Python for Climate Data Analytics Training Course

Participants can clean and transform time-indexed climate observations using pandas.

• Pandas DataFrames constructed from CSV, Excel, JSON, and Parquet files
• Datetime indexes normalized across UTC, local time, calendars, and missing timestamps
• Vectorized transformations replace row loops for climate indicator calculations
• Missing values diagnosed using sentinel codes, masks, and completeness thresholds
• Groupby aggregations produce station, season, year, and catchment summaries
• Merge, join, reshape, and pivot operations preserve analytical keys

Participants can apply documented quality controls to climate and contextual datasets.

• Data dictionaries define variables, units, sources, temporal coverage, and caveats
• Range checks flag impossible temperature, precipitation, humidity, and wind values
• Duplicate records and timestamp gaps identified through scripted validation tests
• Unit conversions use pint-compatible quantities and documented conversion factors
• Quality flags retained rather than silently discarding suspect observations
• Validation reports quantify missingness, outliers, and failed data rules

Participants can produce clear climate charts that support analytical interpretation and decisions.

• Matplotlib figures apply labelled axes, units, legends, annotations, and accessible colours
• Seasonal climatologies plotted with uncertainty bands and baseline comparisons
• Distribution plots reveal skewness, thresholds, extremes, and measurement anomalies
• Small multiples compare stations, scenarios, seasons, and population groups
• Plotly charts support inspection of daily series and threshold exceedances
• Figure specifications record source data, transformations, and intended decisions

Participants can combine vector, raster, and administrative data for spatial climate-risk analysis.

• Coordinate reference systems selected for distance, area, and interpolation tasks
• GeoPandas joins station observations to districts, assets, watersheds, and boundaries
• Rasterio reads gridded hazard layers, nodata masks, and affine transforms
• Zonal statistics summarize rainfall, heat, and drought indicators by polygon
• Spatial joins test exposure of roads, clinics, farms, and settlements
• GeoPackage outputs preserve geometry, attributes, coordinate systems, and metadata

Participants can inspect and process multidimensional climate datasets using xarray and CF metadata.

• NetCDF files inspected through dimensions, coordinates, variables, attributes, and encodings
• CF Conventions interpret standard names, units, calendars, and grid mappings
• xarray DataArrays subset latitude, longitude, time, pressure, and ensemble dimensions
• Chunking strategies manage memory when reading multi-year gridded archives
• Area-weighted averages account for latitude-dependent grid-cell surface areas
• NetCDF outputs retain provenance, compression, coordinate bounds, and CF metadata

Participants can calculate and interpret observed climate changes against defined reference periods.

• Reference periods selected according to WMO climatological normal guidance
• Daily observations aggregated into monthly, seasonal, annual, and hydrological-year series
• Anomalies calculated against fixed baselines with explicit missing-data rules
• Trend estimates use ordinary least squares, Sen slope, and Mann-Kendall
• Confidence intervals distinguish signal, variability, sampling limitations, and data gaps
• Change summaries separate observed conditions from modelled future projections

Participants can calculate climate extremes indicators and interpret extreme-value estimates responsibly.

• ETCCDI indices calculate hot days, warm nights, heavy rain, and dry spells
• Threshold definitions align percentile baselines with operational impact questions
• Annual maximum series prepared for flood, heat, and wind analyses
• Generalized extreme value distributions estimate return levels and uncertainty
• Peaks-over-threshold analysis applies generalized Pareto distributions to exceedances
• Nonstationarity assumptions documented when interpreting changing return periods

Participants can produce drought indicators relevant to water, agriculture, and contingency planning.

• Standardized Precipitation Index calculated across one-, three-, and twelve-month windows
• Standardized Precipitation Evapotranspiration Index incorporates temperature-driven atmospheric demand
• Potential evapotranspiration methods compared for data availability and climate conditions
• Soil moisture anomalies linked to agricultural stress and irrigation triggers
• Drought onset, duration, severity, and recovery extracted from time series
• Indicator thresholds connected to contingency actions and monitoring frequencies

Participants can analyse heat hazards and population exposure for heat adaptation decisions.

• Heat index and wet-bulb temperature calculated from temperature and humidity
• Heatwave definitions specify intensity, duration, seasonality, and local thresholds
• Population exposure combines hazard surfaces with age, health, and housing data
• Urban heat patterns assessed using land surface temperature and impervious cover
• Cooling-centre access mapped against travel time and vulnerable population distribution
• Heat-health warning triggers tested against historical mortality or admission data

Participants can combine rainfall, flood, asset, and service data to assess flood exposure.

