The Indian education system, like many countries, focuses strongly on annual national exams. This results in an emphasis on rote learning, discourages student participation and engagement in lessons, and yields a lack of motivation to learn. Prioritization of rote learning is ubiquitous across India, but is especially detrimental for students from adverse backgrounds and resource-poor settings. Parikrma Humanity Foundation is a non-profit school in Bangalore India that serves students from “the poorest of poor” backgrounds. The school integrates a “360-degree” model that prioritizes student happiness, while emphasizing student and family health as central to learning. Despite these well-established values, students at Parikrma are not exempt from national exam requirements and are also subject to the implications of the heightened importance of exam results. We sought to understand whether key stakeholders perceived that exclusive focus on rote learning hinders development of social-emotional life skills. We conducted over 70 interviews with students, teachers, school faculty, and alumni and analyzed these qualitative results through a method of conceptualization. From this we developed a scale to determine the degree to which the absence of explicit instruction of psychosocial skills impacts overall well-being. Respondents overwhelmingly reported that the lack of material to develop psychosocial skills such as teamwork and active listening results in students with less established life skills and limits access to future opportunities. We aim to develop a curriculum that promotes psychosocial skills key to professional success and overall happiness to be integrated within the context of Parikrma by establishing a cross-aged peer mentorship program that encourages students accountability to each other and themselves. We seek to engage teachers and students through the administration of comprehensive surveys that empowers individuals to self- report the impact of the program in order to assess results.
 

Planning for water resources management (WRM) requires the best available predictions of streamflow. We want to provide actionable predictions that improve WRM as climate change alters streamflows. However, the computer models of these systems result in imperfect predictions despite our best efforts to match observations. These systematic errors reduce the usefulness of model outputs. To improve our ability to plan for WRM, we apply statistical correction techniques to model outputs so that they agree better with observations. The Columbia River Basin, a major river basin in the Northwestern United States, is heavily regulated for a large number of competing uses. In this project, we focus on the Yakima river basin in central Washington, a subbasin of the Columbia River, and use it as a case study for evaluating multiple statistical correction techniques. By comparing streamflow observations to simulations for the same periods we can develop statistical corrections. The application of these statistical corrections is often referred to as “post-processing”. Post-processing adjusts predictions based on previous knowledge of the region as well as the historical observations.These post-processing techniques can then be applied at locations and times without observations. As part of this project, we developed a toolkit for the evaluation of different post-processing techniques. We explore which measures and visualization techniques are adequate at describing key aspects of the streamflow simulations. Our toolkit builds on open source technologies that will allow researchers to reliably measure the statistical performance of these post-processing techniques. Evaluation is performed through exploring different statistical metrics and building a suite of summarizing plotting capabilities in a standalone, open source Python package.

Given that greater than 15% of the world’s population currently relies on snow melt for drinking water, hydrological modeling of landscapes subject to routine snowfall is becoming increasingly important. Up to 60% of snowfall in forested terrain is intercepted by the forest canopy. Therefore, tree interception parameters that quantify how much water or snow a tree collects during a precipitation event are critical for understanding temporal and spatial water fluxes, which most notably determine when and in what volume water is delivered to communities. This project is evaluating whether a new video processing technique can be used to reliably and accurately identify snow interception parameters for coniferous trees. In particular, this project seeks to learn whether the resonant frequency of swaying snow-loaded trees, as determined by processing time lapse videos, can be used to compute the mass of snow in a tree. Having shown that video processing can correctly infer the sway frequency of a tree, we are now deploying cameras on Snoqualmie Pass to observe tree sway events concurrent with snow loading events. To date, we have four cameras monitoring a tree and the snow levels around it. Pending further video data collection, we will begin applying the tree sway processing algorithm, which uses FFT processing to compute changes in tree sway frequency and corresponding changes in tree mass. If successful, this method may improve tree interception parameterization and therefore hydrological forecasting.