Author

Caiyun Tech Team

Engineering

Articles

Likes

Views

Comments

Latest from Caiyun Tech Team

10 recent articles

Caiyun Tech Team

Jan 30, 2026 · Artificial Intelligence

DeepFusion: A Multi-Modal Deep Fusion Network for Short-Term Heavy Rainfall Forecasting

DeepFusion combines radar, historical precipitation, numerical weather prediction, and terrain data through independent encoders and attention‑based multi‑scale fusion, achieving high‑resolution 0‑2 hour heavy‑rain forecasts and demonstrating the benefits of multi‑modal deep learning for short‑term precipitation prediction.

DeepFusionNWPU-Net

0 likes · 13 min read

DeepFusion: A Multi-Modal Deep Fusion Network for Short-Term Heavy Rainfall Forecasting

Caiyun Tech Team

Jan 6, 2025 · Artificial Intelligence

How Domain Knowledge Boosts GNN Accuracy for PM2.5 Forecasting

The article presents PM2.5‑GNN, a graph neural network that incorporates meteorological and geographic domain knowledge via a dynamic directed graph and a GRU module, achieving significantly lower RMSE and MAE than traditional models and demonstrating strong spatial‑temporal prediction capabilities through extensive experiments and case studies.

Air Quality PredictionDomain KnowledgeGRU

0 likes · 13 min read

How Domain Knowledge Boosts GNN Accuracy for PM2.5 Forecasting

Caiyun Tech Team

May 30, 2024 · Big Data

Designing a Sparse‑Matrix Based Format for Radar dBZ Data (smdbz)

The article presents a sparse‑matrix compression scheme called smdbz for high‑resolution radar dBZ data, detailing the dBZ grade encoding, CSR representation, custom Python implementation, bit‑packing techniques, file layout, performance comparisons with PNG and NPY formats, and the method's advantages and limitations.

CSRPythondBZ

0 likes · 21 min read

Designing a Sparse‑Matrix Based Format for Radar dBZ Data (smdbz)

Caiyun Tech Team

Feb 27, 2024 · Fundamentals

cyeva: A Universal Python Toolkit for Weather Forecast Accuracy Evaluation

cyeva is an open‑source Python library that automates deterministic weather forecast accuracy assessment, offering built‑in metrics for temperature, precipitation, wind and other continuous variables, with installation via pip or conda, performance benchmarks, and community‑driven extensions.

MAEPythonRMSE

0 likes · 11 min read

cyeva: A Universal Python Toolkit for Weather Forecast Accuracy Evaluation

Caiyun Tech Team

Feb 22, 2024 · Artificial Intelligence

How to Estimate Visibility Using Atmospheric Tower Photos

This article explains a step‑by‑step method for measuring atmospheric visibility by annotating landmarks in fixed‑angle tower photos, extracting their pixel regions, scoring recognizability with a standard‑deviation‑based metric, and computing the farthest visible object using Python and OpenCV.

OpenCVPythonatmospheric tower

0 likes · 8 min read

How to Estimate Visibility Using Atmospheric Tower Photos

Caiyun Tech Team

Jan 24, 2024 · Backend Development

How We Built a High‑Performance Elevation API with 30 m Resolution

This article details the end‑to‑end process of replacing low‑resolution SRTM data with 30‑meter ASTER elevation data, redesigning storage and access using JuiceFS, Go, and sync.Pool, and achieving sub‑millisecond latency for 1500 QPS in a Kubernetes environment.

GeoTIFFGoJuiceFS

0 likes · 9 min read

How We Built a High‑Performance Elevation API with 30 m Resolution

Caiyun Tech Team

Oct 12, 2023 · Operations

How to Perform Quality Control for Temperature Data

The article explains the challenges of temperature data quality control in weather apps, critiques simple threshold filtering, and presents a statistical‑based solution that builds a probability‑distribution model from a decade of nationwide observations, implements a RESTful API, and evaluates its performance across locations, times, and seasons.

REST APIdata samplingprobability distribution

0 likes · 19 min read

How to Perform Quality Control for Temperature Data

Caiyun Tech Team

Jan 30, 2023 · Backend Development

How Caiyun Weather Optimized Administrative Division Lookups Using H3 Indexing

The article chronicles Caiyun Weather’s evolution of geographic lookup—from a naïve nearest‑adcode algorithm to a high‑precision H3‑based system—detailing the evaluation of Geohash, S2, and H3, performance benchmarks, memory trade‑offs, and the final R‑Tree‑augmented point‑in‑polygon solution.

Geospatial IndexingH3Performance Optimization

0 likes · 10 min read

How Caiyun Weather Optimized Administrative Division Lookups Using H3 Indexing

Caiyun Tech Team

Jan 6, 2023 · Backend Development

Optimizing Geospatial Queries: Finding the K Nearest Weather Stations in Go

The article walks through a real‑world case of locating the K closest national weather stations to a Beijing site, explains why naive full‑scan using Vincenty's formula is slow, and details a tile‑based indexing solution that cuts query time to 900‑3000 ns and reduces CPU usage by about 10 percent.

GoKNNPerformance Optimization

0 likes · 7 min read

Optimizing Geospatial Queries: Finding the K Nearest Weather Stations in Go

Caiyun Tech Team

Dec 30, 2022 · Fundamentals

Understanding Map Projections, Coordinate Systems, and Tile Generation

The article explains why flat maps require projections, compares conformal, equal‑area and equidistant methods, describes the WGS84 datum and the GCJ‑02 “Mars” offset, outlines modern Web Mercator tiling standards, and shows how Python libraries like mercantile, pygrib, xarray, GDAL and rasterio can be used to read gridded data and produce map tiles.

GDALGISPython

0 likes · 14 min read

Understanding Map Projections, Coordinate Systems, and Tile Generation