diff --git a/CM4nowcast/README.md b/CM4nowcast/README.md
new file mode 100644
index 0000000..a8fb85b
--- /dev/null
+++ b/CM4nowcast/README.md
@@ -0,0 +1,90 @@
+# CM4nowcast
+
+## 1. Introduction
+
+CM4nowcast provides the implementation of a Customized Multi-Scale (CM) deep learning framework for storm/precipitation nowcasting.
+The framework is designed to better capture complex multi-scale spatiotemporal variability and improve forecast quality, especially for high-impact storms.
+
+If you use this code in academic work, please cite the paper listed in Section 4.
+
+Repository path:
+`Code/CM4nowcast/` (this folder)
+
+## 2. Required Python Packages
+
+Recommended environment:
+- Python >= 3.8
+- CUDA + PyTorch (if using GPU)
+
+Typical dependencies:
+- numpy
+- scipy
+- pandas
+- xarray
+- matplotlib
+- tqdm
+- pyyaml
+- torch
+- torchvision (optional)
+- tensorboard (optional)
+
+Example installation:
+
+pip install -r requirements.txt
+
+## 3. Usage Instructions
+
+A typical workflow is:
+1) prepare data
+2) train the model
+3) run inference / nowcast
+4) evaluate results
+
+### 3.1 Data preparation
+
+Place your data under:
+
+- data/raw/
+- data/processed/
+
+### 3.2 Training
+
+Example:
+
+python train.py --config configs/train.yaml
+
+### 3.3 Inference / Nowcasting
+
+Example:
+
+python inference.py --config configs/infer.yaml --ckpt checkpoints/best.pt
+
+### 3.4 Evaluation
+
+Example:
+
+python evaluate.py --config configs/eval.yaml
+
+## 4. Citation
+
+If you use this code in your research, please cite:
+
+Yang, S., & Yuan, H. (2023). A Customized Multi-Scale Deep Learning Framework for Storm Nowcasting.
+Geophysical Research Letters, 50, e2023GL103979.
+https://doi.org/10.1029/2023GL103979
+
+BibTeX:
+
+@article{{yang2023cm4nowcast,
+  title   = {{A Customized Multi-Scale Deep Learning Framework for Storm Nowcasting}},
+  author  = {{Yang, Shangshang and Yuan, Huiling}},
+  journal = {{Geophysical Research Letters}},
+  volume  = {{50}},
+  pages   = {{e2023GL103979}},
+  year    = {{2023}},
+  doi     = {{10.1029/2023GL103979}}
+}}
+
+## 5. License
+
+Add a license file (e.g., MIT / Apache-2.0) if distributing publicly.