• Design rainfall hyetographs derived from intensity-duration-frequency relationships and storm profiles
• Antecedent rainfall indicators characterize catchment saturation before flood events
• River gauge series aligned with rainfall observations and upstream timing
• Flood extent rasters intersect assets, land use, population, and critical services
• Exposure tables quantify affected lengths, counts, areas, and replacement values
• Event catalogues link observed impacts to hazard magnitude and warning lead times

Participants can select, summarise, and communicate CMIP6 climate projection information.

• CMIP6 ensembles selected by experiment, model family, resolution, and variable
• Shared Socioeconomic Pathways distinguished from emissions scenarios and policy forecasts
• Ensemble summaries report median, spread, agreement, and outlier model behaviour
• Future periods compared against baselines using consistent calendars and units
• Model independence and performance considerations inform ensemble selection rationale
• Projection statements separate scenario uncertainty from internal climate variability

Participants can prepare climate projection inputs using appropriate bias-correction and downscaling methods.

• Delta change applies projected monthly shifts to observed baseline series
• Quantile mapping corrects distributional bias while preserving tail behaviour
• Bias correction validation uses withheld periods, diagnostics, and seasonal stratification
• Downscaled data limitations documented for convective rainfall and local topography
• Calendar harmonization handles 360-day, noleap, and Gregorian model outputs
• Scenario datasets packaged with assumptions, provenance, versioning, and uncertainty notes

Participants can acquire climate datasets responsibly and record sufficient provenance for reuse.

• Climate Data Store requests retrieve ERA5 reanalysis and CMIP6 subsets
• STAC catalogues discover satellite assets through spatial and temporal queries
• APIs ingested with pagination, authentication, rate limits, and response validation
• Station data reconciled against agency metadata, relocation history, and instruments
• Open licensing checked before redistribution of climate, census, and asset datasets
• Provenance logs record download dates, query parameters, versions, and licences

Participants can structure climate risk analyses that connect data outputs to adaptation priorities.

• Climate risk chains connect hazard, exposure, vulnerability, and adaptive capacity
• IPCC AR6 risk framing distinguishes hazard drivers from socioeconomic vulnerability
• Impact pathways translate rainfall, heat, or drought indicators into service disruptions
• Vulnerability indicators combine sensitivity, coping capacity, poverty, and accessibility
• Multi-criteria scoring applies transparent weights, scales, and sensitivity tests
• Risk matrices communicate priority locations without concealing underlying uncertainty

Participants can translate climate analysis into adaptation actions, thresholds, and monitoring arrangements.

• Adaptation options screened against effectiveness, feasibility, equity, cost, and maladaptation
• Decision thresholds define actions for forecast, observation, and projection signals
• Adaptation pathways sequence no-regret, incremental, and transformational interventions
• Monitoring indicators distinguish climate outcomes, outputs, uptake, and maintenance
• Residual risk documented after planned controls and adaptation investments
• Decision logs record assumptions, responsible owners, review dates, and triggers

Participants can package climate analysis into reviewable reports, maps, and operational products.

• Reusable functions generate indicators, maps, tables, and publication-ready figures
• Jupyter notebooks combine narrative, code, outputs, sources, and assumptions
• Quarto reports render reproducible HTML, PDF, and Word decision briefs
• Dashboards present filterable hazards, locations, scenarios, and adaptation indicators
• Peer review checklists test calculations, visuals, metadata, and decision claims
• Versioned releases distinguish draft analysis from approved operational products

Participants can improve performance and repeatability for large climate data workflows.

• Dask parallelizes xarray workflows across chunks, workers, and distributed memory
• Zarr stores cloud-ready multidimensional arrays with efficient partial reads
• Profiling identifies slow functions, unnecessary copies, and inefficient input-output operations
• Pipeline stages cache intermediate results with deterministic file naming
• Command-line scripts schedule repeatable updates for new observation and forecast data
• Failure logs capture exceptions, source outages, and incomplete processing runs

Participants can apply governance and equity considerations to climate adaptation data products.

• FAIR principles assess findability, accessibility, interoperability, and reusability of datasets
• Data protection controls minimize disclosure of household, health, and location records
• Indigenous data sovereignty considered when using community territories and knowledge
• Uncertainty statements avoid false precision in high-stakes adaptation investment decisions
• Equity checks test whether aggregated indicators mask disadvantaged populations
• Analytical audit trails support challenge, revision, approval, and accountability

Participants can build, review, and present a complete climate adaptation analytics product for a defined decision.

• Problem framing defines decision owner, location, climate hazards, and time horizon
• Data inventory records sources, licences, variables, resolution, and quality limitations
• Reproducible Python pipeline integrates observations, projections, spatial exposure, and indicators
• Risk analysis ranks priority impacts using transparent vulnerability and uncertainty assumptions
• Adaptation recommendations connect evidence, thresholds, responsible owners, and monitoring indicators
• Peer review and presentation defend methods, findings, limitations, and implementation actions