Mean Absolute Bias Error Research Articles

Forecasting soil temperature at multiple-depth with a hybrid artificial neural network model coupled-hybrid firefly optimizer algorithm

Forecasting soil temperature at multiple depths is considered to be a core decision-making task for examining future changes in surface and sub-surface meteorological processes, land–atmosphere energy exchange, resilient agricultural systems for improved crop health and eco-environmental risk assessment. The aim of this paper is to estimate monthly soil temperature (ST) at multiple depth: 5, 10, 20, 50 and 100 cm with a hybrid multi-layer perceptron algorithm integrated with the firefly optimizer algorithm (MLP-FFA). To develop the hybrid MLP-FFA model, the monthly ST and relevant meteorological variables for the city of Adana (Turkey) are collated for the period of 2000–2007. Construction of hybrid MLP-FFA model is drawn upon a limited set of predictors, denoted as soil depth, periodicity (or the respective month), air temperature, pressure and solar radiation, while the objective variable for MLP-FFA model is the forecasted ST at multiple depths. To the evaluate MLP-FFA, statistical metrics applied to test the model’s performance are: the root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and mean bias error (MBE) where the sign of the difference is also considered. In conjunction with statistical metrics, a Taylor diagram is utilized to visualize the degree of similarity between the observed and forecasted soil moisture. In terms of the forecasted results, the hybrid MLP-FFA model is seen to outperform the standalone MLP model. The optimal MLP-FFA is attained for soil temperature forecasting at a depth of 20 cm (RMSE, MAPE of 0.546 °C, 2.40%) whereas the optimal MLP is attained for soil temperature forecasting at a depth of 50 cm (RMSE of 0.544 °C, 2.21%). Conclusively, the study advocates through statistical metrics attained the better utility of the hybrid MLP-FFA hybrid model. Given its superior performance, it is ascertained that the hybrid MLP model integrated with Firefly optimizer is a qualified ancillary tool that can be applied to generate precise soil temperature forecasts at multiple soil depths.

Estimation of the daily global solar radiation based on the Gaussian process regression methodology in the Saharan climate

Accurate estimation of solar radiation is the major concern in renewable energy applications. Over the past few years, a lot of machine learning paradigms have been proposed in order to improve the estimation performances, mostly based on artificial neural networks, fuzzy logic, support vector machine and adaptive neuro-fuzzy inference system. The aim of this work is the prediction of the daily global solar radiation, received on a horizontal surface through the Gaussian process regression (GPR) methodology. A case study of Ghardaia region (Algeria) has been used in order to validate the above methodology. In fact, several combinations have been tested; it was found that, GPR-model based on sunshine duration, minimum air temperature and relative humidity gives the best results in term of mean absolute bias error (MBE), root mean square error (RMSE), relative mean square error (rRMSE), and correlation coefficient (r) . The obtained values of these indicators are 0.67 MJ/m2, 1.15 MJ/m2, 5.2%, and 98.42%, respectively.

Application of the hybrid ANFIS models for long term wind power density prediction with extrapolation capability.

In this paper, the suitability and performance of ANFIS (adaptive neuro-fuzzy inference system), ANFIS-PSO (particle swarm optimization), ANFIS-GA (genetic algorithm) and ANFIS-DE (differential evolution) has been investigated for the prediction of monthly and weekly wind power density (WPD) of four different locations named Mersing, Kuala Terengganu, Pulau Langkawi and Bayan Lepas all in Malaysia. For this aim, standalone ANFIS, ANFIS-PSO, ANFIS-GA and ANFIS-DE prediction algorithm are developed in MATLAB platform. The performance of the proposed hybrid ANFIS models is determined by computing different statistical parameters such as mean absolute bias error (MABE), mean absolute percentage error (MAPE), root mean square error (RMSE) and coefficient of determination (R2). The results obtained from ANFIS-PSO and ANFIS-GA enjoy higher performance and accuracy than other models, and they can be suggested for practical application to predict monthly and weekly mean wind power density. Besides, the capability of the proposed hybrid ANFIS models is examined to predict the wind data for the locations where measured wind data are not available, and the results are compared with the measured wind data from nearby stations.

