Satellite-based ensemble intelligent approach for predicting forest fire: a case of the Hyrcanian forest in Iran.

A machine learning-based approach is applied to simulate and forecast forest fires in the Golestan province in Iran. A dataset for no-fire, medium confidence (MC) fire events, and high confidence (HC) fire events is constructed from MODIS-MOD14A2. Nine climate variables from NASA's FLDAS are used as input variables, and 12 dates and 915 study points are considered. Three machine learning ensemble multi-label classifiers, gradient boosting (GBC), random forest (RFC), and extremely randomized tree (ETC), are used for forest fire simulation for the period 2000 to 2021, and ETC is found to be the most accurate classifier. Future fire projection for the near-future period of 2030 to 2050 is carried out with the ETC model, using CMIP6 EC-Earth3-SSP245 General Circulation Model (GCM) data. It is projected that MC forest fire occurrences will decrease, while HC forest fire occurrences will increase, and that the summer months, especially September, will be the most affected by fire.

Developing a generic relation for predicting sediment pick-up rate using symbolic soft computing techniques.

Sediment pick-up rate has been investigated using experimental and numerical approaches. However, the use of soft computing methods for its prediction has received less attention so far. In this study, genetic programming (GP), grammatical evolution (GE), and gradient boosting machine (GBM) algorithms are employed to develop a relation in dimensionless form for predicting sediment pick-up rate in open channel flow based on two experimental datasets. Dimensionless Froude number, particle diameter, and depth-averaged turbulent kinetic energy are input variables for prediction. Prediction performance is evaluated with performance indices (root mean square error, mean absolute error, and coefficient of correlation), visual comparisons (scatter, dot, and Bland-Altman plots), and uncertainty indicators (Tsallis and Renyi entropies). Three mathematical expressions for sediment pick-up rate prediction are obtained, with GE producing the most accurate results.

A robust stochastic approach in correcting the TRMM precipitation product and simulating flood features.

Logical management and decision-making on water resources require reliable weather variables, where precipitation is considered the main weather variable. Accurate estimation of precipitation is the most important topic in hydrological studies. Due to the lack of a dense network and low temporal and spatial resolution levels at ground-level rain gauges, especially in developing countries, remote sensing methods have been used widely. In recent years, a combination of satellite-ground data on precipitation has led to a more accurate insight into precipitation and improved hydrological model performance. In this study, the Kosar Dam Basin in the Khuzestan province of Iran is selected as the research zone. The TRMM satellite data is used on 50 events to analyze the satellite precipitation data. Copula theory is then employed to check the uncertainties of precipitation estimation, and new precipitations are generated through original data and bias errors. A comparison of the results of the improved TRMM, which was bias-corrected by Gaussian copula, and ground-based rainfall demonstrated the efficacy of this method, with nearly 104% and 51% improvement in the CC and RMSE performance indicators, respectively. The HEC-HMS model was used to simulate flood features based on copula-corrected precipitation over different quartiles (10%, 30%, 50%, 70%, and 90%) and rainfall duration (3, 6, 9, and 24 h). The obtained R-factor values show that the associated uncertainty decreases with rainfall duration, down to 46 and 20% for discharge peak and volume, respectively. In general, the copula approach is a robust approach to improve the accuracy of the TRMM precipitation product for simulating hydrological processes.

Application of ensemble machine learning model in downscaling and projecting climate variables over different climate regions in Iran.

This study evaluates the future climate fluctuations in Iran's eight major climate regions (G1-G8). Synoptic data for the period 1995-2014 was used as the reference for downscaling and estimation of possible alternation of precipitation, maximum and minimum temperature in three future periods, near future (2020-2040), middle future (2040-2060), and far future (2060-2080) for two shared socioeconomic pathways (SSP) scenarios, SSP119 and SSP245. The Gradient Boosting Regression Tree (GBRT) ensemble algorithm has been utilized to implement the downscaling model. Pearson's correlation coefficient (CC) was used to assess the ability of CMIP6 global climate models (GCMs) in replicating observed precipitation and temperature in different climate zones for the based period (1995-2014) to select the most suitable GCM for Iran. The suitability of 21 meteorological variables was evaluated to select the best combination of inputs to develop the GBRT downscaling model. The results revealed GFDL-ESM4 as the most suitable GCM for replicating the synoptic climate of Iran for the base period. Two variables, namely sea surface temperature (ts) and air temperature (tas), are the most suitable variable for developing a downscaling model for precipitation, while ts, tas, and geopotential height (zg) for maximum temperature, and tas, zg, and sea level pressure (psl) for minimum temperature. The GBRT showed significant improvement in downscaling GCM simulation compared to support vector regression, previously found as most suitable for the downscaling climate in Iran. The projected precipitation revealed the highest increase in arid and semi-arid regions (G1) by an average of 144%, while a declination in the margins of the Caspian Sea (G8) by -74%. The projected maximum temperature showed an increase up to +8°C in highland climate regions. The minimum temperature revealed an increase up to +4°C in the Zagros mountains and decreased by -4°C in different climate zones. The results indicate the potential of the GBRT ensemble machine learning model for reliable downscaling of CMIP6 GCMs for better projections of climate.

An efficient strategy for predicting river dissolved oxygen concentration: application of deep recurrent neural network model.

Dissolved oxygen (DO) concentration in water is one of the key parameters for assessing river water quality. Artificial intelligence (AI) methods have previously proved to be accurate tools for DO concentration prediction. This study presents the implementation of a deep learning approach applied to a recurrent neural network (RNN) algorithm. The proposed deep recurrent neural network (DRNN) model is compared with support vector machine (SVM) and artificial neural network (ANN) models, formerly shown to be robust AI algorithms. The Fanno Creek in Oregon (USA) is selected as a case study and daily values of water temperature, specific conductance, streamflow discharge, pH, and DO concentration are used as input variables to predict DO concentration for three different lead times ("t + 1," "t + 3," and "t + 7"). Based on Pearson's correlation coefficient, several input variable combinations are formed and used for prediction. The model prediction performance is evaluated using various indices such as correlation coefficient, Nash-Sutcliffe efficiency, root mean square error, and mean absolute error. The results identify the DRNN model ([Formula: see text]) as the most accurate among the three models considered, highlighting the potential of deep learning approaches for water quality parameter prediction.

Estimating the transient storage parameters for pollution modeling in small streams: a comparison of newly developed hybrid optimization algorithms.

The transient storage model (TSM) is a common approach to assess solute transport and pollution modeling in rivers. Several formulas have been developed to estimate TSM parameters. This study develops a new hybrid optimization algorithm consisting of the dragonfly algorithm and simulated annealing (DA-SA) algorithms. This robust method provides accurate formulas for estimating TSM parameters (e.g., kf, T, [Formula: see text]). A dataset gathered by previous scholars from several rivers in the USA was used to assess the proposed formulas based on several error metrics ([Formula: see text] and [Formula: see text]) and visual indicators. According to the results, DA-SA-based formulas adequately estimated the [Formula: see text] ([Formula: see text], [Formula: see text]), [Formula: see text] ([Formula: see text] [Formula: see text]), and [Formula: see text] ([Formula: see text] [Formula: see text]) parameters. Moreover, the DA-SA-1 showed higher accuracy by improving the RMSE and MAE by 98% compared to the DA and DA-SA-1 as alternatives. The formulas developed in this study significantly outperformed the results of previously proposed models by enhancing the NSE up to 70%. The hybrid DA-SA algorithm method proved highly reliable models to estimate the TSM parameters in the water pollution routing problem, which is vital for reactive solute uptake in advective and transient storage zones of stream ecosystems.