Comparative Rainfall Analysis of 3D-Printed Automatic Weather Stations with Co-located Kenya Meteorological Department Stations

In many countries of Africa that are prone to frequent food insecurities, the number of reliable and accessible weather stations has been declining from a peak of 3,300 in 1981 to a current level of 800 in 2023. At the same time, climate change related to anthropogenic impacts is progressing and causing the number of extreme weather events such as prolific floods and droughts to increase. There is a great need for both improved density of surface weather observations in these areas as well as observation reliability to more accurately monitor and predict these weather events and their impacts on food security in these regions. The COMET International Capacity Development Program (ICDP) at the University Corporation for Atmospheric Research (UCAR) and the Climate Hazards Center (CHC) at the University of California Santa Barbara have partnered with several African meteorological departments to implement the Enhancing Meteorological Networks Partnership (EMNP), funded by the Famine Early Warning System (FEWS NET) and USAID. The EMNP acknowledges the 3D-Printed Automatic Weather Station (3D-PAWS) as an affordable and sustainable way to expand local weather station observation networks. The main goal of the project is to establish 3D-PAWS networks in 10 African countries over the next 5 years. Each station includes two rain gauges, in addition to temperature, relative humidity, and air pressure, that transmit data every 15 minutes using local mobile phone networks. This study focuses on the evaluation of rainfall observed from the original 24 3D-PAWS systems that are located throughout Kenya.

Rainfall data collected from the 3D-PAWS were compared to rainfall data observed by co-located Kenya Meteorological Department (KMD) weather stations. The study period ran from September 2023 through May 2024. Data were investigated for overall accuracy and reliability primarily using Python analysis tools to produce scatter plots, box-plots, time series, and bar graphs. Statistical measures, such as correlation coefficient and interquartile range, were then calculated. These data visualizations and statistical measures were analyzed quantitatively to discover patterns and station behavior. The comparative results revealed many stations with at least one well performing rain gauge but also a few stations with at least one poor performing rain gauge when compared to a KMD station. In addition, half of all 3D-PAWS analyzed revealed inconsistent data transmission because of limited network availability in the more remote regions of Kenya, which resulted in data gaps in the analysis. Other factors appeared to impact station overall performance including but not limited to mechanical failures of the rainfall tipping bucket and small discrepancies in observation location.

Understanding 3D-PAWS performance is crucial for ensuring high quality data. The findings of this analysis will provide guidance on when to schedule maintenance for stations with poor performance, and systematic issues affecting a 3D-PAWS system. This effort should help identify when stations need maintenance to ensure overall data integrity. Data produced by high performing stations are ingested into the Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) products to help improve the assessment of extreme rainfall and drought. The CHIRPS data set is a primary tool used by the FEWS NET program to identify and address potential food insecurity around the world. Currently, the 3D-PAWS network is expanding to Zimbabwe in the second half of 2024 to provide additional monitoring in Southern Africa.