Stream salinity prediction in data-scarce regions: Application of transfer learning and uncertainty quantification.

Scarcity of stream salinity data poses a challenge to understanding salinity dynamics and its implications for water supply management in water-scarce salt-prone regions around the world. This paper introduces a framework for generating continuous daily stream salinity estimates using instance-based transfer learning (TL) and assessing the reliability of the synthetic salinity data through uncertainty quantification via prediction intervals (PIs). The framework was developed using two temporally distinct specific conductance (SC) datasets from the Upper Red River Basin (URRB) located in southwestern Oklahoma and Texas Panhandle, United States. The instance-based TL approach was implemented by calibrating Feedforward Neural Networks (FFNNs) on a source SC dataset of around 1200 instantaneous grab samples collected by United States Geological Survey (USGS) from 1959 to 1993. The trained FFNNs were subsequently tested on a target dataset (1998-present) of 220 instantaneous grab samples collected by the Oklahoma Water Resources Board (OWRB). The framework's generalizability was assessed in the data-rich Bird Creek watershed in Oklahoma by manipulating continuous SC data to simulate data-scarce conditions for training the models and using the complete Bird Creek dataset for model evaluation. The Lower Upper Bound Estimation (LUBE) method was used with FFNNs to estimate PIs for uncertainty quantification. Autoregressive SC prediction methods via FFNN were found to be reliable with Nash Sutcliffe Efficiency (NSE) values of 0.65 and 0.45 on in-sample and out-of-sample test data, respectively. The same modeling scenario resulted in an NSE of 0.54 for the Bird Creek data using a similar missing data ratio, whereas a higher ratio of observed data increased the accuracy (NSE = 0.84). The relatively narrow estimated PIs for the North Fork Red River in the URRB indicated satisfactory stream salinity predictions, showing an average width equivalent to 25 % of the observed range and a confidence level of 70 %.

Simulating the effects of low impact development approaches on urban flooding: a case study from Tehran, Iran.

Low impact development (LID) methods have been shown to be efficient in reducing the peak flow and total volume of urban stormwater, which is a top priority for effective urban stormwater management in many municipalities. However, decision-makers need information on the effects of LIDs and their associated costs before allocating limited resources. In this study, the Storm Water Management Model (SWMM) was used to investigate the effects of five different LID scenarios on urban flooding in a district in Tehran, Iran. The LID scenarios included rain barrel (RB) at two sizes, bio-retention cell (BRC), and combinations of the two structures. The results showed that significant node flooding and overflow volume would occur in the study area under the existing conditions, especially for rainfall events with longer return periods. BRC and combinations of BRC and RBs were the most effective options in reducing flooding, while the smaller-size RB was the cheapest alternative. However, normalized cost, obtained through dividing the total cost by the percent reduction in node flooding and/or overflow volume, was smallest for BRC. The results of this study demonstrate how hydraulic modeling can be combined with economic analysis to identify the most efficient and affordable LID practices for urban areas.

Performance Assessment of Five Different Soil Moisture Sensors under Irrigated Field Conditions in Oklahoma.

Meeting the ever-increasing global food, feed, and fiber demands while conserving the quantity and quality of limited agricultural water resources and maintaining the sustainability of irrigated agriculture requires optimizing irrigation management using advanced technologies such as soil moisture sensors. In this study, the performance of five different soil moisture sensors was evaluated for their accuracy in two irrigated cropping systems, one each in central and southwest Oklahoma, with variable levels of soil salinity and clay content. With factory calibrations, three of the sensors had sufficient accuracies at the site with lower levels of salinity and clay, while none of them performed satisfactorily at the site with higher levels of salinity and clay. The study also investigated the performance of different approaches (laboratory, sensor-based, and the Rosetta model) to determine soil moisture thresholds required for irrigation scheduling, i.e., field capacity (FC) and wilting point (WP). The estimated FC and WP by the Rosetta model were closest to the laboratory-measured data using undisturbed soil cores, regardless of the type and number of input parameters used in the Rosetta model. The sensor-based method of ranking the readings resulted in overestimation of FC and WP. Finally, soil moisture depletion, a critical parameter in effective irrigation scheduling, was calculated by combining sensor readings and FC estimates. Ranking-based FC resulted in overestimation of soil moisture depletion, even for accurate sensors at the site with lower levels of salinity and clay.