CMIP6 temperature and precipitation projections in Excel
Many people who need future climate data, such as engineers, planners, consultants and students, work in Excel rather than Python. This guide shows how to get NASA NEX-GDDP-CMIP6 temperature and precipitation projections for a location as an Excel workbook, and how to turn them into useful numbers with standard formulas.
Updated 16 November 2026
1. Pick the right variables
NEX-GDDP-CMIP6 has four temperature and precipitation variables. You can select up to four variables per job, so all of them fit in one request.
| Variable | Name in files | What it tells you | Units you can choose |
|---|---|---|---|
| Daily mean temperature | tas | How warm a typical day is; general warming, seasonal shifts, heating and cooling needs | K, °C, °F |
| Daily maximum temperature | tasmax | How hot the hottest part of the day gets; heatwaves, infrastructure stress, workplace safety | K, °C, °F |
| Daily minimum temperature | tasmin | How cold nights get; frost, winter utility planning, crop protection | K, °C, °F |
| Precipitation | pr | Daily rainfall and snowfall combined; flood risk, drought, water supply | kg m⁻² s⁻¹, mm/day, in/day |
Tip: choose mm/day or in/day for precipitation. Each daily value is then that day's total, which makes sums and averages easy to read.
2. Get the workbook
On the request form, pick your location, select the variables above with the units you want, choose your models, and select Historical plus one future scenario (each job can include two scenarios, and Historical counts as one). When the job is finished, you'll receive an email with links to a daily CSV, an annual-average CSV and an Excel workbook. The CSV download guide walks through the form step by step, and the scenarios and models guide helps you choose.
3. How the workbook is laid out
The workbook has two sheets, Annual Averages and Daily Timeseries, with the same layout:
- Rows 1–9: details of your request (location name, coordinates, models, variables, scenarios, dataset citation).
- Row 11: column headers.
- Row 12 onwards: the data. Column A holds the year (annual sheet) or the date (daily sheet, as real Excel dates).
Every other column is one model, variable and scenario, followed by a multi-model mean for each group. In the Seattle sample, which has two models (ACCESS-CM2 and EC-Earth3), precipitation in in/day and temperature in °F, the columns are:
| Columns | Variable | Scenario | Contents |
|---|---|---|---|
| B, C, D | Precipitation (in/day) | Historical | ACCESS-CM2, EC-Earth3, multi-model mean |
| E, F, G | Precipitation (in/day) | SSP1-2.6 | ACCESS-CM2, EC-Earth3, multi-model mean |
| H, I, J | Temperature (°F) | Historical | ACCESS-CM2, EC-Earth3, multi-model mean |
| K, L, M | Temperature (°F) | SSP1-2.6 | ACCESS-CM2, EC-Earth3, multi-model mean |
Historical columns are empty for future years and future-scenario columns are empty for historical years. Your own workbook's column letters depend on what you selected, so check the headers in row 11 and adjust the formulas below.
4. Worked examples
These examples use the Seattle sample, which covers 1990–1999 (Historical) and 2045–2054 (SSP1-2.6). Paste them into an empty cell on a new sheet.
Change in average temperature
Compare the future period's average with the historical baseline, using the multi-model mean columns (M for the future, J for Historical) on the Annual Averages sheet:
=AVERAGEIFS('Annual Averages'!M:M, 'Annual Averages'!A:A, ">=2045", 'Annual Averages'!A:A, "<=2054")
- AVERAGEIFS('Annual Averages'!J:J, 'Annual Averages'!A:A, ">=1990", 'Annual Averages'!A:A, "<=1999")
In the sample this returns about +4.2 °F (from 54.1 °F to 58.2 °F). Swap in the individual model columns (K and H, then L and I) to see the range across models: +4.7 °F for ACCESS-CM2 and +3.7 °F for EC-Earth3.
Percentage change in precipitation
=AVERAGEIFS('Annual Averages'!G:G, 'Annual Averages'!A:A, ">=2045", 'Annual Averages'!A:A, "<=2054")
/ AVERAGEIFS('Annual Averages'!D:D, 'Annual Averages'!A:A, ">=1990", 'Annual Averages'!A:A, "<=1999") - 1
Format the cell as a percentage. In the sample, average precipitation rises by about 8%.
Annual precipitation total
The Annual Averages sheet holds the average daily precipitation for each year, not the yearly total. To get a total, sum the daily values for that year on the Daily Timeseries sheet:
=SUMIFS('Daily Timeseries'!G:G, 'Daily Timeseries'!A:A, ">="&DATE(2050,1,1), 'Daily Timeseries'!A:A, "<="&DATE(2050,12,31))
In the sample this gives about 44.2 inches for 2050 (multi-model mean). Alternatively, multiply the annual average by the number of days in the year (365 or 366; 360 for models with a 360-day calendar).
Counting hot days
Count how many days in 2045–2054 had a daily mean temperature above 70 °F for ACCESS-CM2 (column K), then divide by 10 for a yearly average:
=COUNTIFS('Daily Timeseries'!A:A, ">="&DATE(2045,1,1), 'Daily Timeseries'!A:A, "<="&DATE(2054,12,31), 'Daily Timeseries'!K:K, ">70") / 10
This returns about 62.9 days a year for ACCESS-CM2 and 56.2 for EC-Earth3 (column L). The same formula on the multi-model mean (column M) gives only 53.0, which is fewer than either model.
Count extremes per model, then average the counts. Averaging models day by day smooths out highs and lows, so threshold counts on a multi-model-mean column underestimate extremes. For heat-stress thresholds, use daily maximum temperature (tasmax) rather than the daily mean.
5. Tips for sound results
- Compare 20- to 30-year periods, not single years. Individual years vary a lot. A common choice is a 1985–2014 baseline (the last 30 years of the Historical run) against a period such as 2041–2070. The Seattle sample uses 10-year periods only to keep the file small.
- Measure change against the models' own Historical run, as in the examples above, so you compare like with like.
- Report the range across models, not just the multi-model mean. More models give a better picture of uncertainty.
- Look at more than one scenario. A lower-emission and a higher-emission scenario show the range of possible futures. Run a second job if you need more than one future scenario.
- Cite the data. See how to cite the NEX-GDDP-CMIP6 dataset and this tool.