From Raw JSON to Kibana: Automated Weather Data Pipeline in Brno

At a Glance

The Brno Weather Dashboard Overview

Monitoring real-time environmental metrics requires a clean visualization layer backed by a resilient data pipeline. This project demonstrates how automated data collection, Elasticsearch indexing, and Kibana dashboarding come together to monitor meteorological conditions in Brno.

At a glance, the top section of the Kibana dashboard offers a quick overview of current weather conditions in Brno. It tracks key meteorological phenomena, including temperature, precipitation, air humidity, and wind speed and direction.

The sidebar provides high-level summary metrics, such as minimum, maximum, and average values, calculated across all monitored variables. Below the overview, detailed time-series charts illustrate how these metrics evolve over time, while integrated benchmarks compare local measurements directly against the Czech national median.

The lower panels of the dashboard extend the view beyond temperature. The air humidity chart plots Brno's actual humidity against the Czech national median, revealing how local conditions diverge from the national baseline across the day. The wind speed panel follows the same comparative pattern — tracking both Brno's measured values and the national median in a shared time frame.

At the bottom, the pipeline health panel ('Stav datové pipeline') confirms the ingestion rate, showing a steady ~2,500 records per hour being processed reliably throughout the day. A flat, consistent line here is a signal of a healthy, uninterrupted pipeline — exactly what a production-grade observability stack should look like.

Dashboard Deep Dive: Detail Analysis Rainfall and Wind Dynamics

A second dashboard takes the analysis a level deeper moving beyond headline weather statistics to expose the structural relationships between rainfall and wind dynamics across a rolling 7-day window.

The centrepiece of this view is a dual-axis time-series chart that overlays 10-minute rainfall bursts against real-time wind speeds, making storm-front signatures immediately visible as correlated spikes across both axes. Alongside it, a wind direction donut chart breaks down dominant airflow patterns over the period — with prevailing westerly and northwesterly flows accounting for the majority of observations. A cross-metric scatter panel then links precipitation totals directly to specific wind direction categories and Beaufort force ranges, revealing a consistent pattern: measurable rainfall in Brno occurs predominantly during lighter wind conditions rather than at peak gust speeds.

The temporal dimension is covered by diurnal wind intensity heatmaps, which aggregate wind speed by hour-of-day across the full 7-day window. The resulting pattern is clear and repeatable: speed ramps up between 09:00 and 20:00 before settling into a calm overnight baseline. This kind of time-bucketed aggregation — applied here to meteorological data — is the same pattern RNX uses to build operational intelligence for enterprise infrastructure: converting raw, high-frequency event streams into structured, human-readable signals that expose what the raw numbers alone cannot.

Data Sourcing & Ingestion

The raw data is sourced from the Czech Hydrometeorological Institute (CHMI) website, which refreshes its public data feed every hour. On-site weather stations record measurements at finer 10-minute intervals.

To process this data automatically, a streamlined four-step pipeline runs without manual intervention:

Technical Foundations: The RNX Approach to Data Infrastructure

This project demonstrates the core technical principles, pipeline architectures, and visualization techniques engineered by RNX.

Summary

By combining lightweight ingestion scripts with the analytical power of Elasticsearch and Kibana, this project highlights how modern data pipelines bring clarity to complex time-series metrics. The Brno weather dashboard is a concrete example of the pipeline architectures and visualization patterns RNX engineers for production environments — scalable, automated, and built to run without interruption.

The same core patterns used here — automated ingestion, schema normalization, time-series aggregation, and comparative benchmarking — are directly transferable to enterprise log management, APM, and infrastructure monitoring use cases.