Estimation and comparison of various global solar radiation models on horizontal surface in a hot and dry region

ABSTRACTIn this study, 54 empirical models including 22 new models were used to estimate the global solar radiation on horizontal surface. The data were collected in Dohuk cities in the northern Iraq. The evaluated models are based on sunshine duration, air temperature, cloud cover, and other meteorological parameters. In the investigated region, the statistical error indicators value for all empirical models of R, R2, mean absolute bias error, root mean square error, mean absolute percentage error, Nash–Sutcliffe equation, and t-test value were low and within an acceptable range. The results of Model 49 based on the other meteorological parameters which is newly developed model with equation of had the most appropriate result with a value of R2 (0.9942) in the 54 empirical models. Besides, 22 new equation models are produced based on sunshine hours, air temperature, and other meteorological parameters. The best accuracy among the new equation was observed in Model 49. Furthermore, the Models 34, 12, and 25 based on cloud cover, sunshine hour, and air temperature in terms of data accuracy were ranked, respectively.

Retrieval of aerosol profiles combining sunphotometer and ceilometer measurements in GRASP code

In this paper we present an approach for the profiling of aerosol microphysical and optical properties combining ceilometer and sun/sky photometer measurements in the GRASP code (General Retrieval of Aerosol and Surface Properties). For this objective, GRASP is used with sun/sky photometer measurements of aerosol optical depth (AOD) and sky radiances, both at four wavelengths and obtained from AErosol RObotic NETwork (AERONET), and ceilometer measurements of range corrected signal (RCS) at 1064 nm. A sensitivity study with synthetic data evidences the capability of the method to retrieve aerosol properties such as size distribution and profiles of volume concentration (VC), especially for coarse particles. Aerosol properties obtained by the mentioned method are compared with airborne in-situ measurements acquired during two flights over Granada (Spain) within the framework of ChArMEx/ADRIMED (Chemistry-Aerosol Mediterranean Experiment/Aerosol Direct Radiative Impact on the regional climate in the MEDiterranean region) 2013 campaign. The retrieved aerosol VC profiles agree well with the airborne measurements, showing a mean bias error (MBE) and a mean absolute bias error (MABE) of 0.3 μm3/cm3 (12%) and 5.8 μm3/cm3 (25%), respectively. The differences between retrieved VC and airborne in-situ measurements are within the uncertainty of GRASP retrievals. In addition, the retrieved VC at 2500 m a.s.l. is shown and compared with in-situ measurements obtained during summer 2016 at a high-atitude mountain station in the framework of the SLOPE I campaign (Sierra Nevada Lidar AerOsol Profiling Experiment). VC from GRASP presents high correlation (r = 0.91) with the in-situ measurements, but overestimates them, MBE and MABE being equal to 23% and 43%.

Spatial Pattern and Temporal Stability of Root-Zone Soil Moisture during Growing Season on a Larch Plantation Hillslope in Northwest China

