I'm interested in visualizing climate data across the conterminous U.S. to determine regions that I would be interested in living in. So far, I've pulled in NOAA data, and I've:

  • averaged the maximum daily temperature measurements per location
  • averaged the minimum daily temperature measurements per location
  • taken the difference between the average maximum temperature measurements and the average minimum temperature measurements to determine how much weather varies across the year per location
  • pulled precipitation in tenths of mm per location
  • pulled snowfall in mm per location

At first, these seemed like "good enough" approximations, but I'm discovering that this approach is somewhat inadequate because it is:

  • Not appropriately representing swings in daytime-nightime temperatures
  • Not appropriately representing the annual variance in temperatures between locations
  • using averages

I was initially seeking to discover areas with a high mean temperature, low variance in temperatures annually and daily, and low rainfall. What metrics could I be using to improve this investigation? Thanks!

  • $\begingroup$ I'm thinking that statistical variance could be a better indication of variance than the range between annual avg max and avg min temps. Is it incorrect to just average all the min and max temps together for the average temp to use in a variance calculation? $\endgroup$ Commented Sep 5, 2018 at 16:58
  • 2
    $\begingroup$ Generally the average weather is comfortable ; it is the actual weather that is often soo bad. for example the averare rain in Houston is about 1.1 in. / week , but last year they got 40.0 in. in a week. $\endgroup$ Commented Sep 5, 2018 at 19:00
  • $\begingroup$ Very good point. May be worth investigating rainfall variance in addition to temperature variance. $\endgroup$ Commented Sep 6, 2018 at 16:08

2 Answers 2


The Köppen climate classification is a classic metric to classify climates:

N-America Köppen map
Source: Wikimedia Commons

It classifies climates based on the number of months with temperature below and/or above certain thresholds, and on whether rainfall is evenly distributed or concentrated in a particular season. You will find details on the definitions on Wikipedia.

An alternative is the Trewartha classification:

Trewartha map Source: Wikimedia Commons.

Note that the climate you seek may not exist:

I was initially seeking to discover areas with a high mean temperature, low variance in temperatures annually and daily, and low rainfall.

A low temperature variance tends to happen in humid climates. Dry climates tend to have large temperature variances. A subtropical highland climate may come closest to what you seek, although it's not particularly dry. In the Köppen classification, this would be Cwb. From the Wikipedia article:

Cwb = Subtropical highland climate or Monsoon-influenced temperate oceanic climate; coldest month averaging above 0 °C (32 °F) (or −3 °C (27 °F)), all months with average temperatures below 22 °C (71.6 °F), and at least four months averaging above 10 °C (50 °F). At least ten times as much rain in the wettest month of summer as in the driest month of winter (an alternative definition is 70% or more of average annual precipitation received in the warmest six months).

I don't think any place with Cwb climate exists in the United States. Quito is famous for being the city of eternal spring, and some areas of Mexico are classified as Cwb, as in Johannesburg. I suspect the closest you'll get in the USA will be somewhere not too high, not too low in the Sierra Nevada.

  • 2
    $\begingroup$ This is super interesting and informative; thanks for sharing. The criteria for classification appears to be 12-month average temperature above/below a certain pre-defined threshold and a threshold number of months with precipitation above/below another pre-defined threshold. Maybe worth clarifying that I'm less interested in broad classifications and more interested in comparing the individual microclimates/locations, which is why I gravitated towards the classification criteria. Surprisingly, this doesn't seem to be too far off from what I'm currently using. Thanks! $\endgroup$ Commented Sep 5, 2018 at 14:32

Zech–Aslan energy statistic stands out as the most recommended metric

Baseline setup and dataset The data that are used represent some aspects of the historical (recent past) and future climates at each grid cell of an array. The reference location corresponds to the grid cell for which analogs are sought, and candidate locations are all grid cells contained within the investigation domain, which includes the reference location itself. As presented in Fig. 1, the investigation domain is restricted to the following geographical limits: 358–508N, 608–1008W (excluding lake and ocean grid cells). When not otherwise stated, the reference location is the grid cell that includes the city of Montreal (centered at approximately 45.68N, 73.58W and identified by a star in Fig. 1). For some calculations, 126 reference locations are used (also identified in Fig. 1). Each location’s climate is synthesized by a set of 90 values, accounting for values of three annual indicators over a 30-yr span. The historical and future periods extend over 1971–2000 and 2041–70, respectively.

enter image description here

An Assessment of Six Dissimilarity Metrics for Climate Analogs by PATRICK GRENIER Ouranos Consortium, Montreal, Quebec, Canada


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.