Free community tools: season condition and flood impact for any district

The community tools answer two questions that local organisations ask often, for a district chosen from a list. Season condition says whether rainfall and vegetation over the last one, three or six months were below, near or above normal. Flood impact maps land that has newly flooded and estimates how many people live in it. Both are free, need only a web browser, and return a short answer in plain English rather than a dataset to be processed.

The tools exist because of the site’s mission: to make GIS and remote sensing usable by organisations that do not have a GIS specialist, a powerful computer or the time for a long workflow. This article explains how the tools work, shows one real answer from each, and sets out what the answers can and cannot be used for.

How the tools work

There is no map to draw on. You choose a country, then a district, province or county from a list; an outline map is offered as a check where the connection allows. The tools cover more than 52,000 districts and provinces in 200 countries and territories, using boundaries from geoBoundaries. One country’s boundaries were left out because their licence does not allow commercial reuse.

You then choose the period. For season condition this is the length of the season, one, three or six months, and the month it ends. For flood impact it is the date of the latest satellite pass you want to include. The tool shows what it will do and how long it is likely to take before you submit, and the work runs on our server: about a minute for season condition, one to three minutes for flood impact.

Every answer has the same parts:

• a one-sentence headline;
• the figures behind it, with a chart or map;
• a statement of how reliable the answer is, with the reasons;
• the limitations that apply to this answer;
• the data sources, their licences and the required credits;
• the date it was produced and a link that can be shared.

Answers can be printed on one A4 page, saved as a PDF and downloaded as CSV. The text is written for a reading age of about twelve, and the pages are kept small so that they load on a slow mobile connection. Answers are kept for 90 days, and anyone asking the same question in that time is given the stored answer at once.

Season condition: Marsabit, Kenya, October to December 2022

Marsabit County in northern Kenya relies on two rainy seasons, the long rains from March to May and the short rains from October to December. By August 2022 the regional climate centre was forecasting a fifth consecutive failed rainy season across the eastern Horn of Africa, with more than 50 million people already acutely food insecure (UN News, August 2022).

The answer reads: rainfall in Marsabit was below normal, and vegetation was well below normal.

• Rainfall was 86 mm against an average of 161 mm for October to December, 54% of normal. That placed 2022 at the 23rd percentile of the 1991–2020 record: seven of those thirty years were drier.
• Vegetation had a mean NDVI of 0.17 against an average of 0.28 for the same months. That is the 1st percentile of the 2001–2020 record, and the lowest value for these months since MODIS records began in 2001.

The pairing matters. Rainfall alone suggests a poor season but not an exceptional one. Vegetation shows the cumulative effect of several failed seasons in a row: pasture had not recovered from earlier droughts, so a below-normal season left the land barer than at any point in two decades. The answer was rated “fairly reliable”: the district covers many CHIRPS rainfall cells, and all five 16-day vegetation images for the period were clear enough to use. Because mapped cropland covers less than 1% of the county, vegetation was measured on all vegetated land rather than on cropland, and the answer says so.

The figures are consistent with what was reported on the ground. The Kenya Drought Response Plan for 2023 records five consecutive below-average rainy seasons and an estimated 2.6 million livestock deaths, and named Marsabit among the counties expected to move from Crisis (IPC Phase 3) to Emergency (IPC Phase 4) by March 2023.

Download the Marsabit answer as a one-page PDF or the figures as CSV. You can run the same question again in the tool.

How the comparison is made

Rainfall comes from CHIRPS version 3, which blends satellite estimates with rain-gauge records. The total for the chosen months is compared with the same months in each year from 1991 to 2020. Vegetation comes from the MODIS vegetation index, compared with 2001 to 2020. Where at least 5% of a district is cropland in ESA WorldCover, vegetation is measured on cropland only, so that a poor harvest is not hidden by forest or grazing land. The result is then placed in one of five classes: well below normal (below the 10th percentile), below normal (10th to 33rd), near normal (33rd to 67th), above normal (67th to 90th) and well above normal.