Soil moisture plays a decisive role for tree growth and forest ecosystems services supply in dryland regions. Hence, it is necessary to clarify the spatio-temporal variation of soil moisture under field conditions. This study selected a hillslope in the Liupan Mountains covered by the plantation of Larix principis-rupprechtii Mayr (larch), a main afforestation tree species in north and northwest China. The volumetric soil moisture (VSM) in root zone layers was monitored with a time interval of about 15 days during the growing season (from May to October) in 2016 at 48 points on this hillslope. The aim was to evaluate the spatial pattern and temporal stability of soil moisture at slope scale. The results showed a moderate spatial variability of VSM in each soil layer, with the variation coefficients range of 17.12–22.63%. The spatial variability of VSM showed a dependence on the soil wetness and a threshold effect, it increased with rising VSM until the VSM reached a threshold of about 15%, but thereafter decreased. The mean relative difference (MRD) among the 48 points ranged from −30.56% to 27.20%, −29.89% to 39.58%, and −28.13% to 33.71% for the soil layers of 0–20, 20–40, and 40–60 cm, respectively. The associated standard deviation (SDRD) (and range) was 11.38% (5.20–26.06%), 8.28% (4.64–15.63%), and 6.51% (2.00–14.16%) for the soil layers of 0–20, 20–40, and 40–60 cm, respectively. The high Spearman’s rank coefficients (p < 0.05) among the measuring dates at each soil layer indicated that the spatial distribution of VSM in the root zone had strong temporal stability. The decrease of Spearman’ rank correlation coefficient and mean SDRD with rising soil depth indicated an increasing temporal stability of VSM with rising soil depth. The mean VSM of the three soil layers on the entire hillslope can be estimated by the direct method (using representative points determined by the index of temporal stability (ITS)) successfully, and these representative points determined by ITS were mainly located at the points with a ratio of field capacity to leaf area index (LAI) close to the slope mean. Moreover, the mean VSM of the three soil layers on the entire hillslope can also be estimated by indirect method (using the time-stable points determined by mean absolute bias error (MABE) and considering the offset between slope mean VSM and observed VSM at time-stable points), and the prediction accuracy of the indirect method was better than the direct method. Significant correlation between MRD and soil bulk density, field capacity, capillary porosity, and LAI were observed for all soil layers, indicating that both the water-retention ability in root zone soil (expressed mainly by field capacity) and water-consumption ability of trees (expressed mainly by canopy LAI) are the main factors controlling the spatial pattern of root-zone VSM on the larch plantation hillslope studied.

Estimation of the Land Surface Temperature over the Tibetan Plateau by Using Chinese FY-2C Geostationary Satellite Data.

During the process of land–atmosphere interaction, one of the essential parameters is the land surface temperature (LST). The LST has high temporal variability, especially in its diurnal cycle, which cannot be acquired by polar-orbiting satellites. Therefore, it is of great practical significance to retrieve LST data using geostationary satellites. According to the data of FengYun 2C (FY-2C) satellite and the measurements from the Enhanced Observing Period (CEOP) of the Asia–Australia Monsoon Project (CAMP) on the Tibetan Plateau (CAMP/Tibet), a regression approach was utilized in this research to optimize the split window algorithm (SWA). The thermal infrared data obtained by the Chinese geostationary satellite FY-2C over the Tibetan Plateau (TP) was used to estimate the hourly LST time series. To decrease the effects of cloud, the 10-day composite hourly LST data were obtained through the approach of maximal value composite (MVC). The derived LST was used to compare with the product of MODIS LST and was also validated by the field observation. The results show that the LST retrieved through the optimized SWA and in situ data has a better consistency (with correlation coefficient (R), mean absolute error (MAE), mean bias (MB), and root mean square error (RMSE) values of 0.987, 1.91 K, 0.83 K and 2.26 K, respectively) than that derived from Becker and Li’s SWA and MODIS LST product, which means that the modified SWA can be applied to achieve plateau-scale LST. The diurnal variation of the LST and the hourly time series of the LST over the Tibetan Plateau were also obtained. The diurnal range of LST was found to be clearly affected by the influence of the thawing and freezing process of soil and the summer monsoon evolution. The comparison between the seasonal and diurnal variations of LST at four typical underlying surfaces over the TP indicate that the variation of LST is closely connected with the underlying surface types as well. The diurnal variation of LST is the smallest at the water (5.12 K), second at the snow and ice (5.45 K), third at the grasslands (19.82 K) and largest at the barren or sparsely vegetated (22.83 K).

New method for estimating daily global solar radiation over sloped topography in China

A new scheme for the estimation of daily global solar radiation over sloped topography in China is developed based on the Iqbal model C and MODIS cloud fraction. The effects of topography are determined using a digital elevation model. The scheme is tested using observations of solar radiation at 98 stations in China, and the results show that the mean absolute bias error is 1.51 MJ m−2 d−1 and the mean relative absolute bias error is 10.57%. Based on calculations using this scheme, the distribution of daily global solar radiation over slopes in China on four days in the middle of each season (15 January, 15 April, 15 July and 15 October 2003) at a spatial resolution of 1 km × 1 km are analyzed. To investigate the effects of topography on global solar radiation, the results determined in four mountains areas (Tianshan, Kunlun Mountains, Qinling, and Nanling) are discussed, and the typical characteristics of solar radiation over sloped surfaces revealed. In general, the new scheme can produce reasonable characteristics of solar radiation distribution at a high spatial resolution in mountain areas, which will be useful in analyses of mountain climate and planning for agricultural production.

