Newsletters
Enhancing Leaf Area Index Simulations With Evidential Deep Learning: A Probabilistic Approach to Uncertainty and Sensitivity
Published: July 16, 2025
Summary:
This article presents a significant advancement in LAI simulation through evidential deep learning, offering high accuracy, reduced computational demands, and detailed uncertainty quantification. It identifies key parameters for model improvement and calls for future integration with observational data, making it a cornerstone for enhancing climate and ecological modeling efforts.
Impact of Soil Hydraulic Parameter Variability on Soil Moisture: An Empirical Orthogonal Function Analysis
Published: February 6, 2025
Summary:
This article employs Empirical Orthogonal Function (EOF) analysis to elucidate how variations in soil hydraulic parameters underpin significant differences in soil moisture simulations. The work highlights the uncertainty introduced by different parameterization schemes and calls for improved representations of soil properties in land surface models.
Can Foot-Flush Reduce Cholera Spreading?
Published: December 5, 2024
Summary:
Drawing on my undergraduate thesis, I proposed a foot-operated toilet valve to reduce hand contact with surfaces and curb the transmission of cholera. Inspired by pandemic-era hands-free designs, the mechanism promises affordability and ease of adoption, pending sponsorship for prototyping.
Africa Needs More Investment in Climate Action
Published: December 3, 2024
Summary:
Highlighting a US$52.7 billion annual funding gap for adaptation, Kachinga Silwimba warns that underinvestment leaves African nations vulnerable to droughts, power crises, and other climate-related risks. He urges stronger collaboration between researchers and policymakers to secure the resources needed for effective mitigation and resilience.
Efficient Analysis of Large Climate Datasets
Published: November 29, 2023
Summary:
Kachinga Silwimba combines climate model emulation with EOF analysis on CLM5 outputs to reduce computational costs. Leveraging HPC, Python, and statistical tools, they extract key patterns from massive datasets, making large-scale climate studies faster and more accessible.