Flood impact: Feni, Bangladesh, August 2024

In the second half of August 2024, heavy rain and water flowing in from upstream in India caused flash floods across eastern Bangladesh. By 30 August, OCHA reported that 5.8 million people in 11 districts were affected and that more than 500,000 people were sheltering in over 3,400 evacuation centres. Feni was among the worst-affected districts (OCHA Situation Report No. 2, 30 August 2024).

The answer reads: about 39 km² of Feni appeared flooded in radar images from 21 to 28 August 2024, and about 4,400 people live in that area, with a range of 1,100 to 8,000.

• Flooded area was about 39 km², 4.4% of the district, mapped at 40 m. The range of 20 to 57 km² comes from running stricter and looser versions of the water rule.
• The reference was January 2024, chosen automatically as the driest month of a normal year in Feni, using six radar passes from the same orbits.
• Coverage was complete: 100% of the district was observed.
• Reliability was rated “use with care”, which is the highest rating the flood tool gives.

Download the Feni answer as a one-page PDF or the figures as CSV, or run it again in the tool.

Why the people figure is much lower than the numbers affected

About 4,400 people is far below the hundreds of thousands reported as affected in Feni. The difference is real, and it is the most important thing to understand about this tool.

• Radar sees open water. Calm open water reflects the radar signal away from the satellite and appears dark. Water among houses, trees, crops and narrow streets does not look like that, because buildings and vegetation scatter the signal back. Floodwater in villages and towns is therefore often missed.
• The figure counts residents, not people affected. It is the number of people whose homes lie where open water was seen. People whose roads were cut, whose fields or wells were flooded, or who fled from nearby land are not counted.
• The timing may miss the peak. Sentinel-1 passes over a given place every six to twelve days. Flash floods can rise and fall between passes, so the images show the water that remained on those dates, not the greatest extent.

For these reasons the tool describes its population figure as “people living in the flooded area” and never as people affected, displaced or in need. It is best used to show where flooding was concentrated and to compare districts or events, alongside reports from the ground.

How the map is made

Each radar pass after the event is compared with passes from the reference month. A pixel is counted as newly flooded if it is dark enough to be open water after the event and has become markedly darker than in the reference. Areas that are water in at least half of all observations since 1984 in the JRC Global Surface Water record, such as rivers, lakes and ponds, are removed, and isolated single pixels are filtered out. People are counted from the GHSL population grid over the flooded pixels.

What the tools should and should not be used for

The tools are designed to help a local organisation describe a situation clearly and quickly: in a report to a funder, a briefing for a district office, a news story or a funding bid. The answers are indicative. They are not a substitute for field assessment, for official drought and food-security classifications such as the IPC, or for damage and needs assessments, and they should not be the only evidence for decisions about who receives help.

No household-level or personal data is used. Season condition reports whole districts, and flood impact reports flooded areas mapped at 20 m or coarser, so no answer identifies an individual home. Both tools state how reliable each answer is, and both list the specific limitations that apply: few rain-gauge cells, cloudy vegetation images, a reference month that had to be moved, or parts of a district the radar did not see.

Licences and reuse

Every dataset the tools use is licensed for commercial use and redistribution of derived outputs, and the licence, its source and the required credit are listed with each answer. Answers may be reused for any purpose, including in publications and funding bids, provided the line “Generated at TechnicalGIS.com” and the data credits are kept. Where a country’s district boundaries are published under a share-alike licence, the answer says so.

Getting help with larger questions

The tools are free and there is no advertising on them. The site is supported by advertising elsewhere and by paid bespoke work. If your organisation needs a larger area, a series of dates, other indicators or help interpreting the results, please contact the team and choose “Community or NGO project”.

To start, open the community tools, choose a tool and a district, and read the reliability note before you use the answer.