Performance evaluation of hybrid adaptive neuro-fuzzy inference system models for predicting monthly global solar radiation

Solar energy plays a vital role in the field of sustainable energy by providing clean, efficient and reliable alternative source of energy. Where, the output of solar energy systems is highly dependent on the solar radiation. Thus, accurate prediction of solar radiation is considered as a very important factor for such applications. In this paper, standalone adaptive neuro-fuzzy inference system and hybrid models have been developed to predict monthly global solar radiation from different meteorological parameters such as sunshine duration S(h), and air temperature. The proposed hybrid models include particle swarm optimization, genetic algorithm and differential evolution. To evaluate the capability and efficiency of the proposed models, several statistical indicators such as; root mean square error, co-efficient of determination and mean absolute bias error are used. All prediction models’ results showed good agreements with measured datasets. The performance evaluation over different statistical indicators showed high correlation for all developed modules. Whereas, hybrid particle swarm optimization has achieved the best statistical indicators over all models in training and testing models. A detailed comparison with other studies is carried out to validate the prediction accuracy and suitability of the proposed models. The results showed that the developed hybrid models have the most reliable and accurate estimation capability and deemed to be the efficient methods for predicting global solar radiation for various applications.

Simple and effective method for evaluating performance of Si based photovoltaic cell technologies

In this research, a methodology is proposed to extract temperature coefficient of open circuit voltage and intensity dependence of short circuit current of PV cells. A holistic combination of experiments and diode models of photovoltaic cells was used. The effects of light intensity and cell temperature on the performance of different solar cell technologies were investigated. Three main photovoltaic cell technologies were investigated, namely monocrystalline, polycrystalline and thin-film. The monocrystalline and polycrystalline technologies were modeled using the one-diode model whereas the thin-film technology was modeled using the two-diode model. The mean absolute bias error was close to zero for the models, which confirmed their reliability. It was observed that light intensity highly affected the current parameters while having little effect on the voltage parameters. It was also noted that cell temperature affected the voltage parameters while having minimal effect on the current parameters. The study revealed that wafer based technology exhibited higher response to irradiance and lower temperature coefficient than thin film technology. Irrespective of cell technology, temperature coefficient is linearly dependent on intensity of light.

MODELING OF MONTHLY OIL PALM YIELD PATTERN BASED ON DEPTH OF RAINFALL AND RAINY DAYS

Oil palm requires evenly distributed rainfall throughout the year to achieve optimum yield. This study was aimed to estimate monthly oil palm yield based on depth of rainfall and rainy days data. Yield data were collected from 12 years old of oil palm grown on mineral soils at 15 plantations in North Sumatra. The yield data were monthly data of 2016 and 2017 for database and comparison, respectively. Data of depth of rainfall and rainy days were from 2012-2016. Data were analysed using linear and non-linear correlation between depth of rainfall versus yield and rainy days versus yield at time lag of 0, 6, 12, 18, 24, 30, 36, 42, and 48 months. The results of correlation analysis were used to construct an equation model for estimating monthly yield patterns. Based on values of Root Mean Square Error (RMSE), MeanAbsolute Bias Error (MABE), and Mean Absolute Percentage Error (MAPE) between estimation and actual monthly yield of 2017, it could be conluded that estimation model based on rainy days were more accurate than when it was based on depth of rainfall. The values of RMSE, MABE, MAPE of estimation model based on rainy days were 0,337; 0,275; 15,482%, respectively; while based on depth of rainfall the values were 0,367; 0,296; 16,594%, respectively.

