<aside>

🌊 What’s New?

August 2025

• Unification of QC methodology across all water level deployment types
• More rigorous testing, improved accuracy, and reduced data artifacts
• Clean data are generated nearly every 15min and are available on a standard 1-min interval </aside>

Revision History

<aside>

Scope and Purpose

Hohonu provides precision water level monitoring and forecasting to help communities prepare for and respond to flooding. The organization maintains a network of nearly 200 water level sensors located in 18 states, and it uses the latest technologies in hardware, software, and data science in order to deliver reliable and accessible water level data to its customers.

Adherence to rigorous quality assurance (QA) and quality control (QC) protocols is essential for providing high-quality data to stakeholders. The U.S. Integrated Ocean Observing System (IOOS) provides authoritative procedures through the Quality Assurance/Quality Control of Real-Time Oceanographic Data (QARTOD) program. Additional resources are available specifically to QC flooding data on a real-time basis. Hohonu implements these procedures within its own QA/QC program.

This document provides an overview of the QC methodology used by Hohonu to ‘clean’ raw sensor data and ensure the provision of high-quality real-time water level measurements to its customers. It is intended as technical support for users of Hohonu’s data, and covers both Hohonu’s tidal and overland sensor networks. While procedures documenting data acquisition QA is outside the scope of this document, additional information on sensor specifications, calibration and maintenance, and leveling are referenced at the end.

</aside>

<aside>

Vision for Hohonu Data

• Real-time, readable signal
  • Efficient provision of data
  • Aligned timestamps
• Bad data are accurately filtered
  • Good data are ‘credible’ (i.e. physical, stable)
  • Tests and parameters are tuned to maximally balance sensitivity and specificity
    • Minimal artifacts introduced to data from the cleaning and imputation process
  • Data structure or trends are not assumed without evidence (e.g. pre-existing flood, climatology)
• Transparent cleaning process
  • Cleaning process is well-tracked (robust flags, clear indication of any replaced value)
  • Raw or suspect data are properly indicated

</aside>

Data QC Methodology

Hohonu’s water level data QC system is divided into two stages:

• Basic QC: Efficient testing and processing of raw data upon sample ingress. Flags and removes obvious noise irrespective of neighboring data points. Semi-clean signal for near-real time data visualization. Unstructured in time.
• Advanced QC: More rigorous but more resource intensive testing. Utilizes common tests and tests for signal volatility. Produces a clean, time-structured signal for both data visualization and analysis.

Most of the individual tests within Hohonu’s QC methodology are borrowed from the IOOS Water Level QARTOD procedures, which provide a hierarchy of 11 QC tests for real-time water level data [ref]. While some individual testing parameters may vary between over land (i.e. inundation) and over water deployment types, the following data pipeline applies to all water level station types.

Overview of the Hohonu Data QC System (as of August 2025)

Pre-Processing

Basic QC

Advanced QC

Data Access

All raw and cleaned sensor data along with QC flags are stored on the Hohonu server and are available for further QC, reporting, manipulation, and forecasting. Due to the tiered approach to QC, data are available in three quality levels:

• QC0 (QC Level 0) - Raw data. Decoded transmission messages with an unstructured timestamp
• QC1 (QC Level 1) - Slightly processed data. Data have undergone Basic QC to remove objective noise outside of the sensor range and distances below a surface (if applicable). No time structure imposed on this level.
• QC2 (QC Level 2) - Clean data. Data have undergone Advanced QC to remove common types of noise and high periods of volatility. Data have been resampled to 1-min mean values.

Users can access these data via the user interface on the Hohonu Dashboard, via a mobile iOS app, or through the Hohonu Public API. References for data access are provided below: