Evapotranspiration (ET) is widely employed to measure amounts of total water loss between land and atmosphere due to its major contribution to water balance on both regional and global scales. Considering challenges to quantifying nonlinear ET processes, machine learning (ML) techniques have been increasingly utilized to estimate ET due to their powerful advantage of capturing complex nonlinear structures and characteristics. However, limited studies have been conducted in subhumid climates to simulate local and spatial ETo using common ML methods. The current study aims to present a methodology that exempts local data in ETo simulation. The present study, therefore, seeks to estimate and compare reference ET (ETo) using four common ML methods with local and spatial approaches based on continuous 17-year daily climate data from six weather stations across the Red River Valley with subhumid climate. The four ML models have included Gene Expression Programming (GEP), Support Vector Machine (SVM), Multiple Linear Regression (LR), and Random Forest (RF) with three input combinations of maximum and minimum air temperature-based (Tmax, Tmin), mass transfer-based (Tmax, Tmin, U: wind speed), and radiation-based (Rs: solar radiation, Tmax, Tmin) measurements. The estimates yielded by the four ML models were compared against each other by considering spatial and local approaches and four statistical indicators; namely, the root means square error (RMSE), the mean absolute error (MAE), correlation coefficient (r2), and scatter index (SI), which were used to assess the ML model’s performance. The comparison between combinations showed the lowest SI and RMSE values for the RF model with the radiation-based combination. Furthermore, the RF model showed the best performance for all combinations among the four defined models either spatially or locally. In general, the LR, GEP, and SVM models were improved when a local approach was used. The results showed the best performance for the radiation-based combination and the RF model with higher accuracy for all stations either locally or spatially, and the spatial SVM and GEP illustrated the lowest performance among the models and approaches.
The most efficient way of soybean (Glycine max (L.) Merrill) iron deficiency chlorosis (IDC) management is to select a tolerant cultivar suitable for the specific growing condition. These cultivars are selected by field experts based on IDC visual ratings. However, this visual rating method is laborious, expensive, time-consuming, subjective, and impractical on larger scales. Therefore, a modern digital image-based method using tree-based machine learning classifier models for rating soybean IDC at plot-scale was developed. Data were collected from soybean IDC cultivar trial plots. Images were processed with MATLAB and corrected for light intensity by using a standard color board in the image. The three machine learning models used in this study were decision tree (DT), random forest (RF), and adaptive boosting (AdaBoost). Calculated indices from images, such as dark green color index (DGCI), canopy size, and pixel counts into DGCI ranges and IDC visual scoring were used as input and target variables to train these models. Metrics such as precision, recall, and f1-score were used to assess the performance of the classifier models. Among all three models, AdaBoost had the best performance (average f1-score = 0.75) followed by RF and DT the least. Therefore, a ready-to-use methodology of image processing with AdaBoost model for soybean IDC rating was recommended. The developed method can be easily adapted to smartphone applications or scaled-up using images from aerial platforms.
The ASCE-EWRI reference evapotranspiration (ETo) equation is recommended as a standardized method for reference crop ETo estimation. However, various climate data as input variables to the standardized ETo method are considered limiting factors in most cases and restrict the ETo estimation. This paper assessed the potential of different machine learning (ML) models for ETo estimation using limited meteorological data. The ML models used to estimate daily ETo included Gene Expression Programming (GEP), Support Vector Machine (SVM), Multiple Linear Regression (LR), and Random Forest (RF). Three input combinations of daily maximum and minimum temperature (Tmax and Tmin), wind speed (W) with Tmax and Tmin, and solar radiation (Rs) with Tmax and Tmin were considered using meteorological data during 2003-2016 from six weather stations in the Red River Valley. To understand the performance of the applied models with the various combinations, station, and yearly based tests were assessed with local and spatial approaches. Considering the local and spatial approaches analysis, the LR and RF models illustrated the lowest rate of improvement compared to GEP and SVM. The spatial RF and SVM approaches showed the lowest and highest values of the scatter index as 0.333 and 0.457, respectively. As a result, the radiation-based combination and the RF model showed the best performance with higher accuracy for all stations either locally or spatially, and the spatial SVM and GEP illustrated the lowest performance among models and approaches.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.