MODELING OF MONTHLY OIL PALM YIELD PATTERN BASED ON DEPTH OF RAINFALL AND RAINY DAYS

Oil palm requires evenly distributed rainfall throughout the year to achieve optimum yield. This study was aimed to estimate monthly oil palm yield based on depth of rainfall and rainy days data. Yield data were collected from 12 years old of oil palm grown on mineral soils at 15 plantations in North Sumatra. The yield data were monthly data of 2016 and 2017 for database and comparison, respectively. Data of depth of rainfall and rainy days were from 2012-2016. Data were analysed using linear and non-linear correlation between depth of rainfall versus yield and rainy days versus yield at time lag of 0, 6, 12, 18, 24, 30, 36, 42, and 48 months. The results of correlation analysis were used to construct an equation model for estimating monthly yield patterns. Based on values of Root Mean Square Error (RMSE), MeanAbsolute Bias Error (MABE), and Mean Absolute Percentage Error (MAPE) between estimation and actual monthly yield of 2017, it could be conluded that estimation model based on rainy days were more accurate than when it was based on depth of rainfall. The values of RMSE, MABE, MAPE of estimation model based on rainy days were 0,337; 0,275; 15,482%, respectively; while based on depth of rainfall the values were 0,367; 0,296; 16,594%, respectively.

Productivity modelling of a developed inclined stepped solar still system based on actual performance and using a cascaded forward neural network model

This paper presents a cascaded forward neural network model for predicting the productivity of a developed inclined stepped solar still system. The actual recorded data of the developed inclined stepped solar still system is used to develop the proposed model. The results of the predicted productivity are compared with that obtained from regression and linear models. In this study, three statistical error terms are used to evaluate the proposed model: root mean square error (RMSE), mean absolute percentage error (MAPE) and mean bias error (MBE). The results show that the proposedcascaded forward neural network (CFNN) model more accurately predicts the productivity of the system than the other modelsmentioned. The RMSE, MAPE and MBE values of the proposed model are 22.48%, 18.51% and −26.46%, respectively. Therefore, the CFNN model provides benefits for modelling the solar still.

Experimental investigation of TiO2-, SiO2- and Al2O3-lubricants for a domestic refrigerator system using LPG as working fluid

This paper experimentally investigated low concentration of TiO2, SiO2 and Al2O3 nanoparticles dispersed in a mineral oil lubricant, in a domestic refrigerator system, using LPG as the working fluid. The performance of the system was investigated with test parameters including: power consumption, thermal conductivity and viscosity at the suction and discharge end of the compressor, evaporator air temperature (pull-down time) was also considered. Findings showed that all nano-lubricant based LPG had lower steady state evaporator air temperatures than LPG with pure mineral oil at 180min. In addition, TiO2 and SiO2 nano-lubricants gave 13 and 12% reductions in power consumptions than the base refrigerant (LPG), whereas higher power consumption of the refrigerator system was found with utilization of Al2O3-lubricant when compared with base refrigerant (LPG). Futhermore, error analysis conducted using Mean Absolute Bias Error (MAPE) and Root Mean Square Error (RMSE) showed that Al2O3-lubricant gave the best fit with pure-lubricant LPG than TiO2 and SiO2. Application of nanoparticles in refrigeration systems was found to improve the performances of the system, but TiO2 followed by SiO2 will be prefered among the nanoparticles, as Al2O3-lubricant was found to be incompartible with hydrocarbon refrigerant having led to a high power consumption.

Application of extreme learning machine for short term output power forecasting of three grid-connected PV systems

The power output (PO) of a photovoltaic (PV) system is highly variable because of its dependence on solar irradiance and other meteorological factors. Hence, accurate PO forecasting of a grid-connected PV system is essential for grid stability, optimal unit commitment, economic dispatch, market participation and regulations. In this paper, a day ahead and 1 h ahead mean PV output power forecasting model has been developed based on extreme learning machine (ELM) approach. For this purpose, the proposed forecasting model is trained and tested using PO of PV system and other meteorological parameters recorded in three grid-connected PV system installed on a roof-top of PEARL laboratory in University of Malaya, Malaysia. The results obtained from the proposed model are compared with other popular models such as support vector regression (SVR) and artificial neural network (ANN). The performance in terms of accuracy and precision of the prediction models is conducted with standard statistical error indicators including: relative root mean square error (RMSE), mean absolute percentage error (MAPE), mean absolute bias error (MABE) and coefficient of determination (R2). The comparison of results obtained from the proposed ELM model to other models showed that ELM model enjoys higher accuracy and less computational time in forecasting the daily and hourly PV output power.

