Accessing Drought & Fire Indicators

Mapping Drought Indices

In this section, we will go step-by-step through how to create maps of drought indices. These are useful for understanding conditions pre-fire. To tie this to a real work example, we will map drought layers prior to the 2007 Angora Fire. We will see that most of the area inside of the fire perimeter was in severe drought prior to the fire start date.

Step 1: Open the Climate Engine App

To open the Climate Engine app, in your browser head to app.climateengine.org. If this is your first time using the app, you will be prompted to create a free account (find directions here). To make a map, ensure you are on the 'Make Map' tab of the panel on the left hand side of the app.

Step 2: Fill out the 'Variable' section

First, you need to select which type of data variable you are interested in. Since we are interested in historical drought, we will select the 'Climate & Hydrology' type. Next, we will search for 'GridMET Drought - 4km - Pentad' as the 'Dataset'.

Once we selected the dataset, the variables for that dataset will populate the 'Variable' drop-down. We will select a '14 Day EDDI', which has been identified as valuable for understanding fire risk in the western US (McEvoy et al., 2019).

Step 3: Fill out the 'Processing' section

Since we selected a Standard Index variable, there will be only one option for 'Statistic (over day range)', we will leave it at 'No Statistic'. For the 'Calculation' we will leave it at 'Standard Index.'

Step 4: Skip the Masking section

This is optional section that we will skip for these examples.

Step 5: Fill out the 'Time Period' section

Since we selected a pre-calculated standard index, all we need to do is update the 'End Date' value. The fire started on June 24, 2007 so we will use that as our end date to visualize the conditions right before the fire (2007-06-24). Then we will click the 'Get Map Layer' button to visualize the results on the mapping panel on the right of the app.

Optional Step: Add Fire Perimeter Overlay

On the right side of the App, above the map, you will see different drop-downs that allow you to interact with the map. We will expand the 'Map Overlays' drop-down and scroll to the 'US Fire Perimeters Layers' section under 'Vector'. We will then check on the 2007 layer and zoom to just below Lake Tahoe.

Interested in a similar workflow using the Climate Engine API?

The Climate Engine API allows users to extend their analysis by exporting larger geotiffs, creating timeseries data over multiple areas, and writing loops to make multiple API calls. You can find a python notebook walking you through a similar workflow here.

Mapping Fire Indicators

In this section, we will go step by step through how to create maps of historical and forecasted fire indicators. These can help with the context of past fire events, and understanding conditions in the near-future. By combining these two datasets, we can also look how near-future values differ from past values.

Example # 1: Mapping Fire Indicators Prior to the Angora Fire

Similar to the workflow detailed for mapping drought indices, we will start with updating the 'Variable' section. We will keep the 'Type' as 'Climate & Hydrology' and update 'Dataset' to 'GridMET - 4km - Daily'. Lastly, we will update the 'Variable' to 'Burning Index'. Moving to the 'Processing' section, we will leave the 'Statistic (over day range)' as 'Mean' and 'Calculation' as 'Values'. Finally, within the 'Time Period' section we will choose to use a 'Custom Date Range'. The 'Start Date' will be updated to '2007-05-10' to capture the mean values over the two-weeks prior to the Angora Fire. The 'End Date' will be kept at '2007-06-24'. Then we will click the 'Get Map Layer' button. We see that part of the perimeter falls within the 30 - 40 value range which indicates moderate fire danger and the other part falls within the 40 to 60 range which indicates high fire danger (burning index categories).

We can further explore what these values mean by calculating anomalies, i.e. looking at them in context of past values. To do this, we can head back to the 'Processing' section and change the calculation from 'Values' to 'Difference From Average Conditions'. This will update the options for the 'Time Period' section, add a 'Year Range for Historical Avg/Distribution' . We will choose to set the range to all of the years leading up to the fire, '1979' - '2006'. Then we will click the 'Get Map Layer' button. This will show the difference in Burning Index values from the average for 1979 - 2006. We see the values of difference range from +2.2 to > +20.

Another option for understanding how the time period right before the fire differed from previous years is map percentiles. To do this, all we update is the 'Calculation' to 'Percentile in Distribution of Past Observations. Here we see the burning index values in the north section of the fire perimeter are in the 95th + percentile of all of the values between 1979 and 2006 for that time period.

