造成预报精度偏差过大,难于有效的指导工程应用。鉴于此,本文提出一种改进的采用局部回归的Elman神经网络方法。并应用到凤滩水库优化调度的径流预报中。结果表明,与回归分析法、BP网络相比较,该方法不仅提高了算法的效率,而且提高了预报的精度,在径流预报中具有有效性和优越性。     

洪水预报信息利用问题研究与讨论

分析了洪水预报研究现状,讨论了实时洪水预报系统研究中信息利用方面存在的问题,提出了值得进一步研究的关键性问题、研究思路和研究技术路线。     

南海热带气旋路径集合预报试验

利用3种不同模式的初始资料，通过它们生成得到16个不同的初始场，分别对2004年南海及其周边地区9个热带气旋个例进行集合预报试验，最后筛选得到了7个南海热带气旋初值集合成员，由此初步探讨了南海热带气旋初值集合成员的生成方法。结果表明，采用不同模式的初始资料生成得到初值集合成员的方法用于集合预报，对南海热带气旋路径预报有一些明显的改进。     

APPLICATION OF BDA SCHEME IN TYPHOON TRACK PREDICTION

The MM5, which is the PSU/NCAR mesoscale nonhydrostatic limited-area model, and its adjoining modeling system are used in this paper. Taking T106 analysis data as background field the authors generate an optimal initial condition of a typhoon by using two bogus data assimilation schemes, and conduct some numerical simulating experiments. The results of No.9608 typhoon (Gloria) show that the optimal initial field have some dramatic improvements, such as inaccurate position of typhoon center, weaker typhoon circulation and incomplete inner structure of the typhoon, which are caused by shortage of data over the sea. Some improvements have been made in the track forecast. Through several comparing experiments, the initialization optimized by BDA scheme is found to be more reasonable than GFDL scheme and its typhoon track forecast is better.     

Application of ensemble kalman filter to geophysical parameters retrieval in remote sensing:A case study of kernel-driven BRDF model inversion

Remote sensing can provide multi-spatial resolution, multi-temporal resolution multi-spectral band and multi-angular data for the observation of land surface. At present, one of research focuses is how to make the best of these data to retrieve geophysical parameters in conjunction with their a priori knowledge and simul-taneously consider the influence of data uncertainties on inversion results[1-5]. The essence of remote sensing lies in inversion. It is difficult to precisely retrieve parame…