Update README.md

README.md

@@ -20,6 +20,29 @@ The original raster tiles at 20 m resolution have been converted into a **single
 This format allows easy analysis with Python (pandas, scikit-learn, etc.), making the dataset suitable for **machine learning, classification, and geospatial studies** without requiring raster-specific libraries.
 
 
+## Date Information
+What the IMCC 2015–2018 dataset represents-
+
+It’s a classified change layer, not a yearly time series.Each pixel encodes whether imperviousness changed between 2015 and 2018, and possibly the type of change (increase, decrease, stable).
+The TIFFs themselves have no time dimension per pixel — the whole dataset is just a spatial snapshot of change across a fixed period.
+
+So, do the dates matter?
+
+**Yes, for context/metadata:** Adding start_date and end_dateis useful when you later merge this dataset with others (e.g., soil, climate, socio-economic data). It tells you that these imperviousness values summarize that 3-year window.
+
+**No, for pixel-level analysis:** Since every pixel shares the same start/end dates, it doesn’t add extra information at the row level. It’s essentially metadata that applies to the whole DataFrame.
+
+**The important dates** - the start & end dates refer to the observation period when this data was collected. 
+
+| Field             | Value          |
+|-------------------|----------------|
+| `dataset`         | IMCC_1518_020m |
+| `start_date`      | 2015-07-17     |
+| `end_date`        | 2018-08-16     |
+| `publication_date`| 2020-07-10     |
+
+
+
 ## Dataset Structure
 
 ### Columns