Complex Data Preparation and Preprocessing for Predicting Forest Pests with GeoAI

Complex Data Preparation and
Preprocessing for Predicting Forest
Pests with GeoAI

Presenters
The
Peak
of
Data
Integration
20
23
Dr. Christopher
Britsch
Consultant
con terra GmbH

The
Peak
of
Data
Integration
20
23
Agenda
1. Introduction
2. Project Background
3. Data Preparation and Preprocessing
4. Conclusion

The
Peak
of
Data
Integration
20
23
Introduction

The
Peak
of
Data
Integration
20
23
GeoAI - Definition
GeoAI is a Machine
Learning technology, which
enables the caption and
analysis of complex patterns
and structures in
(geospatial) data.

The
Peak
of
Data
Integration
20
23
Setup of a „typical“GeoAI Project
Data Preprocessing
AI-Methods
Integration and Operation
Visualization and Application

The
Peak
of
Data
Integration
20
23
Project Background

The
Peak
of
Data
Integration
20
23
“We are the first generation to feel the impact
of climate change and the last generation that
can do something about it”.
Barack Obama
FormerPresidentof the USA

The
Peak
of
Data
Integration
20
23
Project - KINoPro
Künstliche Intelligenzzur
NonnenfalterPrognose
• Research Project of TU Dresden and con terra GmbH
• Data from state forestries Brandenburgand Saxony
• Climate Change Influence
• Trees struggling with dry and hot weather
• Forestpests adapt fasterthanplants
→ Irregularpopulationgrowth/appearence
• Forestry personnel needs to be managed more efficiently
• New prediction models are necessary

The
Peak
of
Data
Integration
20
23
Approach- Overview
Monitoring of
Black Arches
GeoAI-Modell
Influencing
Factors
Data
Research
Web Application
Data
Preparation and
Preprocessing
Validation and survey for additional
parameters

The
Peak
of
Data
Integration
20
23
Data
● Around 6k data points
● Right-skewed distribution
● Data from 70 weeks per
point
| 11

The
Peak
of
Data
Integration
20
23
Parameters
• Target Variable:
Moth Count per Year
• Additional Information:
• Land Cover Classification
• Altitude
• Slope
• Trap Alignment
Air
Temperature
Rainfall
Humidity
Wind
Speed
Depth of
Frost
Sum of last
years Black
Arches
Soil
Moisture
Soil
Temperature

The
Peak
of
Data
Integration
20
23
Output
4 weeks prior to
moth activities
June 15th
August 16th

The
Peak
of
Data
Integration
20
23
Data Preparation and
Preprocessing

The
Peak
of
Data
Integration
20
23
Challenges in the Data
● Data Formats
○ Excel files, Access-DBs, Esri ASCII Grids, GeoTIFFs, netCDF, etc.
● Projections
○ Different projections, partially highly individual projections
● Spatial Correlation/Corrections
○ Traps can slightly change position
○ Positions are, however, critical to connect previous with current year

The
Peak
of
Data
Integration
20
23
● Pest Monitoring Strategies
○ Different states – different strategies
→ data has to be standardized
● Weather Data
○ Changes in temporal resultion, terminologies, accuracies
→ data has to be standardized

The
Peak
of
Data
Integration
20
23
● Data Quality Assurance
○ Filtering of nonsense data
○ Interpolating missing data
● Performance
○ Large raster files are minimized by clipping

The
Peak
of
Data
Integration
20
23
● Most AI algorithms expect vectors of numerical data as
input
● The input for the KINoPro model is composed of features
from various data sources and with different resolutions
● Data must be transformed to a consistent format and
resolution

The
Peak
of
Data
Integration
20
23
Data Preprocessingwith FME
● FME supports a wide range of input data formats
● No need for…
○ …multiple scripts
○ …a variety of different libraries
○ …researching best solutions for each and every format

The
Peak
of
Data
Integration
20
23
GeoTIFFs
ASCII Grids
netCDF
Gridded Binary

The
Peak
of
Data
Integration
20
23
Conclusion

The
Peak
of
Data
Integration
20
23
● All relevant transformations could be performed
● Workspace is easy to read and understand
● Every step is reproducible and if necessary, easy to
adjust to new parameters
● All in one place, everything connected

Thank You!
c.britsch@conterra.de

Complex Data Preparation and Preprocessing for Predicting Forest Pests with GeoAI

More Related Content

Similar to Complex Data Preparation and Preprocessing for Predicting Forest Pests with GeoAI (20)

More from Safe Software (20)

Recently uploaded (20)

Complex Data Preparation and Preprocessing for Predicting Forest Pests with GeoAI