Comparison of two air temperature gridding methods over complex terrain in China

This paper evaluates the monthly mean air temperature over complex terrain in China determined using two methods: Australian National University spline and GridMet_Climate_Temperature (grid meteorological and climatic temperature (GMCT)) models. It is found that the macroscopic patterns of the temperature distributions by the two methods are very similar. The comparison of modelled temperatures with observations from 40 verification stations shows that both methods have the same value of total mean absolute bias errors (MABE) of 0.43 °C. Since the observational data used are from meteorological observational stations which are located on horizontal flat and open areas, these evaluation results only represent those on the flat open areas. In order to evaluate the effects of topographical factors on the temperature, a concept of relief amplitude is defined. The temperatures are grouped according to range of the relief amplitudes, and variation of temperature with relief amplitudes is worked out. It is found that the temperature difference between the two models varies significantly with relief amplitudes. The averaged maximum difference can be as large as 12 °C in January when the relief amplitude is greater than 700 m. The reasons for the temperature difference are investigated, and they are due to the effects of topographical slopes and aspects. One model (GMCT) includes these effects while the other does not. The evaluation results demonstrate that it is necessary to include the effects of topographical factors in the model simulation in order to produce realistic temperature distributions in the complex terrain areas. The distributions of temperature on the northern and southern slopes are determined using the GMCT model, and the results show that the temperatures on the southern slopes are clearly greater than those of the northern slopes, especially in winter. These results can be used as a guide for the reasonable and sustainable utilization of heat resources in the mountainous area.

Temporal stability of soil moisture in an oasis of northwestern China

AbstractTemporal stability of soil moisture has been widely used in hydrological monitoring since it emerged. However, the spatial analysis of temporal stability at the landscape scale is often limited because of insufficient sampling numbers. This work made an effort to investigate the spatial variations of temporal stability of soil moisture in an oasis landscape. The specific objectives of the study were to explore the spatial patterns of temporal stability and to determine the controlling factors of temporal stability in the desert oasis. A time series of soil moisture measurements were gathered on 23 occasions at 118 locations over 3 years in a rectangular transect of approximately 100 km2. The nonparametric Spearman's rank correlation coefficient, standard deviation of relative difference (SDRD), and mean absolute bias error (MABE) were used to quantify the temporal stability of soil moisture. Results showed that the temporal stability of soil moisture was depth dependent and season dependent. The spatial pattern of soil moisture in a deep soil layer and between two same seasons generally had a high temporal stability. SDRD and MABE were spatially autocorrelated and exhibited strong spatial structures in the geographic space. The concept of temporal stability can be extended to describe the time‐stable areas of soil moisture with geostatistics. There were great differences between SDRD and MABE in describing the temporal stability of soil moisture and in identifying the controlling factors of temporal stability. In this case, MABE was a better alternative to estimate the areal mean soil moisture using representative locations than SDRD. Land use type, soil moisture condition, and soil particle composition were the dominant controls of temporal stability in the oasis. These insights could help to better understand the essence of temporal stability of soil moisture in arid regions.

