Lesson 0: introduction

Author: Su Ye (remotesensingsuy@gmail.com)

This tutorial was made to present the examples for using pyxccd for multiple remote sensing and ecological applications using multi-source time-series datasets.

Special thanks to Tianjia Chu, Ronghua Liao, Yingchu Hu, and Yulin Jiang for preparing tutorial datasets.

Preparation

First, please install pyxccd. In a Jupyter notebook cell, run:

pip install pyxccd

Additionally, you need to install the visualization package:

pip install seaborn

Download the recent source codes of pyxccd in the devel branch, unzipped it, and under the directory /pyxccd/tutorial, the directory should look like:

└── notebooks
└── datasets

Learning Pyxccd with Examples

To illustrate the utilities of pyxccd, we prepared multiple notebook examples using multivariate satellite-based time series across a wide range of applications in this tutorial:

No.	Topics	Applications	Location	Time series	Resolution	Density
1	Break detection	Forest fire	Sichuan, China	HLS2.0	30m	2-3 days
2	Parameter selection	Forest Insects	CO & MA, United States	Landsat	30m	8-16 days
3	Flexible choice for inputs	Crop dynamics	Henan, China	Sentinel-2	10m	5 days
4	Tile-based processing	General disturbances	Zhejiang, China	HLS2.0	30m	2-3 days
5	State analysis 1	Greening	Tibet, China	MODIS	500m	16 days
5	State analysis 2	Precipitation seasonality	Arctic	GPCP	2.5°	Monthly
6	Anomalies vs. breaks	Agricultural drought	Rajasthan, India	GOSIF	0.05°	8 days
7	Near real-time monitoring	Forest logging	Sichuan, China	HLS2.0	30m	2-3 days
8	Gap filling	Soil moisture	Henan, China	FY3B	25km	Daily

Note:

1. The tutorial primarily provides pixel-based time series examples for educational purposes; however, in practical applications, analyses are typically performed on image-based datasets. In Lesson 4, we will specifically demonstrate the procedures for applying pyxccd to real-world image-based time series;

2. All date columns in the tutorial are formatted as Gregorian proleptic ordinal numbers, representing the number of days elapsed since 0001-01-01. Users can convert the ordinal date format to human-readable date format using the Python function datetime.date.fromordinal().