Day 34: Time Series Analysis & Morphological Change – Masterpiece Edition | River Warrior

📉 DAY 34: TIME SERIES ANALYSIS & MORPHOLOGICAL CHANGE

⏱️ Estimated Reading Time: 16 Minutes | 🎓 Level: Professional Hydrographer / Geomorphologist

From Static Snapshots to Dynamic Change – Understanding Seabed Evolution Over Time

Instructor: Engr. Rokib Hossain | River Warrior Academy

🏠 Course Homepage

📖 Table of Contents (Serialised)

Why Time Series Analysis Is Critical
Preparing Multi‑Temporal Bathymetric Data
Change Metrics: Net, Erosion, Deposition
Rate of Change Calculation & Simulator
Trend Analysis (Linear Regression)
Seasonal Patterns & Monsoon Effects
Case Study: 5‑Year Jamuna River Morphology
Predictive Extrapolation & Early Warning
Time Series Analysis Checklist
Software & Tools
Frequently Asked Questions
Action Items & Next Steps

1. Why Time Series Analysis Is Critical

A single bathymetric survey gives a snapshot. Multiple surveys over time reveal the dynamics of the seabed: erosion, deposition, migration of sandbars, and infill of dredged areas. Time series analysis helps:

Predict future dredging volumes.
Identify erosion hotspots before they become navigation hazards.
Evaluate the effectiveness of coastal protection structures.
Understand seasonal morphological cycles.

🧠 Golden Rule: Consistent survey parameters (datums, grid resolution, tide reduction) are essential for meaningful time series comparisons. Re‑process older surveys if necessary.

🌊 River Warrior Pro-Tip: Jamuna Time Series

We maintain a time series of 12 surveys per year in the Jamuna River (monthly). After five years, we identified a cyclic 18‑month period of sandbar migration – a pattern now used to schedule dredging campaigns. Without time series, this would have remained invisible.

2. Preparing Multi‑Temporal Bathymetric Data

To compare surveys over time, you must ensure:

Same horizontal & vertical datums (geoid, tide correction).
Same grid origin and cell size (or interpolate to a common grid).
Same coverage area (clip to the common extent).
Consistent outlier cleaning (avoid bias from different cleaning levels).

In Qimera or QGIS, you can create a “stack” of rasters and compute statistics per cell.

Grids from different years stacked for per‑cell analysis.

3. Change Metrics: Net, Erosion, Deposition

For each cell or for the whole area, compute:

Net change: Final depth – Initial depth (can be positive or negative).
Erosion volume: Sum of negative changes (material removed).
Deposition volume: Sum of positive changes (material added).
Budget: Deposition – Erosion (net).

In QGIS, use Raster Calculator: `("survey_2025" - "survey_2020")`. Then compute histogram to separate positive and negative changes.

📊 Example: If a 1 km² area shoaled by 0.2 m, the deposition volume = 200,000 m³.

📈 Rate of Change Calculator (Linear)

Enter two surveys to compute annual rate of change:

Depth at time 1 (m): Date of survey 1 (years from now or epoch): Depth at time 2 (m): Date of survey 2 (years):

Rate of change = +0.60 m/year (deposition)

4. Trend Analysis (Linear Regression)

With three or more surveys, you can fit a linear trend (depth vs time) for each cell or the whole domain. The slope gives the rate of change (m/year); the R² indicates how consistent the trend is.

Tools: Excel, Python (scipy.stats.linregress), or QGIS with “Raster Time Series” plugins (e.g., Raster Trend).

Scatter plot of mean depth vs year with linear trend line.

5. Seasonal Patterns & Monsoon Effects

In rivers like the Jamuna, strong seasonal signals exist: monsoon floods cause erosion and sediment transport, dry season shows deposition in some areas. To identify seasonal patterns:

Subtract the long‑term trend to get seasonal residuals.
Plot monthly averages to see annual cycles.
Use harmonic analysis (Fourier) to extract periodic components.

🌊 In Jamuna, we found a 0.5 m annual depth variation: deeper in October (post‑flood), shallower in April (dry season). This guides when to schedule maintenance dredging.

6. Case Study: 5‑Year Jamuna River Morphology (2021–2025)

Objective: Track morphological evolution of a 15 km reach with 20 surveys (quarterly).

Key finding 1: The navigation channel experienced net shoaling of 0.8 m over 5 years, requiring annual dredging.
Key finding 2: Erosion hotspot near a confluence migrated 300 m upstream – risk to a riverbank protection structure.
Key finding 3: Strong seasonal component: 70% of annual deposition occurs in 3 months of the monsoon.
Action: Dredging now concentrated in pre‑monsoon months, reducing re‑deposition by 25%.

📈 Lesson: Time series analysis converted raw data into actionable dredge strategy – saving millions in maintenance costs.

7. Predictive Extrapolation & Early Warning

Using linear trend or more advanced models (e.g., ARIMA), you can forecast future depths. For example, if a navigation channel is shoaling at 0.2 m/year and the minimum required depth is 5 m, you can predict when dredging will be needed.

Alert threshold: Set a “danger depth” and compute the time to reach it based on the trend.

⏳ Time to Threshold Calculator

Current depth (m): Rate of change (m/year, negative if shoaling): Threshold depth (m):

Time until threshold: 40 months (≈3.3 years).

8. Time Series Analysis Checklist

All surveys share same projection, datum, grid origin, cell size.
Tide and SVP corrections consistently applied (or reprocessed).
Raster stack created (e.g., with QGIS or Python).
Net change map and volume computed (erosion vs deposition).
Rate of change map generated (linear slope per cell).
Seasonal pattern analysed (if monthly data).
Trend significance tested (p‑value).
Predictive extrapolation performed for critical areas.
Time series report created with graphs and recommendations.

Click items to track progress (saved in browser).

9. Software & Tools

.htmlPython (xarray, scipy, rasterio)那样Full flexibility, seasonal decomposition, ARIMA那样Open source.htmlCARIS HIPS那样Time series of surfaces, difference volumes那样teledynecaris.com

Tool	Time Series Capabilities	Link
QGIS (free) + Raster Trend plugin那样Cell‑wise linear regression, trend maps那样qgis.org
QPS Qimera那样Change detection, volume tracking, time series export那样qps.nl/qimera

10. Frequently Asked Questions

How many surveys are needed for a reliable trend?

At least 4‑5 surveys spread over 2‑3 years to distinguish seasonal noise from long‑term trend.

What is the difference between net change and erosion/deposition volumes?

Net change = final – initial (single value). Erosion volume = sum of negative changes; deposition = sum of positive changes; net = deposition – erosion.

How do I handle missing data (e.g., no survey in some months)?

Interpolate linearly or use only common timestamps. Avoid extrapolating over large gaps.

Can time series analysis be done with single‑beam data?

Yes, but only along lines. Use profile‑based analysis (depth vs distance) at fixed transects.

How do I present time series results to non‑technical clients?

Use animated maps (difference through time) and simple line graphs of mean depth or volume.

11. Action Items & Next Steps

📌 Gather at least 3 bathymetric surveys of the same area (or use synthetic data).
📌 Create a raster stack in QGIS and compute the mean depth per year.
📌 Use the rate calculator above with your data to estimate annual change.
📌 Proceed to Day 35: Bridge Scour Inspection.

⏳ Day 34: Time-Series Analysis and Morphological Change