The River Remembers
See when the White River ran higher or lower than usual across 95 years.
Try the project ↓How to use this project
Help
How to explore
The White River changes every day as water moves through Indianapolis. This project compares one streamgage's complete years without treating one peak as the whole story.
Start with 1941
Follow the longest low stretch in the saved record, then move the date control to see the exact flow and its usual range.
Choose another year
Use the month rows to find a year with many lower or higher months. Choosing a row loads its daily line above.
Compare long stretches
Switch between all unusual days and the longest unbroken stretch to separate repeated events from one sustained period.
Compare seasons
Read the two 30-year profiles, then check the limits before drawing a conclusion about why they differ.
- Discharge
- The volume of water passing the streamgage in a set amount of time; this project calls it flow in the operating copy.
- Cubic feet per second
- The source unit for flow; one cubic foot is about 7.48 gallons.
- Daily mean
- The average of the streamgage's readings across one day, as published by USGS.
- Median
- The middle value after observations are put in order, used here to describe a typical day or month.
- Percentile threshold
- A cutoff based on ordered observations; the 90th-percentile threshold has roughly nine tenths of values below it.
White River · Indianapolis · 1931–2025
The River
Remembers
Choose a year and see when the White River ran higher or lower than usual for that date.
A streamgage near downtown records how much water passes each second. One year can hold months of low water, short surges, or both.
01 / Follow one year
How did the river move through 1941?
The line shows daily mean flow. The pale band holds the middle half of recorded values for each calendar date across 1931–2025. The chart uses a logarithmic scale so low and high flows both remain visible.
02 / Scan 95 years
Which months stood apart?
Each row is one complete year. A minus marks a month whose median flow was among the lower quarter for that month; a plus marks the upper quarter. Choose a row to load it in the daily chart.
This view compares January with other Januaries, February with other Februaries, and so on. It does not compare one month directly with the next.
03 / Compare long stretches
How long did unusual flow last?
A high day is above the 90th-percentile threshold for its calendar date. A low day is below the 10th-percentile threshold. Switch the measure, then choose any year.
04 / Compare seasons
How does an earlier 30-year period compare with a recent one?
For each month, the bars show the middle daily flow across all 30 years in that period. The recent-period median is higher in every month, but this one gage cannot tell us why.
Read the monthly values
| Month | 1931–1960 | 1996–2025 |
|---|---|---|
| Jan | 842 | 1,290 |
| Feb | 1,150 | 1,255 |
| Mar | 1,570 | 2,060 |
| Apr | 1,490 | 1,835 |
| May | 1,060 | 1,565 |
| Jun | 792 | 1,395 |
| Jul | 484 | 738 |
| Aug | 261 | 388 |
| Sep | 196 | 309 |
| Oct | 212 | 357 |
| Nov | 295 | 520 |
| Dec | 450 | 856 |
Why this exists
The question
How different did whole years look at the White River streamgage in Indianapolis, and how long did unusually high or low flow last?
What to try
Start with 1941, move through its daily flow, choose another year from the 95-row month view, then compare total unusual days with the longest unbroken stretches.
Method and limits
About the data
I selected the U.S. Geological Survey's daily mean discharge series for station 03353000, White River at Indianapolis, after checking the current Water Data for the Nation location page, modern Water Data API documentation, and USGS public-domain terms. I retrieved the API response on September 25, 2026. The 5,828,964-byte CSV contains 35,781 approved observations through 2025 and has SHA-256 493a5f834e7da935541ac9bd8d830d656ad76b6ea3a85f700120d136e392a5fe.
The checked-in reduction keeps all 34,699 daily mean values in the uninterrupted calendar-year span from 1931 through 2025. USGS marked 1,768 retained observations as estimated; 43 more carry ice, revision, equipment, or forced-interpolation qualifiers. I retain those published observations. The reproducible Python script validates the source series, unit, approval status, checksum, dates, and row completeness before writing a deterministic 219,448-byte snapshot. The page imports that snapshot and makes no runtime network requests.
For each calendar date, I calculate the 10th, 25th, 50th, 75th, and 90th percentiles across the 95 complete years with linear interpolation. February 29 uses the 24 leap-year observations. A high day is strictly above its date's 90th-percentile threshold; a low day is strictly below its 10th-percentile threshold. Longest stretches require consecutive calendar dates within one year. For the month rows, I calculate each year's monthly median, then compare it with the distribution of the other medians for the same month. The seasonal view pools all daily values in a month within 1931–1960 or 1996–2025 and takes the median.
USGS lists the streamgage's drainage area as 1,635 square miles. The record describes water passing that one point, not the whole river or watershed. Rain, snow, drought, reservoirs, withdrawals, land use, channel changes, instruments, and calculation practices can affect flow. USGS may revise approved data. These comparisons do not establish causes, a climate trend, flood frequency, ecological condition, or a safety threshold.
Skills used
- Data analysis
- Data engineering
- Statistical reasoning
- Systems design
- Visualization
- Information design
- Interaction design
- Visual design
- Creative coding
- Product thinking
- Editorial storytelling
- Accessibility
- Performance
What the data shows
- 1941 has the lowest annual mean flow among the 95 complete years at 252.8 cubic feet per second. It has 227 unusually low days, including the record's longest low stretch: 104 days from February 19 through June 2.
- 1990 has the most unusually high days at 110. The longest single high stretch occurs in 2001 and lasts 32 days, showing that the annual count and the longest continuous period answer different questions.
- The 1996–2025 monthly median is higher than the 1931–1960 median in all 12 months. June has the largest absolute difference, 1,395 versus 792 cubic feet per second, but the streamgage record alone cannot explain the change.