This sequence of analyses can be completed for the Energy Release Component, 100-HR Dead Fuel Moisture (FM-100), and 100-HR Dead Fuel Moisture (FM-100) variables by only updating the 'Variable' drop-down of the 'Variable' section. If you've made multiple maps in this session, you can click 'Reset' before going on to the next example. This helps ensure that none of the variables from the previous map get caught, and leave you with a map you didn't expect. It is helpful to do this after every 5 or so mapping requests.

Example #2: Mapping Forecasted Fire Indicators

Similar to the workflow for mapping historical fire indicators, we will start in the 'Variable' section. We will start by updating 'Type' to 'Forecasts'. This will load a new set of datasets into the 'Dataset' drop-down. We are going to select 'CFS gridMET - 4km - 1to28day' and the move to 'Variable' where we are going to select 'Energy Release Component (ERC)'. In the 'Processing' section, a new option called 'Statistic (over forecast ensemble)'. This allows you to select specific ensembles or Median/Mean/Minimum/Maximum/25th Percentile/75th Percentile of the 48 ensembles. We will leave all of the defaults for that section for now. Lastly, we will move to the 'Time Period' section. The options for this section are a bit different for the forecasts. We provide various summarization options. We will choose 'Week 1 (Next 1-7 Days of Data)'.

We can then evaluate the forecasted week of values vs a past time period to see how much those values differ. We will head back to the 'Calculation' drop-down in the 'Processing' section. We will update the 'Calculation' to 'Difference from Average Conditions'. We will then head to the 'Time Period' section, where 'Year Range for Historical Avg/Distribution' has been added. We will update the year range to '1991' - '2020'. The CFS gridMET forecast values will be differenced from the gridMET values.

This sequence of analyses can be completed for other forecast variables as well.

Interested in a similar workflow using the Climate Engine API?

The Climate Engine API allows users to extend their analysis by exporting larger geotiffs, creating timeseries data over multiple areas, and writing loops to make multiple API calls. You can find a python notebook walking you through a similar workflow here.

Plotting Fire Indicators

In this section, we will shift gears and go step by step through how to charts of historical and forecasted fire indicators.

Example #1: Plotting Forecasted Fire Indicators

To start making a chart, ensure you are on the 'Make Graph' tab of the panel on the left hand side of the app. We will start at the 'Time Series Calculation' section. We will keep the first drop-down on 'Native Time Series' and second drop-down on 'One Variable Analysis'. Next, we will move to the 'Region' section where we will select 'US Regions' using the drop down. It will then add another drop-down where we will select US Hydrologic Unit Codes(HUC) 8. It will then populate the last drop-down with huc-8 options, and we will select Colorado Headwaters. Finally, we will move to 'Variable 1'. If you just requested a map, the details will automatically populate this section. We will work our way through it. First, we will ensure that 'Type' is 'Forecasts', 'Dataset' is 'CFS gridMET - 4km - 1to28day', and 'Variable' is 'Energy Release Component (ERC)'. Next, we will leave the 'Computation Resolution (Scale)' at its default which is 4000m, 'Statistic (over region)' on 'Mean', and 'Statistic (over forecast ensemble)' at '48-Ensemble Median' . We will skip the 'Masking' section for now. Lastly, we will select 'Week 1-4 (Next 1-28 Days of Data)' for 'Time Period'. Then we will click the 'Get Map Layer' button. Then we will click the 'Get Time Series' button. To download the chart, hit the 'Download' button on the top right corner of the chart.

Example #2: Plotting Historical Fire Indicators

To plot the last 28 days of Energy Release Component (ERC) for this hub-8, we will leave the first two sections the same. We will then head to the 'Variable 1' section, where we will update 'Type' to 'Climate & Hydrology' , 'Dataset' to 'GridMET - 4km - Daily' , and 'Variable' to 'Energy Release Component (ERC)'. We will leave the next two drop-downs the alone, and head to the 'Time Period' section. We will select 'Custom Date Range' and update 'Start Date' to '2025-02-07' and 'End Date' to '2025-03-06'. Then we will click the 'Get Time Series' button.