UNDER CONSTRUCTION (10/15/26)

TL;DR:

In this project, I systematically measure watermelons using 38 different parameters in an effort to predict watermelon sweetness based on its externally measurable characteristics. Generally speaking, the metrics may be grouped as follows. In addition, a public survey was conducted of more than 1,000 respondants asking which methods they believe are key to choosing a sweet watermelon.

Parts List:

TOOLS:

• Audio-Technica AT2020: A professional-grade microphone for recording audio data [Link]
• Analog Brix Refractometer: An optics-based manual tool to measure sugar content [Link]
• Digital Brix Refractometer: A digital tool to measure sugar content [Link]
• Audacity: Free audio recording and processing software [Link]
• Diglta calipers: :High-precision metal calipers with digital readout [Link]
• Pipette: Plastic pipettes to collect watermelon juice [Link]

Introduction

Every summer, people flock to grocery stores in search of a refreshing seasonal favorite, the watermelon. Regardless of age, sex, or ethnic background, picking a sweet watermelon presents a dilemma for us all: Without cutting it open, how do I pick out a sweet watermelon?

With no clear answer, theories abound. Some say “knocking” on the watermelon is key, listening for a hollow-sounding response. Others swear by the size of the ‘yellow spot’ where the watermelon rested on the ground as it grew. Still others claim a sweet, ripe watermelon should be able to stand on its end without help. But which metrics withstand scientific rigor? That’s where this research comes in.

In this multi-year survey, I systematically measure more than 35 metrics across 30+ watermelons, compiling their results into a highly unique database. By directly correlating visual, acoustic, and structural parameters of each specimen with its sweetness in five or more locations, the database can be used to identify data trends, parameter relationships, and analyze metrics using conventional- or AI-based methods..

Data Collection

Introduction

The collected data metrics can be cleanly separated from the data collection process, since the former is a list, while the latter is a workflow. They are presented in this order below.

Basic metrics

In this context, “basic metrics” refer to non-measured watermelon parameters, including:

• Measured
  • Purchase date
  • Purchase venue
  • Purchaser
  • Watermelon type (mini or standard)
• Derived
  • Seasonal parameter fluctuations

Visual metrics

Visual metrics refer to features on the outside of the watermelon that give each its distinct look, including:

• Measured
  • Number of dark green bands, counted around the equator
  • Number of dark green bands, counted at the non-stem end
  • Yellow spot diameter along major and minor axes [mm]
  • Scarring images
• Derived
  • Surface area of the largest scarred regions

Acoustic metrics

Acoustic metrics refer to features of the audio recorded from “knocking” on the watermelon under different circumstances. Although only a small number of knocks are recorded, many acoustic properties can be derived from the data, including:

• Measured
  • Knocks recorded under three different circumstances:
    • Watermelon held on the bottom, three knocks at each of the top, middle, and bottom
    • Watermelon held on the side, three knocks at each of the top, middle, and bottom
    • Watermelon held on the top, three knocks each at the bottom, middle, and top
    • Watermelon held on the bottom, knocks randomly around the watermelon
• Derived
  • Per knock
    • Frequency of the first, second, and third loudest spectral peak [Hz]
    • Lowest resonance still within 20 dB of the loudest peak [Hz]
    • Dominant peak measured after skipping 15 ms (excludes impact click) [Hz]
    • Power-weighted mean frequency [Hz]
    • Power-weighted spread around the centroid [Hz]
    • Amplitude 1/e time [ms]
    • Resonance sharpness (Q-factor)
    • How long the level holds up after the peak (sustain) [ns]
  • Per melon
    • Median low frequency across all knocks [Hz]
    • Median dominant frequency [Hz]
    • Median low frequency across all strikes [Hz]
    • Median low frequency from random strike locations [Hz]
    • Spread of low frequency across the whole melon [Hz]
    • Knock decay time [ms]
    • Cooke and Rand firmness index removing the effect of size
    • Ratio of knock frequency of top and bottom

Structural metrics

Structural metrics refer to the size of various features on the watermelon, including:

• Measured
  • Weight [lbs., oz]
  • Whether the watermelon can stand upright on the non-stem end [0,1]
  • Circumference around the latitudinal and longitudinal axes [mm]
  • Diameter of the major and minor axes of the non-stem end [mm]
  • Distance from the yellow spot to the nearest end [mm]
  • Width of the thickest and thinnest part of the cross-section rine [mm]
  • Weight of the rine after watermelon flesh has been removed [lbs., oz]
  • Number of elongated ridges typically running along the dark bands

Data Collection

For

When

Transcript parsing

After

Data Analysis

Concurrent viewers

Concurrent

Additionally,

The impact of events

During

Fig. 4 s

Event timing

The

asdf

Analyzing the data left to right reveals some interesting points:

Results

Once

Of

Conclusion

The