From c47c8823c6c026e0bbcdab3f9aa762d63da06cf3 Mon Sep 17 00:00:00 2001
From: rasmaormane <164026305+rasmaormane@users.noreply.github.com>
Date: Fri, 20 Jun 2025 14:14:54 +0100
Subject: [PATCH] Update LPS_training_exercise2.ipynb

Few minor changes
---
 LPS_training_exercise2.ipynb | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/LPS_training_exercise2.ipynb b/LPS_training_exercise2.ipynb
index 0ce27c3..3fe9f38 100644
--- a/LPS_training_exercise2.ipynb
+++ b/LPS_training_exercise2.ipynb
@@ -248,7 +248,7 @@
         "id": "tsWrx8i9W235"
       },
       "source": [
-        "Uncertainty variables have an assocaited error-correlation structure -- since we didn't define this for `u_noise`, it is assumed to be random (i.e., errors are uncorrelated between pixels).\n",
+        "Uncertainty variables have an associated error-correlation structure -- since we didn't define this for `u_noise`, it is assumed to be random (i.e., errors are uncorrelated between pixels).\n",
         "\n",
         "Next let's add a calibration uncertainty component, `u_bt_cal`, with a more complicated error-correlation structure using the `err_corr` attribute. This uses the [error-correlation parameterisations](https://comet-toolkit.github.io/unc_website/specification/draft-v0.1/unc_specification.html#appendix-a-error-correlation-parameterisations) defined by the draft UNC Specfication (it is also possible to add custom error-correlation parameterisations).\n",
         "\n",
@@ -386,7 +386,7 @@
         "\n",
         "Thermal Infrared multi-spectral data, like our example AVHRR dataset, is used to develop Level 2 (L2) Climate Data Records (CDRs) such as Sea or Land Surface Temperature (SST or LST). SST/LST retriavals account for the atmosphere to evaluate the surface temperature from the top of atmosphere L1 brightness temperature.\n",
         "\n",
-        "A widely approach for this is called the \"split window\" method. A simplified form of this algorithm could be represented as,\n",
+        "A widely used approach for this is called the \"split window\" method. A simplified form of this algorithm could be represented as,\n",
         "\n",
         "$SST = a T_{11} - b T_{12}$\n",
         "\n",
@@ -424,7 +424,7 @@
         "\n",
         "**punpy**'s [`MeasurementFunction`](https://punpy.readthedocs.io/en/latest/content/punpy_digital_effects_table.html#measurement-function) class enables a much simpler method for propagating the uncertainties of measurement datasets defined using **obsarray**. It is an alternative interface to the **punpy** propagation functions we used in Step 3.\n",
         "\n",
-        "For this approach instead defining the measurement function as a python function, we define a measurement function class which should be a subclass of the **punpy** `MeasurementFunction` class. We can then use the class's [`propagate_ds`](https://punpy.readthedocs.io/en/latest/content/punpy_digital_effects_table.html#functions-for-propagating-uncertainties) method to propagate all dataset uncertainties in one go!"
+        "For this approach instead of defining the measurement function as a python function, we define a measurement function class which should be a subclass of the **punpy** `MeasurementFunction` class. We can then use the class's [`propagate_ds`](https://punpy.readthedocs.io/en/latest/content/punpy_digital_effects_table.html#functions-for-propagating-uncertainties) method to propagate all dataset uncertainties in one go!"
       ]
     },
     {
@@ -518,7 +518,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "# **link to Next Exercise**\n",
+        "# **Link to next exercise**\n",
         "\n",
         "In [Exercise 3](https://colab.research.google.com/github/comet-toolkit/comet_training/blob/main/LPS_training_exercise3.ipynb) you will either try to apply these tools to propagate uncertainties for your own usecase, or alternatively try to implement uncertainty propagation for a usecase we have provided. "
       ]