Updates to the Climatic Water Deficit Toolbox part II

Here I walk through an example using the West Walker River watershed in Nevada/California USA. The watershed is 2575 square kmand took 5 minutes and 9 seconds to run for the 30 year normals. Lower elevations are extremely arid (7 inches or 170 mm) while upper elevations in the Sierra Nevada can receive significant snowfall (> 9000 mm at high elevations). Below is the DEM used as well as a SSURGO soils map (light colors represent higher soil water holding capacity).

The first steps involve partitioning precipitation into snow vs. rain fractions using a melt factor and then determining the amount of snowmelt + rain to get the water supply. The maps below show water supply for March, April, and May moving upslope due to snowmelt. The purple areas are the areas of greatest water availability.

We use the heat load index of McCune and Keon (2002) to apply a fine-scale topographic influence on potential evapotranspiration. The heat load index is fixed in space and gets applied as a multiplier in the potential evapotranspiration calculations. In the map below reds show areas of high heat load (e.g. south-facing slopes).

Potential evapotranspiration is calculated for each month. In the maps below yellow represents the highest PET and the months are April, June, and August.

Updates to the Climatic Water Deficit Toolbox for ArcGIS 10.1

Recently I posted about a new tool for calculating day length from a raster of latitude and mentioned that this improvement would be incorporated into the Climatic Water Deficit Toolbox.  I'm happy to announce here it is.  The updated version of the Climatic Water Deficit Toolbox no longer requires the cumbersome day length lookup table, which allows it to be run on much larger areas. In fact colleagues and I have recently generated a 60 year dataset for the entire western USA.

You may ask "What is the Climatic Water Deficit Toolbox?". In a nutshell, it is a toolbox that runs in ArcGIS that calculates potential and actual evapotranspiration based on a Thornthwaite water balance approach and water supply as the sum of snowmelt and rainfall. In essence, it quantifies the demand that a plant experiences on water (potential evapotranspiration), the supply of water (water supply), and the amount of biological activity expected given the limitation of water (actual evapotranspiration). The unmet demand (difference between potential and actual evapotranspiration) is referred to as the climatic water deficit. The tool uses PRISM climate data (30 year normals or monthly rasters) plus SSURGO soils for available water supply and a DEM to incorporate the effects of  fine-scale variation in solar radiation on potential evapotranspiration. The tool accounts for snowpack and soil water storage, runs on a monthly time scale, but does not simulate movement of water through the soil (e.g. groundwater model) nor does it measure actual evapotranspiration via remote sensing. Despite these limitations, the approach is expected to provide a more proximate measure of water stress on plants compared to just using temperature and precipitation variables.

For those interested I highly recommend reading the following papers:

Lutz, J.A., van Wagtendonk, J.W., Franklin J.F. 2010. Climatic water deficit, tree species ranges, and

climate change in Yosemite National Park. Journal of Biogeography 37:936-950. (see also www.wmrs.edu/projects/cerec/pdfs/Lutz3CERECposter.pdf)

Dyer, J.M. 2009. Assessing topographic patterns in moisture use and stress using a water balance approach, Landscape Ecology 24: 391-403.

Stephenson, N. L. 1998. Actual evapotranspiration and deficit: biologically meaningful correlates of vegetation distribution across spatial scales. Journal of Biogeography 25: 855-870.

 Here is how the tool works in ArcMap:

There are two versions of the tool. One that runs on 30 year normals (average climate over a 30 year period) and the other that runs on monthly PRISM data. There are some really basic tools for downscaling the 4 km monthly data to 800 m (the resolution of the normals).  There are seven basic steps in the calculation which can be run together in the all-in-one model or separately.

The outputs are stored in the following folders. Be forewarned, however, that these files can take up a lot of disk space.

Common question - How to combine raster and vector data and deal with NoData

This one is probably the most common question I get.  It goes something along the line of  "I've got some polygon/line data that I want to combine with a raster layer, but how do I deal with the NoData outside the polygon?". Other situations that this apply to include other raster outputs that use NoData as their background, such as outputs from Brownian Bridges, or remote sensing images in which part of the image may be masked. Hence, nearly everyone who works with ArcMap in a natural resource context is likely to encounter this problem at least once (if not nearly daily in my case).

Here is a hypothetical example and solution. In the picture to the right we've got a habitat model (raster) and buffered roads which are not suitable habitat for our hypothetical species. We wish to combine the two layers to yield a habitat model in which all areas near roads are not suitable habitat (value = 0).

In ArcGIS the first step would be to convert the polygon to raster. We might do this using the Polygon to Raster Tool.  It doesn't matter what the Value Field is set to but you probably want to set the Cellsize to the same as the habitat model layer (see the picture to the right).

Then use the IsNull tool  to convert all areas that are NoData to 1.  The IsNull tool is in the Spatial Analyst toolbox and is also available as Map Algebra in the Raster Calculator tool. IsNull is a really simple, but also one of the most powerful tools in ArcMap.

The final step involves the use of the Con tool in Spatial Analyst to combine the IsNull output with the original raster (see below). The input conditional raster should be the output from the IsNull tool. The expression selects all cells with a value of 0 (i.e. in this case they are within the road buffer). The true value is set to 0 to indicate that areas near roads are unsuitable habitat. The false value uses the original habitat suitability raster to indicate that for area not near roads the habitat is the same as the habitat model.

The final result  is map that looks like the one below. Areas near roads all have a value set to zero. You can also combine IsNull and Con in raster calculator using the following type of statement: Con(IsNull[Raster],0,[Raster]) or use ModelBuilder to streamline the process.

 

New tool - Day Length Tools for ArcGIS

Day length turns out to be one of the critical calculations for calculating potential evapotranspiration in the Climatic Water Deficit Toolbox .

In order to improve the ease of use of the Climatic Water Deficit Toolbox I've decided to automate the calculation of day length for any given latitude for any given Julian day. The result is the Day Length Tools for ArcGIS . There is also a version of the tool that calculates day length for all months using the 15th day of each month. I can imagine that there is potential for this tool beyond just climatic water deficit. For example, maybe this tool could be used by people studying plant phenology to map areas where day length exceeds a certain number of days.

One question that people may have is "where can I get a latitude raster?". I created one in ArcGIS using lines of latitude and then performing a second order trend analysis using the Trend tool in Spatial Analyst (ArcToolbox - Spatial Analyst Tools - Interplotation - Trend).