Two ultraviolet radiation datasets that cover China

Ultraviolet (UV) radiation has significant effects on ecosystems, environments, and human health, as well as atmospheric processes and climate change. Two ultraviolet radiation datasets are described in this paper. One contains hourly observations of UV radiation measured at 40 Chinese Ecosystem Research Network stations from 2005 to 2015. CUV3 broadband radiometers were used to observe the UV radiation, with an accuracy of 5%, which meets the World Meteorology Organization’s measurement standards. The extremum method was used to control the quality of the measured datasets. The other dataset contains daily cumulative UV radiation estimates that were calculated using an all-sky estimation model combined with a hybrid model. The reconstructed daily UV radiation data span from 1961 to 2014. The mean absolute bias error and root-mean-square error are smaller than 30% at most stations, and most of the mean bias error values are negative, which indicates underestimation of the UV radiation intensity. These datasets can improve our basic knowledge of the spatial and temporal variations in UV radiation. Additionally, these datasets can be used in studies of potential ozone formation and atmospheric oxidation, as well as simulations of ecological processes.

Comparison of GOME-2 UVA Satellite Data to Ground-Based Spectroradiometer Measurements at a Subtropical Site

The ultraviolet A (UVA) (315–400 nm) daily exposures and maximum daily irradiances from the Global Ozone Monitoring Experiment (GOME)-2 satellite have been compared over three years to the corresponding data from a ground-based spectroradiometer for a subtropical Southern Hemisphere site. This is one of the first such comparisons for the GOME-2 UVA waveband in the Southern Hemisphere. For the UVA daily exposures and the maximum daily irradiances, the comparisons were undertaken for all sky conditions and for cloud-free conditions. Under cloud-free conditions the $R^{2}$ of the fit regression line for the comparisons was 0.93 for the exposures and the irradiances. The influence of cloud reduced the $R^{2}$ values to 0.86 and 0.70 for the daily exposures and maximum irradiances, respectively. The relative root-mean-square error (rRMSE), mean absolute bias error (MABE), and mean bias error (MBE) for the maximum daily UVA irradiances on the cloud-free days were 0.08, 6.59 ± 7.32%, and −1.04 ± 9.83%, respectively. Similarly, for the daily UVA exposures on the cloud-free days, the rRMSE, MABE, and MBE were 0.10, 5.19 ± 6.42%, and −0.79 ± 8.24%, respectively. For the all-sky conditions, the corresponding values were 0.20, 15.23 ± 14.90%, and −0.79 ± 8.24% for the maximum daily irradiances and 0.19, 14.17 ± 14.56%, and −4.63 ± 20.60% for the daily exposures. In all studies of the influence of UVA on human health, this can complement ground-based measurements that provide the higher temporal and spatial resolution available only at a limited number of surface monitoring sites.

Characterizing and modeling regional-scale variations in soil salinity in the arid oasis of Tarim Basin, China

Soil spatial variations are scale dependent and can be controlled by many environmental factors. Numerous factors have been related to variations of soil salinity, some of them combined empirical mode decomposition (EMD) method and correlation analysis. However, environmental factors that essentially affect soil-water balance are not given enough attention. In addition, further analysis is needed in exploring how well the environmental factors can interpret the variations in soil salinity at different scales, especially in arid oasis areas and at large scales. This paper explores the potential of modeling variations in soil salinity via the EMD and Random Forest modeling of remote sensing based environmental factors. A case study is presented for Tarim basin, Xinjiang, China, using land surface temperature (LST), evapotranspiration (ET), TRMM precipitation (TRM) and digital elevation model (DEM) products. Soil salinity and its decompositions were first correlated with environmental factors for feature selection. Then, those selected environmental factors and their decompositions were correlated and coupled with their counterparts of soil salinity to evaluate their synchronization. Finally, those IMF components of environmental factors that had high correlation coefficients and were coupled well with corresponding IMF components of soil salinity were identified and divided into different feature sets for modeling. Mean absolute error and mean bias error were adopted for accuracy assessment of the models. Our results indicate that soil salinity series can be separated into eight scales ranging from 170km to 480km. IMF components 5–7 account for most of the variation and can be modeled using the corresponding IMF components of different combinations of DEM, ET, LST and TRM. IMF components 6–7 are well coupled with LST and ET at approximately 475km scale. Overall, regional-scale modeling of variations in soil salinity based on remote sensing products is possible. Reasonably accurate results can be obtained in arid oasis areas where researchers and policy makers must focus on preventing the loss of agricultural productivity and ecosystem health.