Tutorial Outline
From the primer, we can see a few common tasks for implementing DA. These include:
Organizing climate data
Building a state vector ensemble,
Running forward models on the ensemble,
Estimating proxy uncertainties, and
Running assimilation algorithms
In this tutorial, we will use the DASH toolbox to implement all these tasks. We’ll practice using DASH using some demo datasets, which we’ll provide a bit later. The demo datasets are from published assimilations, and should give sense of typical DASH workflows. The remainder of the tutorial is organized as follows:
- DASH Introduction
We’ll start by introducing the DASH toolbox. We’ll briefly examine its command style, layout, and how to access the documentation.
- Demo Datasets
Next, we’ll introduce and download the demo datasets. We’ll also use a few DASH utilities to examine metadata for these datasets.
- Catalogue Climate Datasets
At this point, we’ll begin using DASH to implement an assimilation. We’ll start by using DASH’s
gridMetadata
andgridfile
classes to describe and catalogue climate datasets.- Build a State Vector
Once our data is catalogued, we’ll use DASH’s
stateVector
class to design and build a state vector ensemble.- Manipulate Ensembles
Next, we’ll examine the
ensemble
class and use it to manipulate our state vector ensembles. We also briefly explore theensembleMetadata
class, which helps locate and organize data within an ensemble.- Implement Forward models
Next, we’ll use the
PSM
interface to run forward models on climate variables in the ensemble. We’ll also use a number ofensembleMetadata
methods to locate forward model inputs.- Run DA Algorithms
With this setup complete, we can now start running DA algorithms. We’ll examine the
kalmanFilter
,particleFilter
, andoptimalSensor
classes and use each of them to run an assimilation.